Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Common genetic variants, acting additively, are a major source of risk for autism

Common genetic variants, acting additively, are a major source of risk for autism Background: Autism spectrum disorders (ASD) are early onset neurodevelopmental syndromes typified by impairments in reciprocal social interaction and communication, accompanied by restricted and repetitive behaviors. While rare and especially de novo genetic variation are known to affect liability, whether common genetic polymorphism plays a substantial role is an open question and the relative contribution of genes and environment is contentious. It is probable that the relative contributions of rare and common variation, as well as environment, differs between ASD families having only a single affected individual (simplex) versus multiplex families who have two or more affected individuals. Methods: By using quantitative genetics techniques and the contrast of ASD subjects to controls, we estimate what portion of liability can be explained by additive genetic effects, known as narrow-sense heritability. We evaluate relatives of ASD subjects using the same methods to evaluate the assumptions of the additive model and partition families by simplex/multiplex status to determine how heritability changes with status. Results: By analyzing common variation throughout the genome, we show that common genetic polymorphism exerts substantial additive genetic effects on ASD liability and that simplex/multiplex family status has an impact on the identified composition of that risk. As a fraction of the total variation in liability, the estimated narrow-sense heritability exceeds 60% for ASD individuals from multiplex families and is approximately 40% for simplex families. By analyzing parents, unaffected siblings and alleles not transmitted from parents to their affected children, we conclude that the data for simplex ASD families follow the expectation for additive models closely. The data from multiplex families deviate somewhat from an additive model, possibly due to parental assortative mating. Conclusions: Our results, when viewed in the context of results from genome-wide association studies, demonstrate that a myriad of common variants of very small effect impacts ASD liability. Keywords: Narrow-sense heritability, Multiplex, Simplex, Quantitative genetics * Correspondence: [email protected] Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA Full list of author information is available at the end of the article © 2012 Klei et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Klei et al. Molecular Autism 2012, 3:9 Page 2 of 13 http://www.molecularautism.com/content/3/1/9 Background Simplex Collection (SSC) [19] and the other from the Beliefs about the genetic architecture of autism Autism Genome Project (AGP) [20]. Importantly the spectrum disorders (ASD) have changed dramatically analysis of these two cohorts allows for an estimate of over the past few decades. Early twin studies produced the heritability of ASD in simplex versus multiplex heritability estimates approaching 90% [1,2] and, while families as well as an assessment of how well the data fit no specific risk loci were known at the time, it was predictions for an additive model of inheritance [21]. believed that liability was conferred by a handful of When all risk variation acts additively, for example, and genes of large effect. Later, data on the distribution of no other forces alter the covariance of relatives, the ASD within families, together with results from linkage liability for relatives of an affected individual consistently analyses, were interpreted to mean that liability arose halves for each degree of separation from the proband. from many genes [3]. Recent work has definitively Therefore, we also evaluate heritability tracing to SSC demonstrated the substantial contribution of de novo and AGP parents and SSC unaffected siblings, evaluat- variation [4-11]. Indeed multiple studies of rare single ing the empirical results against simulation-derived nucleotide and copy number variants (CNVs) have sug- expectations. Finally we use the same techniques to gested that 15% or more of liability traces to de novo ask what residual heritability is contained in what the mutation, effects that are genetic but not inherited [2]. field calls pseudo-controls, which are genotypes formed Importantly, despite notable recent successes in gene from the alleles that parents did not transmit to their discovery efforts, key questions remain regarding the affected offspring. overall nature and scale of the genetic contribution to ASD liability. For example, the contribution of genetics Methods is still debated: a recent large-scale twin study [12] esti- ASD families mated only 38% of liability was accounted for by additive DNA samples from SSC and AGP family members genetic effects, while common environmental factors genotyped on the Illumina Infinium 1Mv3 (duo) micro- accounted for 55% of the variance; whereas most studies array or the Illumina Infinium 1Mv1 microarray were of twins find much higher heritability, including studies analyzed here. Specifically qualifying SSC samples were of phenotypes in the broader spectrum (see [13,14] for genotyped on the Illumina Infinium 1Mv3 (duo) micro- review). Moreover, despite a near-consensus that com- array (71.8%) while most AGP samples were genotyped mon and transmitted variation must confer liability, on the Illumina Infinium 1Mv1 microarray (98.7%). multiple genome-wide association studies have so far Both arrays genotype roughly 1,000,000 single nucleotide not revealed replicable common polymorphisms [15] polymorphisms (SNPs) and the overlap between the associated with ASD, and studies of rare structural and SNP sets is almost perfect. sequence mutations have largely failed to account for The SSC sample [19] includes >2,000 genotyped fam- the anticipated risk associated with transmitted variation ilies. However, our analyses targeted a homogeneous [6,7]. Finally, since the earliest CNV studies in ASD, subset of these data. First, we included only samples it has been postulated that the architecture of simplex genotyped on an Illumina 1M array; families had to be and multiplex autism would be strikingly different [4]. ‘quads’ consisting of an unaffected mother and father, an However not all studies have found marked disparities affected proband and an unaffected sibling; and all mem- in the rate of de novo mutation in simplex versus bers of a quad had to have complete genotypes (>95% multiplex families, and large effect de novo mutations completion rate). Only samples of European ancestry have been characterized in both multiplex and simplex were included. European ancestry for the SSC families families [9,16]. was determined using GemTools [22,23] for all available Consequently, to gain insight into the broad questions SSC probands. To conduct the ancestry analysis we regarding the nature of the genetic factors underlying selected 5,156 SNPs with at least 99.9% calls for geno- ASD, we have estimated how much of the population types, had minor allele frequency MAF >0.05, and were variability in liability can be traced to inherited variation, at least 0.5 Mb apart. Individuals were clustered into specifically the narrow-sense heritability of ASD. Yang nine ancestry groups based on four significant dimen- et al. [17] proposed elegant methods in which the herit- sions of ancestry. The central five clusters, which held ability of liability can be estimated as a function of the a total of 1,686 families, were identified as being of covariance between trait values, in this instance affection European descent. The ancestry cluster information status [18], and the genome wide genetics of the sub- combined with complete genotype information yielded a jects. This contrasts with the usual approach of esti- total of 965 SSC families for the analysis. mating heritability from the distribution of trait values The AGP Stage 1 dataset [16,20] comprised 1,471 in pedigrees. In the present study, these methods are families, of which 1,141 were previously identified to applied to two ASD data sets, one from the Simons be of European ancestry [20]. European ancestry was Klei et al. Molecular Autism 2012, 3:9 Page 3 of 13 http://www.molecularautism.com/content/3/1/9 confirmed by analyses identical to those applied to the MAF > 0.01, and produce a P-value for Hardy-Weinberg SSC families (see Additional file 1: Figure S1). equilibrium > 0.005. Following these QC steps, data from 965 SSC quad families and 1,141 AGP families were ana- Clinical evaluation lyzed using genotypes from 713,259 SNPs. Probands for the SSC and AGP cohorts were diagnosed in a similar manner (for diagnostic protocol for SSC, see Statistical calculations and motivation [19]; for AGP, see [16,20]). All SSC parents were Estimating heritability as a case-control contrast screened for Autism Spectrum Disorder by the Broad Heritability of ASD from probands versus controls was Autism Phenotype Questionnaire [24] (self-report) and estimated using GCTA software [28], which encodes the Social Reciprocity Scale - Adult Research Version the theory laid out in [17,18]. Prevalence of ASD was [25] (informant report). Moreover, family history evalu- taken to be 1% [27]. For each of the analyses, Genetic ation excluded first-, second-, or third-degree relatives Relationship Matrices (GRM) were determined for each who met diagnostic criteria for ASD or intellectual dis- of the 23 chromosomes using the –make-grm option ability. For AGP families 46.2% were known to be multi- in GCTA [28]. These were then combined in an over- plex, another 38.2% were identified as simplex on the all matrix, using the –mgrm option in GCTA. The first basis of a family history indicating no known first- to 10 principal components of ancestry were determined third-degree relatives with ASD, and the remaining 15.6% using –pca in GCTA. These 10 PCA were then used as were of unknown status. Note that most AGP parents covariates for estimating the heritability using –reml in were not systematically evaluated for ASD, unlike those GCTA. A prevalence of 0.01 for autism spectrum dis- from the SSC, and when AGP parents were systematically orders was used to transform the heritability on the evaluated, the results were not used to screen out affected observed scale to the heritability on the liability scale. individuals and thus multiplex families. In addition, while all available SSC family members were genotyped, only The logic of estimating heritability from unaffected parent-proband trios were genotyped for the AGP even family members when additional siblings were available. Due to the screening of SSC samples, no SSC parents would meet criteria for ASD. Given that is the case, what Control subjects is the justification for assigning them to be affected and Controls derived from a convenience sample, specifically contrasting them to controls to estimate the heritability 1,663 individuals from HealthABC [26]. Control samples in the parental generation? Under the additive heritabil- were also genotyped on the Illumina Infinium 1Mv3 ity model parents transmit many genetic variants of (duo) array, like most of the AGP data, providing excel- small effect to their offspring, with the expectation that lent comparability with the case dataset. Moreover, we half would be transmitted from each parent. The parents reasoned that ASD is sufficiently rare (approximately 1% of probands are thus more similar at liability loci than [27]) that screened and unscreened controls would yield expected by chance, and our goal is to estimate this similar results. increased genetic similarity. Calling the parents affected and contrasting their genotypes to that of controls is a Filtering natural approach to estimating their genetic contribution To make heritability estimates comparable, we filtered to liability and it has precedence in quantitative genetics, all families and control subjects based on the follow- such as estimation of the heritability of milk production ing criteria: all were of European descent as determ- from its covariance arising from bulls, when only the ined by genetically-estimated ancestry (Additional file 1: bull’s female progeny give milk (for example [29]). Figure S1); genotypes for all family members met strin- A similar argument follows for unaffected siblings gent quality control (QC) criteria; and control samples from SSC families. These siblings should receive a ran- met identical QC criteria. dom sample of the parent’s genomes and, in expectation, For the three data sets we first chose SNPs genotyped this sampling would include half the liability alleles car- on all platforms. Then ambiguous AT, TA, CG, and GC ried by each parent. Thus the unaffected offspring SNPs were removed. A total of 813,960 SNP across the 22 should mirror the average liability carried by the parents autosomes and chromosome X were included for further and this level can be estimated by calling them affected quality evaluation. At the level of individuals, we required and contrasting their genotypes to those from controls. that genotyping completion rate be greater than 98%, that there be no discrepancy regarding nominal and genotype- Simulations to compute expected heritability for parents inferred sex, and no individuals in different families and pseudo-controls were closely related. At the level of individual SNPs, each While the literature contains numerous references to SNP must have a genotype completion rate > 98%, have the burden of risk variants carried by parents of simplex Klei et al. Molecular Autism 2012, 3:9 Page 4 of 13 http://www.molecularautism.com/content/3/1/9 versus multiplex families, we could not find quantitative 3. Proband and second child are both affected, no genetics analyses of it as a function of ascertainment restriction on the other individuals in the family (there is related work on the impact of multi-locus (unscreened multiplex family); inheritance on the power of candidate gene association 4. A mixture of 60% unscreened simplex families and studies [30,31]). We therefore evaluated the expected 40% unscreened multiplex families. heritability for parents, unaffected siblings, and pseudo- controls on the basis of simulations and the theory of By using rejection sampling, a total of 1000 families quantitative genetics regarding the selection differential were generated for each scenario and this procedure was (for ASD, approximately 1%) and the response to selec- repeated 100 times per scenario and proband heritability tion (expected change in the population’s mean liability). (50 and 75%). To obtain the heritability estimates for the The simulations are designed to mimic ascertainment family members, the average phenotype of the primary for simplex and multiplex families. probands on the liability scale (S) were compared to the One thousand SNPs having an impact on liability average phenotype of the family member of interest on were simulated. The allele frequency for SNP i, p , varied the liability scale (R). The heritability estimate based on ^ R between 0.01 and 0.99. Overall heritability h across all the family member was estimated as h ¼ . Note that n = 1000 SNPs was set to be either 0.50 or 0.75 for we also checked the heritability estimated from the pro- probands with ASD. The relative importance of each bands as a function of the reduction in genetic variance SNP, w , was determined by first selecting a fraction in the selected group. For probands, estimated heritabil- t between 0 and 1 at random using a uniform distri- ity was always close to 50% when that was the desired bution. These 1000 values were added to obtain T, heritability and always close to 75% when that was the and each SNP was weighted by w =t /T. The allele sub- i i desired heritability. stitution effect for each SNP i was then determined as From theoretical considerations we expected assorta- qffiffiffiffiffiffiffiffiffiffiffiffiffiffi w h tive mating to elevate the expected liability of pseudo- a ¼ . For each simulation 1000 families were 2pðÞ 1p i i controls and evaluated its impact by a simple experiment generated consisting of a father, mother, and one child using the simulation structure just described. Rather (AGP simplex) or two children (SSC simplex or AGP than randomly assign genotypes to mates, we first multiplex). Genotypes for the parents were assigned at randomly chose the paternal genotypes at the 1,000 random using the allele frequencies, while children liability SNPs, then assigned maternal genotypes on the received alleles from the parents using the rules of basis of the toss of a fair coin: heads the genotype was Mendelian inheritance. Likewise a pseudo-control was chosen at random, tails it was taken to be the father’s generated by comparing the genotype of the parents genotype. All simulations procedures were as described to that of the proband and assigning the un-transmitted above, except we conducted two simulations: for sim- allele of each parent as the alleles for the pseudo- ulation (a) the heritability of probands from simplex control’s genotype. After all genotypes in a family were families was taken to be 50% and ascertainment followed assigned, the genetic contribution to the underlying scenario 2 above; and for simulation (b) the heritability liability phenotype for each individual j in the family of probands from multiplex families was set to 75% and was determined by G = x a − μ in which x is j i =1 i i G i ascertainment followed scenario 3 above. the allele count for SNP i and μ = p (1 − p )a G i =1 i i i is the average genetic contribution over all genotypes. Robustness of results To simulate the environmental influence on the phe- To evaluate the robustness of the results, 1,986 indivi- notype of individual j, e , we drew a random number duals of European descent from the Neurogenetics from a normal distribution with mean 0 and variance Research Consortium [32] (NGRC) were available through (1-h ). The liability phenotype was then determined as dbGap [33] and used as a second control sample. For y = G + e . Affection status was then assigned based on j j j the NGRC study, genotypes were produced using the not affected when y < 2:326 affection status ¼ Illumina Infinium Human Omni2.5 microarray. There- affected when y ≥ 2:326 fore, to combine all four data sets, we performed QC representing a disease risk of 1% in the population. on 444,200 SNPs genotyped on all platforms, yielding Four different scenarios were simulated: 391,425 SNPs for analyses. 1. Primary child in the family is affected (proband), and father, mother, and designated sibling were Assessing the potential for experimental bias not-affected (SSC family); To explore the impact of different cohorts and genotyp- 2. Proband is affected, no restriction on the other ing protocols on estimated heritability, we conducted a individuals in the family (unscreened simplex family); series of contrasts between SSC and AGP samples of the Klei et al. Molecular Autism 2012, 3:9 Page 5 of 13 http://www.molecularautism.com/content/3/1/9 same relationship type – contrasting probands, mothers, Results and discussion fathers, and pseudo-controls – as well as HealthABC Estimates of heritability (h ) versus NGRC controls. Heritability of SSC probands, measured against HealthABC controls, was found to be 39.6% (Figure 1A, Table 1). SSC mothers, fathers and siblings, when contrasted to controls, Determining genomic coverage yielded an estimated heritability approximately half that of While 713,259 SNPs were used for primary analyses, probands (Figure 1A, Table 1), consistent with expected they constitute a small fraction of the SNPs in the values from theoretical analyses of an additive model human genome. Hence the heritability presented could (Figure 1A). We also generate a “pseudo-control” from the underestimate total heritability. On the other hand, alleles that parents did not transmit to their affected because genotypes of SNPs in close proximity tend to be offspring by using the program Plink [37]. When these correlated due to linkage disequilibrium, it does not pseudo-controls were contrasted to the unrelated control follow that the coverage of the genome by the SNPs sample they produce estimates roughly one-quarter of used here estimate only a small fraction of the heritabil- that identified in probands and close to the theoretical ity. To determine the shortfall in “genomic coverage” expectation, zero (Figure 1A), demonstrating that the and how it impacts estimates of heritability, we per- probands received the majority of risk alleles carried formed an experiment using data from the 1,000 by parents. Genomes project [34], under the assumption that cover- When heritability is estimated using AGP probands age of common variants in the 1,000 Genomes data (Figure 1B, Table 1), the point estimates are larger than is perfect. Assessing all SNPs genotyped in our data, those from SSC (h =55.2% versus 39.6%) although the as well as subsets thereof, we estimated heritability 95% confidence intervals overlap. Moreover the decline of liability. Using the same subsets, but in 1,000 Gen- in heritability for AGP parents relative to probands is omes subjects, we estimated levels of genomic coverage. 30% (55% for probands, 37% for parents), instead of the We can then relate estimated heritability to genomic 50% seen for SSC, and heritability estimated from coverage to develop a functional relationship between pseudo-controls is also higher (38%), consistent with the two. parental values (Figure 1B, Table 1). These results sug- We performed the experiment assessing “genomic gest that AGP parents carry a greater load of additive coverage” as follows. We assumed genomic coverage of risk variants than SSC parents and thus are, on average, SNPs with MAF > 0.1 would be essentially complete for closer to the threshold of being affected. the 379 European samples analyzed by the 1,000 A major difference between the SSC and AGP samples Genomes project. From these genomes we selected 50 was the ascertainment and assessment process. SSC par- 1Mb regions in which at least 500 SNPs in the 1,000 ents were systematically screened on two instruments to Genomes samples had MAF > 0.10. Coverage of these ensure they did not meet criteria for a spectrum diagno- regions by the 713,259 SNPs was calculated as a func- sis. Most parents from AGP families were not evaluated tion of the number of other SNPs with MAF > 0.1 that in this way, and a small fraction of those parents met were tagged by (correlated with) them; call the set of criteria for ASD [9,16]. While not as systematic as the M = 713,259 SNPs “tagSNP”. The tagging evaluation SSC phenotyping assessment, most AGP families did was implemented using Hclust [35]. Forcing tagSNP to have available information about simplex versus multi- be in the set of selected tag SNPs from the region, plex status. Consequently, we were able to compare Hclust evaluated how many more independent SNPs heritability of probands from AGP multiplex versus sim- N were required to cover the region when the min- plex families (Figure 1D, Table 2). The former was esti- imum linkage disequilibrium [36] r amongst tags could mated at 65.5% by comparison to HealthABC, whereas be no less than X,where X = {0.5, 0.7, and 0.9}. Then, probands for AGP simplex families it was 49.8%. Thus for each value of X, M/(M+N) estimates the coverage. estimates of heritability for AGP simplex probands Next we randomly sampled 50, 25 and 12.5% of the are somewhat closer to those from SSC probands 713,259 SNPs (356,630, 178,315, and 89,158 SNPs (Figure 1C) than to estimates for AGP multiplex pro- respectively) five times and each time estimated coverage bands. Moreover, for multiplex families and the mixed for these subsets. set of AGP families (simplex/multiplex/unknown), both the observed and expected heritability for first-degree relatives was higher than that seen in simplex families Human subjects research statement (Figure 1). These results comport with the literature The research described here is in compliance with the showing that unaffected relatives from multiplex families Helsinki Declaration, including appropriate informed tend to exhibit more features of the broader autism consent or assent [16,19,20,26,32,33]. phenotype than relatives in simplex families [38-40] (see Klei et al. Molecular Autism 2012, 3:9 Page 6 of 13 http://www.molecularautism.com/content/3/1/9 A B Pr Mo Fa Si Pc Pr Mo Fa Si Pc C D Pr Mo Fa Si Pc Pr Mo Fa Si Pc Figure 1 Estimated heritability for Autism Spectrum Disorders from ASD probands (Pr), as well as for their mothers (Mo), fathers (Fa), siblings (Si) and pseudo-controls (Pc). Blue dotted reference line is set to the estimated heritability from probands; the black line marks the expected heritability for first degree relatives; and the gray line marks the expected heritability from pseudo-controls. Expected values derived from simulations mimicking the recruitment strategy producing the samples for (A)-(D). (A) Simons Simplex Collection or SSC data; (B) Autism Genome Project or AGP data; (C) AGP data, only simplex families; (D) AGP data, only multiplex families. Additional file 2: Table S1 for estimates from combined estrogen/testosterone balance) in the face of a similar simplex samples). degree of genetic risk. A curious observation from AGP multiplex families Our results support either the first or second hypo- was that fathers generate larger heritability than theses but are not consistent with the third. The first mothers. We reasoned that this could be explained by hypothesis is impossible to rule out given the limited three plausible hypotheses: (1) the confidence intervals sample size. For the second hypothesis, if AGP fathers of the paternal and maternal estimates overlap, so there were simply carrying greater risk, some of those is no true difference; (2) the load of risk variants is, in additional risk alleles would be carried by the pseudo- fact, greater for AGP fathers; or (3) fathers carry a larger controls and the heritability obtained from the contrast number of both liability and protective alleles. The last of probands and pseudo-controls should be substantially of these requires some elaboration. Males are at much smaller than that observed from probands versus greater risk for ASD than females (4:1 or greater) and controls. Indeed the values are substantially smaller: parents carry additive risk factors, yet AGP fathers and 10.9% vs. 39.6% for SSC; 14.5% vs. 55.2% for all mothers are largely unaffected. It is possible, then, that AGP; 0.0% vs. 49.8% for simplex AGP, and 27.1% vs. the increased allele sharing in unaffected fathers is due 65.5% for multiplex AGP. Finally, if (3) were true, to a greater proportion of protective alleles, with females then contrasting probands to pseudo-controls would being resilient for some other reason (for example, produce substantial estimates of heritability because of Table 1 Heritability estimates and their standard errors (se) based on contrasts to HealthABC controls using genotypes from 713,259 SNPs SSC AGP Simplex All Simplex Multiplex Estimate se Estimate se Estimate se Estimate se Probands 0.396 0.082 0.552 0.068 0.498 0.118 0.655 0.139 Mothers 0.199 0.082 0.371 0.070 0.314 0.119 0.377 0.141 Fathers 0.196 0.084 0.370 0.070 0.352 0.119 0.666 0.143 Siblings 0.158 0.082 – – –– –– Pseudo controls 0.090 0.082 0.381 0.070 0.317 0.120 0.503 0.146 Heritability 0.0 0.4 0.8 0.0 0.4 0.8 0.0 0.4 0.8 0.0 0.4 0.8 Klei et al. Molecular Autism 2012, 3:9 Page 7 of 13 http://www.molecularautism.com/content/3/1/9 Table 2 Heritability estimates and their standard errors (se) based on contrasts to HealthABC and NGRC controls using genotypes from 391,425 SNPs SSC AGP HealthABC NGRC HealthABC NGRC Estimate se Estimate se Estimate se Estimate se Probands 0.395 0.082 0.378 0.073 0.553 0.068 0.586 0.063 Mothers 0.200 0.082 0.232 0.074 0.371 0.070 0.342 0.065 Fathers 0.196 0.084 0.153 0.073 0.373 0.070 0.518 0.063 Siblings 0.158 0.082 0.170 0.073 –– –– Pseudo controls 0.090 0.082 0.107 0.073 0.380 0.070 0.446 0.065 the differentiation induced by protective alleles, but this families, heritability estimated from X is less than that pre- is not observed. dicted by its size. This is noteworthy because chromosome X has been cited as a possible source of sex-differential li- Distribution of liability alleles in the genome ability for ASD [41]. Our results suggest that common var- If the additive variation for liability to ASD conforms iants affecting liability do not cluster on chromosome X. to the traditional polygenic or infinitesimal model, then liability variants should be distributed at random over the genome. The implication is that if heritability were Evaluating robustness of results estimated for each chromosome, the resulting estimates To evaluate the robustness of our results, we first should be correlated with the lengths of the chromo- contrasted the genotypes of SSC and AGP probands to somes. On the other hand, if the heritability traced to a a second large set of controls, 1,986 individuals from relatively small number of variants, even a few dozen, the Neurogenetics Research Consortium [32,33]. These such a correlation would be unlikely. In fact, we observe samples, genotyped on the Illumina Infinium Human significant correlation between per-chromosome herit- Omni2.5, were filtered and subjected to QC in an identi- ability and chromosome length (Figure 2), both for sim- cal fashion to the HealthABC control set. There was ex- plex (r = 0.46, P value = 0.028) and multiplex (r= 0.54, cellent agreement of heritability estimates for ASD from P value = 0.0075) families. the two control samples (Tables 2 and 3) despite differ- In Figure 2the deviationfrompredictionfor chromo- ences in ascertainment of the controls and the different some X is surprising. For both multiplex and simplex genotyping platforms. Multiplex Simplex 50 100 150 200 250 300 Chromosome length (cM) Figure 2 Estimated heritability per chromosome for simplex and muliplex families. In this figure chromosome X is marked distinctly, but each chromosome is mapped by its length. Heritability 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Klei et al. Molecular Autism 2012, 3:9 Page 8 of 13 http://www.molecularautism.com/content/3/1/9 Next, the impact of different cohorts and genotyping dimensional space of allele frequencies, phenotypes and platforms on estimates of heritability was explored by their interrelationships. Therefore even if two controls conducting a series of contrasts between SSC and AGP groups evoke similar estimates of ASD heritability from samples of the same relationship type: contrasting the same sample of probands, the controls themselves probands, mothers, fathers, and pseudo-controls. Note need not be close in the multidimensional space of allele that most SSC samples were genotyped on the Illumina frequencies. What generates the differentiation between 1Mv3 (duo) microarray (71.8%) while most AGP controls is unknown. It could arise from the different samples were genotyped on the Illumina Infinium genotyping platforms or from differences in ascertain- 1Mv1 microarray (98.7%). Contrasts between SSC and ment. In light of this difference, the fact that both con- AGP samples of the same relationship type (Additional trols sets give rise to nearly identical estimates of file 3: Table S2) produce estimates close to the difference heritability for all proband subsets is remarkable and between their control-based heritability. Indeed the suggests that the similarity amongst cases overwhelms estimates from direct contrasts were usually smaller than differences between the controls. the difference of control-based heritability (for probands, 0.08 vs. 0.15 ≈ 0.552-0.396 from Table 1; for mothers, Heritability of pseudo-controls 0.11 vs. 0.17; for fathers, 0.19 vs. 0.17; and for pseudo- There remains an unexplained feature of the results: controls, 0.22 vs. 0.29). Thus these results are not estimates of heritability for pseudo-controls tend to be consistent with effects attributable to genotyping plat- elevated over their theoretical values (Figure 1). Several form or ascertainment beyond multiplex/simplex status. genetic forces could be at play. The simulations to derive Implicit in these results is common genetic liability - the distribution of liability in families also produce esti- SSC and AGP probands must share many liability mates for pseudo-controls. Those results show (Figure 1) variants despite their differences in ascertainment. that while the expected heritability for simplex families Indeed when AGP multiplex probands are contrasted to is zero, multiplex status raises the expected value to SSC probands the resulting heritability is 0.23, quite 20%. It is not unreasonable to assume that the simplex similar to that expected by the difference in their esti- collections analyzed here contain families with unreal- mated heritability (0.66 - 0.40 = 0.26); and when AGP ized multiplex potential, and that might be especially simplex probands are contrasted to SSC probands, the true for AGP families that had ascertainment criteria less resulting estimated heritability, 0.0, is below that of the dif- stringent than those for SSC families. ference in their estimated heritability (0.50 - 0.40 = 0.10). A factor that will elevate the expected heritability in These results suggest that the difference between multi- pseudo-controls is positive assortative mating (hence- plex and simplex families is largely a matter of degree (see forth assortative mating). Assortative mating on pheno- also [42]), namely the number of liability alleles carried by types related to ASD liability has been previously parents, rather than a fundamental difference in the gen- reported [39]. When parents are genetically similar at etic architecture [4,43]. liability loci and they bear affected offspring, their Given the remarkable similarities of heritability esti- gametes will tend to be highly enriched for risk alleles, mates obtained for either set of control samples (Tables 2 even those that are not transmitted to affected offspring. and 3), one might anticipate there would be little, if Simple simulations mimicking assortative mating show any, difference between these controls. When we con- that it can exert an impact similar to the difference be- trasted these control samples, however, they produced a tween simplex and multiplex status. When simplex pro- heritability of 26.5% (Additional file 3: Table S2). Math- bands had heritability of 50% (that is, simulation a in ematically, estimates of heritability arise from a high Methods), the expected heritability of pseudo-controls Table 3 Heritability estimates and their standard errors (se) based on contrasts to HealthABC and NGRC controls using genotypes from 391,425 SNPs but separating the AGP data into multiplex and simplex families for estimation AGP multiplex AGP simplex HealthABC NGRC HealthABC NGRC Estimate se Estimate se Estimate se Estimate se Probands 0.650 0.139 0.710 0.140 0.503 0.117 0.494 0.114 Mothers 0.369 0.141 0.387 0.136 0.311 0.119 0.268 0.117 Fathers 0.664 0.143 0.693 0.140 0.359 0.119 0.520 0.113 Pseudo controls 0.497 0.146 0.524 0.140 0.323 0.120 0.438 0.117 Klei et al. Molecular Autism 2012, 3:9 Page 9 of 13 http://www.molecularautism.com/content/3/1/9 was 11.3% – versus 0% without assortative mating. direct estimate of the proportion of liability attribut- When multiplex probands had heritability of 75%, the able to additive genetic effects, whereas twin studies ob- expected heritability of pseudo-controls was 42.8% – tain their estimates by relying on assumptions that are versus 20.2% without assortative mating. These simple approximations. For example, Zuk et al. [45] point out experiments were not intended to cover the range of that non-additive genetic effects are almost surely a plausible scenarios for assortative mating relevant to component of the genetic architecture of any trait, but ASD, which would be impossible, but rather to demon- these effects cannot be captured by twin designs. Yet strate the effect of such mating on the nature of pseudo- for autism and other psychiatric disorders non-additive controls. Thus assortative mating could be an important genetic effects could be an integral component [46-48]. and salient source of enrichment. Whether these forces Twin designs also fail to capture other features, explain all of the elevated heritability for pseudo- such as maternal effects [49] and de novo mutations, controls will require further data and analyses. which are an important component of ASD genetic architecture [4-11]. Impact of genome coverage A recent ASD twin study [12] estimates 38% of ASD Because the set of SNPs used for primary analyses con- liability traces to additive genetic effects while 55% stitute a small fraction of the SNPs in the human gen- traces to common environment. Our point estimates ome, estimates of heritability (Figure 1) could be biased would be close to theirs if ascertainment of their families downward. Still, due to linkage disequilibrium, the was like that for SSC families, but not like that for AGP degree of bias is not trivial to estimate. Therefore families. A substantial fraction of their dizygotic twins, we performed an experiment to evaluate the shortfall however, are multiplex for ASD. Thus their point esti- in genomic coverage and how it impacts estimates of mate for heritability from additive genetic effects is low heritability. Results from the experiment are shown relative to ours. If rare inherited variation contributes in Additional file 4: Figure S2, in which estimated herit- substantially to liability for ASD, this makes the 38% ability was plotted against estimated coverage. These estimate seem lower still because twin studies should results suggest that heritability estimates from probands, capture these effects whereas our estimates cannot. as shown in Figure 1, are good approximations. They Genomewide association studies [18,50-52] have represent only slight underestimates of what would be detected only a handful of SNPs, all of small effect and obtained had the entire genome been sampled. none replicating reliably. Teaming this observation with In total our results demonstrate that a substantial our estimates of heritability (Figure 1) and the fact that portion of ASD liability arises from inherited variation these studies are underpowered to detect genetic var- acting additively. This pattern holds both for simplex iants of small effect size, but are otherwise well powered and multiplex families, with the burden of liability [15], we conclude there must be thousands of SNPs scat- greater in multiplex families, consistent with theoretical tered across the genome with common liability alleles. and empirical [38-40] results. The modeling reported Analyses of chromosome-specific heritability support here does not differentiate between additive effects due this conclusion (Figure 2). Employing analyses like those to common versus rare variation. Nonetheless it is rea- proposed by Stahl et al. [53] could estimate this distribu- sonable to assume that most of the estimated heritability tion of effects. traces to common variants because linkage disequilib- Because these loci have small effect, samples far larger rium between the common variants analyzed and rare than exist today will be required to identify a substantial liability variants should, on average, be small [44]. Thus fraction of them using standard genome-wide asso- the additive contribution of rare variants to ASD liability ciation methodology. Hence, for the immediate future, is likely underestimated. Imperfect coverage must also ample “missing heritability” for ASD will remain. Ingeni- ous designs will be required in the near term [54] have an impact, but our analyses suggest its impact is not large (Additional file 4: Figure S2). to identify SNPs affecting liability. In the longer Our analyses cannot address other features of the gen- term GWAS of a large number of ASD subjects, at least on the order of that performed for schizophrenia etic architecture of ASD, including non-additive genetic effects, which add to ASD’s broad-sense heritability [45], [55-57], should be one of the priorities for the field of and de novo mutations. In addition, because they under- ASD genetics. One way forward is to exploit shared liability across estimate the impact of rare inherited variation, they dif- fer from family-based estimates, such as from twin psychiatric disorders, taking advantage of larger samples studies, that do capture these effects. Still our findings of [58] afforded by cross-disorder meta-analysis. There is now sound evidence for common variants affecting substantial heritability are consistent with the majority of twin studies [1,2] and are richer in some ways because liability for schizophrenia [55-57], including a study the analytic technique [17,18] used here provides a similar to ours [46]. Given the documented sharing of Klei et al. Molecular Autism 2012, 3:9 Page 10 of 13 http://www.molecularautism.com/content/3/1/9 rare variants affecting risk for both disorders (for ex- study that reveals a plausible biological link to ASD ample [59]), it would not be surprising to find that some liability [67]. common variants affect liability to both schizophrenia The genetic architecture of ASD has numerous compo- and ASD. nents: additive, non-additive and de novo genetic effects, as The estimated heritability for schizophrenia using well as gene-gene and gene-environment interactions. The methods similar to ours is 23% [46]; for bipolar disorder results shown here are relevant to only one of these compo- and similar methods it is 40% [60]; and for major de- nents. Other components, such as de novo events, are also pression it is 32% [61]. None of these studies separate known to make a substantial contribution to liability [4-11], out simplex and multiplex families, so in that sense whileothersremain tobethoroughlyinvestigated[45]. they are most comparable to the estimate obtained over Already analyses of rare variation of major effect has all AGP families, 55%, although the representation of revealed a substantial number of genes affecting liability multiplex families in the AGP sample is likely larger [8-11,68-70]; it is reasonable to predict that common var- than for the other samples. Regardless of the differences iants regulating expression of those ASD genes could also in simplex/multiplex representation, these estimates affect liability [71]. We hypothesize that the interplay of are stochastically similar, in view of their standard rare and common variants is critical not only to liability it- errors, emphasizing that common variants affect liability self, but to the expression of ASD or other relevant psychi- for most if not all psychiatric disorders. Moreover their atric and developmental disorders. The dynamics of this impact appears to be similar in magnitude across dis- interplay will likely be an important area for future autism orders, as measured by heritability estimated from research. common variants. That ASD shows the largest estimated heritability is Conclusions notable and could reflect the fact that the sibling recur- Common genetic polymorphisms exert substantial additive rence risk is, on average, higher for siblings of an ASD genetic effects on ASD liability and their impact differs by proband than for siblings of probands diagnosed with ascertainment strategies used to recruit families. For sim- schizophrenia, bipolar disorder or major depression. Sib- plex families, who have only a single affected individual in ling recurrence risk is a ratio, defined as the probability multiple generations, approximately 40% of liability traces of a sibling being affected, given that the proband is to additive effects whereas this narrow-sense heritability affected, divided by the prevalence of the disorder in the exceeds 60% for ASD individuals from multiplex families. general population. Recent studies put this recurrence Data for simplex ASD families follow the expectation for risk at almost 20 for ASD [62], whereas for schizophre- additive models closely. Data from multiplex families nia it is 6 to 10 fold [63], for bipolar disorder it is 4 to 10 deviate somewhat from an additive model. This result is fold [64], and for major depression it is roughly twofold consistent with what would beexpectedfrom positiveas- [64]. The larger heritability could also trace to differences sortative mating, but our data do not prove such a among studies. It is possible that our estimates of herit- pattern of mating occurred. In light of results from ability are inflated by unknown differences between our genome-wideassociation studies, theremust bemanycom- case and control samples, including ascertainment biases mon variants of very small effect affecting liability to ASD. and genotype quality. Regarding the latter, we selected case and control samples genotyped on the same geno- Availability of supporting data typing platform to minimize differences and we did not The data sets supporting the results of this article detect any large differences in allele frequencies, but we are available in the repositories: Simons Foundation cannot rule out subtle differences in quality. Autism Research Initiative, SFARI [http://sfari.org/sfari- Regarding identification of common variants affecting initiatives/simons-simplex-collection]; and the National liability, our results suggest that the contrast of case and Institutes of Health database of Genotypes and Pheno- pseudo-control genotypes, the “family-based” analysis, is types, dbGaP [http://www.ncbi.nlm.nih.gov/gap]. not optimal. In many samples pseudo-controls carry a substantial burden of risk variants and their presence Additional files degrades the power of family-based analysis to detect risk SNPs (see also [30,31]). Instead it appears that Additional file 1: Figure S1. Ancestry projects for principal component 1 (PC.1) versus principal component 2 (PC.2) for the samples used in the population-based controls contrasted with ASD cases analysis of heritability. Red dots represent subjects with an ASD diagnosis would be a more powerful design [65], even after and blue are controls. HealthABC=HABC. adjusting for ancestry [66]. In this regard it is intriguing Additional file 2: Table S1. Heritability estimates and their standard that the earliest GWAS of ASD [50] used population- errors (se) using 391,425 SNP when AGP and SSC simplex family data are combined or only multiplex AGP families are analyzed. Analyses include based controls to identify a single locus at 5p14.1, and all HealthABC and NGRC control samples. this result has since garnered support from a functional Klei et al. Molecular Autism 2012, 3:9 Page 11 of 13 http://www.molecularautism.com/content/3/1/9 Swedish Science Council; The Centre for Applied Genomics, Canada; Utah Additional file 3: Table S2. Heritability estimates and their standard Autism Foundation, USA; Core award 075491/Z/04. Wellcome Trust, UK. errors (se) obtained when contrasting AGP and SSC samples of the same Genotype and phenotype data were obtained from dbGap, as provided by relationship type, as well as contrasting HealthABC versus NGRC controls. AGP Study Investigators. Additional file 4: Figure S2. Heritability for ASD probands as a function HealthABC: These controls were obtained from Database for Genotypes and of estimated “genomic coverage” for varying levels of r . Coverage is Phenotypes (dbGap) at http://www.ncbi.nlm.nih.gov/gap. Funding support for the estimated as the fraction of all known SNPs identified by 1000 Genomes “CIDR Visceral Adiposity Study” (Study accession number: phs000169.v1.p1) was with minor allele frequency > 0.1 tagged by the set of SNPs used to provided through the Division of Aging Biology and the Division of Geriatrics and estimate heritability for probands; see Methods for more details. From Clinical Gerontology, NIA. The CIDR Visceral Adiposity Study includes a genome- the left points map onto 12.5%, 25%, 50%, and 100% of the SNPs used to wide association study funded as part of the Division of Aging Biology and the estimate heritability. Top line is for probands from multiplex families, Division of Geriatrics and Clinical Gerontology, NIA. Assistance with phenotype bottom for probands from simplex families. harmonization and genotype cleaning, as well as with general study coordination, was provided by Heath ABC Study Investigators. NGRC: We also used the NINDS dbGaP database from the CIDR: NGRC Abbreviations Parkinson’s Disease Study (dbGap accession number phs000196.v2.p1). The AGP: Autism Genome Project; ASD: Autism Spectrum Disorders; CNVs: Copy genetic arm of the study has been funded by NIH since 1998 (R01 NS36960, Number Variants; GCTA: Genome-Wide Complex Trait Analysis, Software used Haydeh Payami, PI). In 2004, the consortium was formalized as a Michael J to estimate heritability, amongst others; GRM: Genetic Relationship Matrices; Fox Foundation Funded Global Genetic Consortium, and an epidemiologic HealthABC: A sample of subjects used as controls and genotyped on the ® arm was implemented. Genotype and phenotype data were obtained from Illumina Infinium 1Mv3 (duo) array; MAF: Minor Allele Frequency; dbGap, as provided by NGRC Parkinson’s Disease Study Investigators. NGRC: Neurogenetics Research Consortium, a sample of subjects used as ® For both the HealthABC and NGRC studies, genotyping services were controls and genotyped on the Illumina Infinium Human Omni2.5 provided by the Center for Inherited Disease Research (CIDR). CIDR is funded microarray; QC: Quality Control; SNPs: Single Nucleotide Polymorphisms; through a federal contract from the National Institutes of Health to The SSC: Simons Simplex Collection. Johns Hopkins University, contract number HHSN268200782096C and HHSN268201100011I. Competing interests The authors declare no competing financial interests. Author details Authors’ contributions Department of Psychiatry, University of Pittsburgh School of Medicine, MWS supervised the overall project, EHC its phenotypic portions; LK, KR and Pittsburgh, Pennsylvania, USA. Program on Neurogenetics, Yale University BD conceived of the analyses; LK implemented the analyses; EHC, KR, MWS, School of Medicine, New Haven, Connecticut, USA. Child Study Center, Yale SJS, and BD wrote the first draft of the manuscript; all others authors University School of Medicine, New Haven, Connecticut, USA. Department commented on and refined it. Most authors recruited families, produced or of Psychiatry, Yale University School of Medicine, New Haven, Connecticut, evaluated data and commented on the manuscript. All authors read and USA. Department of Genetics, Yale University School of Medicine, New approve the final manuscript. Haven, Connecticut, USA. Department of Psychology, University of Michigan, Ann Arbor, MI, USA. Neurogenetics Program, Department of Neurology and Center for Autism Research and Treatment, Semel Institute, David Geffen Acknowledgments School of Medicine, University of California Los Angeles, Los Angeles, Research supported by grants from the Simons Foundation and MH057881. California, USA. Department of Human Genetics, Emory University School of SSC: We are grateful to all of the families participating in the Simons Medicine, Atlanta, Georgia, USA. Division of Genetics, Children's Hospital Foundation Autism Research Initiative (SFARI) Simplex Collection (SSC). This Boston, Harvard Medical School, Boston, Massachusetts, USA. Department work was supported by a grant from the Simons Foundation. We wish to of Psychiatry, McGill University, Montreal Children's Hospital, Montreal, QC thank the SSC principal investigators A.L. Beaudet, R. Bernier, J. Constantino, H3Z 1P2, Canada. Department of Psychiatry, Mount Sinai School of E.H. Cook, Jr., E. Fombonne, D. Geschwind, D.E. Grice, A. Klin, D.H. Ledbetter, Medicine, New York, New York, USA. Geisinger Health System, Danville, C. Lord, C.L. Martin, D.M. Martin, R. Maxim, J. Miles, O. Ousley, B. Peterson, Pennsylvania, USA. Center for Autism and the Developing Brain, Weill J. Piggot, C. Saulnier, M.W. State, W. Stone, J.S. Sutcliffe, C.A. Walsh, and Cornell Medical College, White Plains, New York, USA. Yale Center for E. Wijsman; the coordinators and staff at the SSC sites; the SFARI staff, in Genome Analysis, Orange, Connecticut, USA. Departments of Pediatrics and particular M. Benedetti; Prometheus Research; the Yale Center of Genomic Human Genetics, The University of Michigan Medical Center, Ann Arbor, Analysis staff, in particular M. Mahajan, S. Umlauf, I. Tikhonova and A. Lopez; Michigan, USA. Department of Molecular Biology, Cell Biology and T. Brooks-Boone, N. Wright-Davis and M. Wojciechowski for their help in Biochemistry, Brown University, Providence, Rhode Island, USA. Department administering the project at Yale; I. Hart for support; and G.D. Fischbach, of Psychiatry and Human Behavior, Brown University, Providence, Rhode A. Packer, J. Spiro, M. Benedetti and M. Carlson for their helpful suggestions Island, USA. Howard Hughes Medical Institute and Division of Genetics, throughout. Approved researchers can obtain the SSC population data set Children's Hospital Boston, and Neurology and Pediatrics, Harvard Medical described in this study by applying at https://base.sfari.org. School Center for Life Sciences, Boston, Massachusetts, USA. Department of AGP: We used data from the Autism Genome Project (AGP) Consortium - Molecular Physiology & Biophysics, Center for Molecular Neuroscience, Whole Genome Association and Copy Number Variation Study of over 1,500 Vanderbilt University, Nashville, Tennessee, USA. Institute for Juvenile Parent-Offspring Trios - Stage I (dbGaP Study Accession: phs000267.v1.p1). Research, Department of Psychiatry, University of Illinois at Chicago, Chicago, Funding for AGP was provided from National Institutes of Health (HD055751, Illinois, USA. Department of Statistics, Carnegie Mellon University, HD055782, HD055784, HD35465, MH52708, MH55284, MH57881, MH061009, Pittsburgh, Pennsylvania, USA. MH06359, MH066673, MH080647, MH081754, MH66766, NS026630, NS042165, NS049261); The Canadian Institutes for Health Research (CIHR); Received: 20 August 2012 Accepted: 4 October 2012 Assistance Publique - Hôpitaux de Paris, France; Autism Speaks UK; Canada Published: 15 October 2012 Foundation for Innovation/Ontario Innovation Trust; Grant: Po 255/17-4. Deutsche Forschungsgemeinschaft, Germany; EC Sixth FP AUTISM MOLGEN; Fundação Calouste Gulbenkian, Portugal; Fondation de France; Fondation References FondaMental, France; Fondation Orange, France; Fondation pour la 1. Bailey A, Le Couteur A, Gottesman I, Bolton P, Simonoff E, Yuzda E, Recherche Médicale, France; Fundação para a Ciência e Tecnologia, Portugal; Rutter M: Autism as a strongly genetic disorder: evidence from a British The Hospital for Sick Children Foundation and University of Toronto, Canada; twin study. Psychol Med 1995, 25:63–77. INSERM, France; Institut Pasteur, France; Convention 181 of 19.10.2001. Italian 2. Devlin B, Scherer SW: Genetic architecture in autism spectrum disorder. Ministry of Health; John P Hussman Foundation, USA; McLaughlin Centre, Curr Opin Genet Dev 2012, 22:229–237. Canada; Rubicon 825.06.031. Netherlands Organization for Scientific Research; 3. Risch N, Spiker D, Lotspeich L, Nouri N, Hinds D, Hallmayer J, Kalaydjieva L, TMF/DA/5801. Royal Netherlands Academy of Arts and Sciences; Ontario McCague P, Dimiceli S, Pitts T, Nguyen L, Yang J, Harper C, Thorpe D, Ministry of Research and Innovation, Canada; Seaver Foundation, USA; Vermeer S, Young H, Hebert J, Lin A, Ferguson J, Chiotti C, Wiese-Slater S, Klei et al. Molecular Autism 2012, 3:9 Page 12 of 13 http://www.molecularautism.com/content/3/1/9 Rogers T, Salmon B, Nicholas P, Petersen PB, Pingree C, McMahon W, copy number variation in autism spectrum disorders. Nature 2010, Wong DL, Cavalli-Sforza LL, Kraemer HC, et al: A genomic screen of 466:368–372. autism: evidence for a multilocus etiology. Am J Hum Genet 1999, 17. Yang J, Benyamin B, McEvoy BP, Gordon S, Henders AK, Nyholt DR, 65:493–507. Madden PA, Heath AC, Martin NG, Montgomery GW, Goddard ME, 4. Sebat J, Lakshmi B, Malhotra D, Troge J, Lese-Martin C, Walsh T, Yamrom B, Visscher PM: Common SNPs explain a large proportion of the heritability Yoon S, Krasnitz A, Kendall J, Leotta A, Pai D, Zhang R, Lee YH, Hicks J, for human height. Nat Genet 2010, 42:565–569. Spence SJ, Lee AT, Puura K, Lehtimäki T, Ledbetter D, Gregersen PK, 18. Lee SH, Wray NR, Goddard ME, Visscher PM: Estimating missing heritability Bregman J, Sutcliffe JS, Jobanputra V, Chung W, Warburton D, King MC, for disease from genome-wide association studies. Am J Hum Genet 2011, Skuse D, Geschwind DH, Gilliam TC, Ye K, et al: Strong association of de 88:294–305. novo copy number mutations with autism. Science 2007, 316:445–449. 19. Fischbach GD, Lord C: The Simons simplex collection: a resource 5. Marshall CR, Noor A, Vincent JB, Lionel AC, Feuk L, Skaug J, Shago M, for identification of autism genetic risk factors. Neuron 2010, Moessner R, Pinto D, Ren Y, Thiruvahindrapduram B, Fiebig A, Schreiber S, 68:192–195. Friedman J, Ketelaars CE, Vos YJ, Ficicioglu C, Kirkpatrick S, Nicolson R, 20. Anney R, Klei L, Pinto D, Regan R, Conroy J, Magalhaes TR, Correia C, Sloman L, Summers A, Gibbons CA, Teebi A, Chitayat D, Weksberg R, Abrahams BS, Sykes N, Pagnamenta AT, Almeida J, Bacchelli E, Bailey AJ, Thompson A, Vardy C, Crosbie V, Luscombe S, Baatjes R, et al: Structural Baird G, Battaglia A, Berney T, Bolshakova N, Bölte S, Bolton PF, Bourgeron T, variation of chromosomes in autism spectrum disorder. Am J Hum Genet Brennan S, Brian J, Carson AR, Casallo G, Casey J, Chu SH, Cochrane L, 2008, 82:477–488. Corsello C, Crawford EL, Crossett A, et al: A genome-wide scan for 6. Levy D, Ronemus M, Yamrom B, Lee YH, Leotta A, Kendall J, Marks S, common alleles affecting risk for autism. Hum Mol Genet 2010, Lakshmi B, Pai D, Ye K, Buja A, Krieger A, Yoon S, Troge J, Rodgers L, 19:4072–4082. Iossifov I, Wigler M: Rare de novo and transmitted copy-number variation 21. Falconer DS: Introduction to Quantitative Genetics. London: Longman; 1981. in autistic spectrum disorders. Neuron 2011, 70:886–897. 22. Lee AB, Luca D, Klei L, Devlin B, Roeder K: Discovering genetic ancestry 7. Sanders SJ, Ercan-Sencicek AG, Hus V, Luo R, Murtha MT, Moreno-De-Luca D, using spectral graph theory. Genet Epidemiol 2009, 34:51–59. Chu SH, Moreau MP, Gupta AR, Thomson SA, Mason CE, Bilguvar K, 23. Klei L, Kent BP, Melhem N, Devlin B, Roeder K: GemTools: a fast and Celestino-Soper PB, Choi M, Crawford EL, Davis L, Wright NR, Dhodapkar RM, efficient approach to estimating genetic ancestry; 2011. http://arxiv.org/pdf/ DiCola M, DiLullo NM, Fernandez TV, Fielding-Singh V, Fishman DO, Frahm S, 1104.1162.pdf. Garagaloyan R, Goh GS, Kammela S, Klei L, Lowe JK, Lund SC, et al: Multiple 24. Hurley RS, Losh M, Parlier M, Reznick JS, Piven J: The broad autism recurrent de novo CNVs, including duplications of the 7q11.23 Williams phenotype questionnaire. J Autism Develop Dis 2007, 37:1679–1690. syndrome region, are strongly associated with autism. Neuron 2011, 25. Constantino JN, Gruber CP: The Social Responsiveness Scale manual. 70:863–885. Los Angeles, CA: Western Psychological Services; 2005. 8. Sanders SJ, Murtha MT, Gupta AR, Murdoch JD, Raubeson MJ, Willsey AJ, 26. HealthABC data. http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study. Ercan-Sencicek AG, DiLullo NM, Parikshak NN, Stein JL, Walker MF, Ober GT, cgi?study_id=phs000169.v1.p1. Teran NA, Song Y, El-Fishawy P, Murtha RC, Choi M, Overton JD, 27. Autism and Developmental Disabilities Monitoring Network Surveillance Bjornson RD, Carriero NJ, Meyer KA, Bilguvar K, Mane SM, Sestan N, Year 2008 Principal Investigators; Centers for Disease Control and Lifton RP, Günel M, Roeder K, Geschwind DH, Devlin B, State MW: De novo Prevention: Prevalence of autism spectrum disorders--Autism and mutations revealed by whole-exome sequencing are strongly associated Developmental Disabilities Monitoring Network, 14 sites, United States, with autism. Nature 2012, 485:237–241. 2008. MMWR Surveill Summ 2012, 61:1–19. 9. Neale BM, Kou Y, Liu L, Ma'ayan A, Samocha KE, Sabo A, Lin CF, Stevens C, 28. Yang J, Lee SH, Goddard ME, Visscher PM: GCTA: a tool for genome-wide Wang LS, Makarov V, Polak P, Yoon S, Maguire J, Crawford EL, Campbell NG, complex trait analysis. Am J Hum Genet 2011, 88:76–82. Geller ET, Valladares O, Schafer C, Liu H, Zhao T, Cai G, Lihm J, Dannenfelser R, 29. http://digitalcommons.unl.edu/cgi/viewcontent.cgi? Jabado O, Peralta Z, Nagaswamy U, Muzny D, Reid JG, Newsham I, Wu Y, et al: article=1425&context=animalscifacpub. Patterns and rates of exonic de novo mutations in autism spectrum 30. Risch N: Implications of multilocus inheritance for gene-disease disorders. Nature 2012, 485:242. association studies. Theor Popul Biol 2001, 60:215–220. 10. O'Roak BJ, Vives L, Girirajan S, Karakoc E, Krumm N, Coe BP, Levy R, Ko A, 31. Ferreira MA, Sham P, Daly MJ, Purcell S: Ascertainment through family Lee C, Smith JD, Turner EH, Stanaway IB, Vernot B, Malig M, Baker C, Reilly B, history of disease often decreases the power of family-based association Akey JM, Borenstein E, Rieder MJ, Nickerson DA, Bernier R, Shendure J, studies. Behav Genet 2007, 37:631–636. Eichler EE: Sporadic autism exomes reveal a highly interconnected 32. Hamza TH, Zabetian CP, Tenesa A, Laederach A, Montimurro J, Yearout D, protein network of de novo mutations. Nature 2012, 485:246–250. Kay DM, Doheny KF, Paschall J, Pugh E, Kusel VI, Collura R, Roberts J, 11. Iossifov I, Ronemus M, Levy D, Wang Z, Hakker I, Rosenbaum J, Yamrom B, Griffith A, Samii A, Scott WK, Nutt J, Factor SA, Payami H: Common genetic Lee YH, Narzisi G, Leotta A, Kendall J, Grabowska E, Ma B, Marks S, variation in the HLA region is associated with late-onset sporadic Rodgers L, Stepansky A, Troge J, Andrews P, Bekritsky M, Pradhan K, Parkinson's disease. Nat Genet 2010, 42:781–785. Ghiban E, Kramer M, Parla J, Demeter R, Fulton LL, Fulton RS, Magrini VJ, 33. Neurogenetics Research Consortium data. http://www.ncbi.nlm.nih.gov/ Ye K, Darnell JC, Darnell RB, et al: De novo gene disruptions in children projects/gap/cgi-bin/study.cgi?study_id=phs000196.v2.p1. on the autistic spectrum. Neuron 2012, 74:285–299. 34. Clarke L, Zheng-Bradley X, Smith R, Kulesha E, Xiao C, Toneva I, Vaughan B, 12. Hallmayer J, Cleveland S, Torres A, Phillips J, Cohen B, Torigoe T, Miller J, Preuss D, Leinonen R, Shumway M, Sherry S, Flicek P, 1000 Genomes Project Fedele A, Collins J, Smith K, Lotspeich L, Croen LA, Ozonoff S, Lajonchere C, Consortium: The 1000 Genomes Project: data management and Grether JK, Risch N: Genetic heritability and shared environmental factors community access. Nat Methods 2012, 9:459–462. among twin pairs with autism. Arch Gen Psychiatry 2011, 68:1095–1102. 35. Rinaldo A, Bacanu SA, Devlin B, Sonpar V, Wasserman L, Roeder K: 13. Ronald A, Hoekstra RA: Autism spectrum disorders and autistic traits: a Characterization of multilocus linkage disequilibrium. Genet Epidemiol decade of new twin studies. Am J Med Genet B Neuropsychiatr Genet 2011, 2005, 28:193–206. 156B:255–274. 36. Devlin B, Risch N: A comparison of linkage disequilibrium measures for 14. Taniai H, Nishiyama T, Miyachi T, Imaeda M, Sumi S: Genetic influences on fine-scale mapping. Genomics 1995, 29:311–322. the broad spectrum of autism: study of proband-ascertained twins. 37. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Am J Med Genet B Neuropsychiatr Genet 2008, 147B:844–849. Maller J, Sklar P, de Bakker PI, Daly MJ, Sham PC: PLINK: a tool set for 15. Devlin B, Melhem N, Roeder K: Do common variants play a role in whole-genome association and population-based linkage analyses. risk for autism? Evidence and theoretical musings. Brain Res 2011, Am J Hum Genet 2007, 81:559–575. 1380:78–84. 38. Bernier R, Gerdts J, Munson J, Dawson G, Estes A: Evidence for broader autism phenotype characteristics in parents from multiple-incidence 16. Pinto D, Pagnamenta AT, Klei L, Anney R, Merico D, Regan R, Conroy J, autism families. Autism Res 2012, 5:13–20. Magalhaes TR, Correia C, Abrahams BS, Almeida J, Bacchelli E, Bader GD, Bailey AJ, Baird G, Battaglia A, Berney T, Bolshakova N, Bölte S, Bolton PF, 39. Virkud YV, Todd RD, Abbacchi AM, Zhang Y, Constantino JN: Familial Bourgeron T, Brennan S, Brian J, Bryson SE, Carson AR, Casallo G, Casey J, aggregation of quantitative autistic traits in multiplex versus simplex Chung BH, Cochrane L, Corsello C, et al: Functional impact of global rare autism. Am J Med Genet B Neuropsychiatr Genet 2009, 150B:328–334. Klei et al. Molecular Autism 2012, 3:9 Page 13 of 13 http://www.molecularautism.com/content/3/1/9 40. Szatmari P, MacLean JE, Jones MB, Bryson SE, Zwaigenbaum L, Bartolucci G, 59. Dolcetti A, Silversides CK, Marshall CR, Lionel AC, Stavropoulos DJ, Mahoney WJ, Tuff L: The familial aggregation of the lesser variant in Scherer SW, Bassett AS: 1q21.1 Microduplication expression in adults. biological and nonbiological relatives of PDD probands: a family history Genet Med 2012, in press. study. J Child Psychol Psychiatry 2000, 41:579–586. 60. Lee SH, DeCandia TR, Ripke S, Yang J, Schizophrenia Psychiatric 41. Marco EJ, Skuse DH: Autism-lessons from the X chromosome. Soc Cogn Genome-Wide Association Study Consortium (PGC-SCZ); International Affect Neurosci 2006, 1:183–193. Schizophrenia Consortium (ISC); Molecular Genetics of Schizophrenia 42. Spiker D, Lotspeich LJ, Dimiceli S, Myers RM, Risch N: Behavioral Collaboration (MGS), Sullivan PF, Goddard ME, Keller MC, Visscher PM, phenotypic variation in autism multiplex families: evidence for a Wray NR: Estimating the proportion of variation in susceptibility to continuous severity gradient. Am J Med Genet 2002, 114:129–136. schizophrenia captured by common SNPs. Nat Genet 2012, 44:247–250. 43. Zhao X, Leotta A, Kustanovich V, Lajonchere C, Geschwind DH, Law K, 61. Lubke GH, Hottenga JJ, Walters R, Laurin C, de Geus EJ, Willemsen G, Law P, Qiu S, Lord C, Sebat J, Ye K, Wigler M: A unified genetic theory Smit JH, Middeldorp CM, Penninx BW, Vink JM, Boomsma DI: Estimating for sporadic and inherited autism. Proc Natl Acad Sci USA 2007, the genetic variance of major depressive disorder due to all single 104:12831–12836. nucleotide polymorphisms. Biol Psychiatry 2012, 72:707–709. 62. Ozonoff S, Young GS, Carter A, Messinger D, Yirmiya N, Zwaigenbaum L, 44. Sun X, Namkung J, Zhu X, Elston RC: Capability of common SNPs to tag Bryson S, Carver LJ, Constantino JN, Dobkins K, Hutman T, Iverson JM, rare variants. BMC Proc 2011, 5(Suppl 9):S88. Landa R, Rogers SJ, Sigman M, Stone WL: Recurrence risk for autism 45. Zuk O, Hechter E, Sunyaev SR, Lander ES: The mystery of missing spectrum disorders: a Baby Siblings Research Consortium study. heritability: genetic interactions create phantom heritability. Proc Natl Pediatrics 2011, 128:e488–e495. Acad Sci USA 2012, 109:1193–1198. 63. Kendler KS, Diehl SR: The genetics of schizophrenia: a current, 46. Risch N: Linkage strategies for genetically complex traits. I. Multilocus genetic-epidemiologic perspective. Schizophr Bull 1993, 19:261–285. models. Am J Hum Genet 1990, 46:222–228. 64. Smoller JW, Finn CT: Family, twin, and adoption studies of bipolar 47. Sanders AR, Duan J, Gejman PV: Complexities in psychiatric genetics. disorder. Am J Med Genet C Semin Med Genet 2003, 123C:48–58. Int Rev Psychiatry 2004, 16:284–293. 65. Bacanu S-A, Devlin B, Roeder K: The power of genomic control. Am J Hum 48. Slatkin M: Exchangeable models of complex inherited diseases. Genetics Genet 2000, 66:933–944. 2008, 179:2253–2261. 66. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D: 49. Devlin B, Daniels M, Roeder K: The heritability of IQ. Nature 1997, 388:468–471. Principal components analysis corrects for stratification in genome-wide 50. Wang K, Zhang H, Ma D, Bucan M, Glessner JT, Abrahams BS, Salyakina D, association studies. Nat Genet 2006, 38:904–909. Imielinski M, Bradfield JP, Sleiman PM, Kim CE, Hou C, Frackelton E, 67. Kerin T, Ramanathan A, Rivas K, Grepo N, Coetzee GA, Campbell DB: Chiavacci R, Takahashi N, Sakurai T, Rappaport E, Lajonchere CM, Munson J, A noncoding RNA antisense to moesin at 5p14.1 in autism. Sci Transl Estes A, Korvatska O, Piven J, Sonnenblick LI, Alvarez Retuerto AI, Herman EI, Med 2012, 4:128ra40. Dong H, Hutman T, Sigman M, Ozonoff S, Klin A, et al: Common genetic 68. Berkel S, Marshall CR, Weiss B, Howe J, Roeth R, Moog U, Endris V, Roberts variants on 5p14.1 associate with autism spectrum disorders. Nature W, Szatmari P, Pinto D, Bonin M, Riess A, Engels H, Sprengel R, Scherer SW, 2009, 459:528–533. Rappold GA: Mutations in the SHANK2 synaptic scaffolding gene in 51. Weiss LA, Arking DE, Daly MJ, Chakravarti A: A genome-wide linkage and autism spectrum disorder and mental retardation. Nat Genet 2010, association scan reveals novel loci for autism. Nature 2009, 461:802–808. 42:489–491. 52. Anney R, Klei L, Pinto D, Almeida J, Bacchelli E, Baird G, Bolshakova N, 69. Vaags AK, Lionel AC, Sato D, Goodenberger M, Stein QP, Curran S, Ogilvie C, Bölte S, Bolton PF, Bourgeron T, Brennan S, Brian J, Casey J, Conroy J, Ahn JW, Drmic I, Senman L, Chrysler C, Thompson A, Russell C, Prasad A, Correia C, Corsello C, Crawford EL, de Jonge M, Delorme R, Duketis E, Walker S, Pinto D, Marshall CR, Stavropoulos DJ, Zwaigenbaum L, Duque F, Estes A, Farrar P, Fernandez BA, Folstein SE, Fombonne E, Gilbert J, Fernandez BA, Fombonne E, Bolton PF, Collier DA, Hodge JC, Roberts W, Gillberg C, Glessner JT, Green A, et al: Individual common variants exert Szatmari P, Scherer SW: Rare deletions at the neurexin 3 locus in autism weak effects on risk for autism spectrum disorders. Hum Mol Genet 2012, spectrum disorder. Am J Hum Genet 2012, 90:133–141. in press. 70. Sato D, Lionel AC, Leblond CS, Prasad A, Pinto D, Walker S, O'Connor I, 53. Stahl EA, Wegmann D, Trynka G, Gutierrez-Achury J, Do R, Voight BF, Kraft P, Russell C, Drmic IE, Hamdan FF, Michaud JL, Endris V, Roeth R, Delorme R, Chen R, Kallberg HJ, Kurreeman FA, Diabetes Genetics Replication and Huguet G, Leboyer M, Rastam M, Gillberg C, Lathrop M, Stavropoulos DJ, Meta-analysis Consortium; Myocardial Infarction Genetics Consortium, Anagnostou E, Weksberg R, Fombonne E, Zwaigenbaum L, Fernandez BA, Kathiresan S, Wijmenga C, Gregersen PK, Alfredsson L, Siminovitch KA, Roberts W, Rappold GA, Marshall CR, Bourgeron T, Szatmari P, Scherer SW: Worthington J, de Bakker PI, Raychaudhuri S, Plenge RM: Bayesian SHANK1 deletions in males with autism spectrum disorder. Am J Hum inference analyses of the polygenic architecture of rheumatoid arthritis. Genet 2012, 90:879–887. Nat Genet 2012, 44:483–489. 71. Davis LK, Gamazon ER, Kistner-Griffin E, Badner JA, Liu C, Cook EH, Sutcliffe 54. Melhem N, Devlin B: Shedding new light on genetic dark matter. Genome JS, Cox NJ: Loci nominally associated with autism from genome-wide Med 2010, 2:79. analysis show enrichment of brain expression quantitative trait loci but 55. Lee SH, Decandia TR, Ripke S, Yang J, Schizophrenia Psychiatric not lymphoblastoid cell line expression quantitative trait loci. Mol Autism Genome-Wide Association Study Consortium (PGC-SCZ), The International 2012, 3:3. Schizophrenia Consortium (ISC), The Molecular Genetics of Schizophrenia Collaboration (MGS), Sullivan PF, Goddard ME, Keller MC, Visscher PM, doi:10.1186/2040-2392-3-9 Wray NR: Estimating the proportion of variation in susceptibility to Cite this article as: Klei et al.: Common genetic variants, acting Schizophrenia captured by common SNPs. Nat Genet 2012, 44:831. additively, are a major source of risk for autism. Molecular Autism 2012 56. International Schizophrenia Consortium, Purcell SM, Wray NR, Stone JL, 3:9. Visscher PM, O'Donovan MC, Sullivan PF, Sklar P: Common polygenic variation contributes to risk of Schizophrenia and bipolar disorder. Nature 2009, 460:748. 57. Ripke S, Sanders AR, Kendler KS, Levinson DF, Sklar P, Holmans PA, Lin DY, Duan J, Ophoff RA, Andreassen OA, Scolnick E, Cichon S, St Clair D, Corvin A, Gurling H, Werge T, Rujescu D, Blackwood DH, Pato CN, Malhotra AK, Purcell S, Dudbridge F, Neale BM, Rossin L, Visscher PM, Posthuma D, Ruderfer DM, Fanous A, Stefansson H, Steinberg S, et al: Genome-wide association study identifies five new Schizophrenia loci. Nat Genet 2011, 43:969–976. 58. Sullivan PF: The psychiatric GWAS consortium: big science comes to psychiatry. Neuron 2010, 68:182–186. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Molecular Autism Springer Journals

Common genetic variants, acting additively, are a major source of risk for autism

Download PDF
13 pages

Loading next page...
 
/lp/springer-journals/common-genetic-variants-acting-additively-are-a-major-source-of-risk-cbQntfeNig

References (80)

Publisher
Springer Journals
Copyright
Copyright © 2012 by Klei et al.; licensee BioMed Central Ltd.
Subject
Medicine & Public Health; Neurology; Neurosciences; Neuropsychology; Psychiatry; Pediatrics; Human Genetics
eISSN
2040-2392
DOI
10.1186/2040-2392-3-9
pmid
23067556
Publisher site
See Article on Publisher Site

Abstract

Background: Autism spectrum disorders (ASD) are early onset neurodevelopmental syndromes typified by impairments in reciprocal social interaction and communication, accompanied by restricted and repetitive behaviors. While rare and especially de novo genetic variation are known to affect liability, whether common genetic polymorphism plays a substantial role is an open question and the relative contribution of genes and environment is contentious. It is probable that the relative contributions of rare and common variation, as well as environment, differs between ASD families having only a single affected individual (simplex) versus multiplex families who have two or more affected individuals. Methods: By using quantitative genetics techniques and the contrast of ASD subjects to controls, we estimate what portion of liability can be explained by additive genetic effects, known as narrow-sense heritability. We evaluate relatives of ASD subjects using the same methods to evaluate the assumptions of the additive model and partition families by simplex/multiplex status to determine how heritability changes with status. Results: By analyzing common variation throughout the genome, we show that common genetic polymorphism exerts substantial additive genetic effects on ASD liability and that simplex/multiplex family status has an impact on the identified composition of that risk. As a fraction of the total variation in liability, the estimated narrow-sense heritability exceeds 60% for ASD individuals from multiplex families and is approximately 40% for simplex families. By analyzing parents, unaffected siblings and alleles not transmitted from parents to their affected children, we conclude that the data for simplex ASD families follow the expectation for additive models closely. The data from multiplex families deviate somewhat from an additive model, possibly due to parental assortative mating. Conclusions: Our results, when viewed in the context of results from genome-wide association studies, demonstrate that a myriad of common variants of very small effect impacts ASD liability. Keywords: Narrow-sense heritability, Multiplex, Simplex, Quantitative genetics * Correspondence: [email protected] Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA Full list of author information is available at the end of the article © 2012 Klei et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Klei et al. Molecular Autism 2012, 3:9 Page 2 of 13 http://www.molecularautism.com/content/3/1/9 Background Simplex Collection (SSC) [19] and the other from the Beliefs about the genetic architecture of autism Autism Genome Project (AGP) [20]. Importantly the spectrum disorders (ASD) have changed dramatically analysis of these two cohorts allows for an estimate of over the past few decades. Early twin studies produced the heritability of ASD in simplex versus multiplex heritability estimates approaching 90% [1,2] and, while families as well as an assessment of how well the data fit no specific risk loci were known at the time, it was predictions for an additive model of inheritance [21]. believed that liability was conferred by a handful of When all risk variation acts additively, for example, and genes of large effect. Later, data on the distribution of no other forces alter the covariance of relatives, the ASD within families, together with results from linkage liability for relatives of an affected individual consistently analyses, were interpreted to mean that liability arose halves for each degree of separation from the proband. from many genes [3]. Recent work has definitively Therefore, we also evaluate heritability tracing to SSC demonstrated the substantial contribution of de novo and AGP parents and SSC unaffected siblings, evaluat- variation [4-11]. Indeed multiple studies of rare single ing the empirical results against simulation-derived nucleotide and copy number variants (CNVs) have sug- expectations. Finally we use the same techniques to gested that 15% or more of liability traces to de novo ask what residual heritability is contained in what the mutation, effects that are genetic but not inherited [2]. field calls pseudo-controls, which are genotypes formed Importantly, despite notable recent successes in gene from the alleles that parents did not transmit to their discovery efforts, key questions remain regarding the affected offspring. overall nature and scale of the genetic contribution to ASD liability. For example, the contribution of genetics Methods is still debated: a recent large-scale twin study [12] esti- ASD families mated only 38% of liability was accounted for by additive DNA samples from SSC and AGP family members genetic effects, while common environmental factors genotyped on the Illumina Infinium 1Mv3 (duo) micro- accounted for 55% of the variance; whereas most studies array or the Illumina Infinium 1Mv1 microarray were of twins find much higher heritability, including studies analyzed here. Specifically qualifying SSC samples were of phenotypes in the broader spectrum (see [13,14] for genotyped on the Illumina Infinium 1Mv3 (duo) micro- review). Moreover, despite a near-consensus that com- array (71.8%) while most AGP samples were genotyped mon and transmitted variation must confer liability, on the Illumina Infinium 1Mv1 microarray (98.7%). multiple genome-wide association studies have so far Both arrays genotype roughly 1,000,000 single nucleotide not revealed replicable common polymorphisms [15] polymorphisms (SNPs) and the overlap between the associated with ASD, and studies of rare structural and SNP sets is almost perfect. sequence mutations have largely failed to account for The SSC sample [19] includes >2,000 genotyped fam- the anticipated risk associated with transmitted variation ilies. However, our analyses targeted a homogeneous [6,7]. Finally, since the earliest CNV studies in ASD, subset of these data. First, we included only samples it has been postulated that the architecture of simplex genotyped on an Illumina 1M array; families had to be and multiplex autism would be strikingly different [4]. ‘quads’ consisting of an unaffected mother and father, an However not all studies have found marked disparities affected proband and an unaffected sibling; and all mem- in the rate of de novo mutation in simplex versus bers of a quad had to have complete genotypes (>95% multiplex families, and large effect de novo mutations completion rate). Only samples of European ancestry have been characterized in both multiplex and simplex were included. European ancestry for the SSC families families [9,16]. was determined using GemTools [22,23] for all available Consequently, to gain insight into the broad questions SSC probands. To conduct the ancestry analysis we regarding the nature of the genetic factors underlying selected 5,156 SNPs with at least 99.9% calls for geno- ASD, we have estimated how much of the population types, had minor allele frequency MAF >0.05, and were variability in liability can be traced to inherited variation, at least 0.5 Mb apart. Individuals were clustered into specifically the narrow-sense heritability of ASD. Yang nine ancestry groups based on four significant dimen- et al. [17] proposed elegant methods in which the herit- sions of ancestry. The central five clusters, which held ability of liability can be estimated as a function of the a total of 1,686 families, were identified as being of covariance between trait values, in this instance affection European descent. The ancestry cluster information status [18], and the genome wide genetics of the sub- combined with complete genotype information yielded a jects. This contrasts with the usual approach of esti- total of 965 SSC families for the analysis. mating heritability from the distribution of trait values The AGP Stage 1 dataset [16,20] comprised 1,471 in pedigrees. In the present study, these methods are families, of which 1,141 were previously identified to applied to two ASD data sets, one from the Simons be of European ancestry [20]. European ancestry was Klei et al. Molecular Autism 2012, 3:9 Page 3 of 13 http://www.molecularautism.com/content/3/1/9 confirmed by analyses identical to those applied to the MAF > 0.01, and produce a P-value for Hardy-Weinberg SSC families (see Additional file 1: Figure S1). equilibrium > 0.005. Following these QC steps, data from 965 SSC quad families and 1,141 AGP families were ana- Clinical evaluation lyzed using genotypes from 713,259 SNPs. Probands for the SSC and AGP cohorts were diagnosed in a similar manner (for diagnostic protocol for SSC, see Statistical calculations and motivation [19]; for AGP, see [16,20]). All SSC parents were Estimating heritability as a case-control contrast screened for Autism Spectrum Disorder by the Broad Heritability of ASD from probands versus controls was Autism Phenotype Questionnaire [24] (self-report) and estimated using GCTA software [28], which encodes the Social Reciprocity Scale - Adult Research Version the theory laid out in [17,18]. Prevalence of ASD was [25] (informant report). Moreover, family history evalu- taken to be 1% [27]. For each of the analyses, Genetic ation excluded first-, second-, or third-degree relatives Relationship Matrices (GRM) were determined for each who met diagnostic criteria for ASD or intellectual dis- of the 23 chromosomes using the –make-grm option ability. For AGP families 46.2% were known to be multi- in GCTA [28]. These were then combined in an over- plex, another 38.2% were identified as simplex on the all matrix, using the –mgrm option in GCTA. The first basis of a family history indicating no known first- to 10 principal components of ancestry were determined third-degree relatives with ASD, and the remaining 15.6% using –pca in GCTA. These 10 PCA were then used as were of unknown status. Note that most AGP parents covariates for estimating the heritability using –reml in were not systematically evaluated for ASD, unlike those GCTA. A prevalence of 0.01 for autism spectrum dis- from the SSC, and when AGP parents were systematically orders was used to transform the heritability on the evaluated, the results were not used to screen out affected observed scale to the heritability on the liability scale. individuals and thus multiplex families. In addition, while all available SSC family members were genotyped, only The logic of estimating heritability from unaffected parent-proband trios were genotyped for the AGP even family members when additional siblings were available. Due to the screening of SSC samples, no SSC parents would meet criteria for ASD. Given that is the case, what Control subjects is the justification for assigning them to be affected and Controls derived from a convenience sample, specifically contrasting them to controls to estimate the heritability 1,663 individuals from HealthABC [26]. Control samples in the parental generation? Under the additive heritabil- were also genotyped on the Illumina Infinium 1Mv3 ity model parents transmit many genetic variants of (duo) array, like most of the AGP data, providing excel- small effect to their offspring, with the expectation that lent comparability with the case dataset. Moreover, we half would be transmitted from each parent. The parents reasoned that ASD is sufficiently rare (approximately 1% of probands are thus more similar at liability loci than [27]) that screened and unscreened controls would yield expected by chance, and our goal is to estimate this similar results. increased genetic similarity. Calling the parents affected and contrasting their genotypes to that of controls is a Filtering natural approach to estimating their genetic contribution To make heritability estimates comparable, we filtered to liability and it has precedence in quantitative genetics, all families and control subjects based on the follow- such as estimation of the heritability of milk production ing criteria: all were of European descent as determ- from its covariance arising from bulls, when only the ined by genetically-estimated ancestry (Additional file 1: bull’s female progeny give milk (for example [29]). Figure S1); genotypes for all family members met strin- A similar argument follows for unaffected siblings gent quality control (QC) criteria; and control samples from SSC families. These siblings should receive a ran- met identical QC criteria. dom sample of the parent’s genomes and, in expectation, For the three data sets we first chose SNPs genotyped this sampling would include half the liability alleles car- on all platforms. Then ambiguous AT, TA, CG, and GC ried by each parent. Thus the unaffected offspring SNPs were removed. A total of 813,960 SNP across the 22 should mirror the average liability carried by the parents autosomes and chromosome X were included for further and this level can be estimated by calling them affected quality evaluation. At the level of individuals, we required and contrasting their genotypes to those from controls. that genotyping completion rate be greater than 98%, that there be no discrepancy regarding nominal and genotype- Simulations to compute expected heritability for parents inferred sex, and no individuals in different families and pseudo-controls were closely related. At the level of individual SNPs, each While the literature contains numerous references to SNP must have a genotype completion rate > 98%, have the burden of risk variants carried by parents of simplex Klei et al. Molecular Autism 2012, 3:9 Page 4 of 13 http://www.molecularautism.com/content/3/1/9 versus multiplex families, we could not find quantitative 3. Proband and second child are both affected, no genetics analyses of it as a function of ascertainment restriction on the other individuals in the family (there is related work on the impact of multi-locus (unscreened multiplex family); inheritance on the power of candidate gene association 4. A mixture of 60% unscreened simplex families and studies [30,31]). We therefore evaluated the expected 40% unscreened multiplex families. heritability for parents, unaffected siblings, and pseudo- controls on the basis of simulations and the theory of By using rejection sampling, a total of 1000 families quantitative genetics regarding the selection differential were generated for each scenario and this procedure was (for ASD, approximately 1%) and the response to selec- repeated 100 times per scenario and proband heritability tion (expected change in the population’s mean liability). (50 and 75%). To obtain the heritability estimates for the The simulations are designed to mimic ascertainment family members, the average phenotype of the primary for simplex and multiplex families. probands on the liability scale (S) were compared to the One thousand SNPs having an impact on liability average phenotype of the family member of interest on were simulated. The allele frequency for SNP i, p , varied the liability scale (R). The heritability estimate based on ^ R between 0.01 and 0.99. Overall heritability h across all the family member was estimated as h ¼ . Note that n = 1000 SNPs was set to be either 0.50 or 0.75 for we also checked the heritability estimated from the pro- probands with ASD. The relative importance of each bands as a function of the reduction in genetic variance SNP, w , was determined by first selecting a fraction in the selected group. For probands, estimated heritabil- t between 0 and 1 at random using a uniform distri- ity was always close to 50% when that was the desired bution. These 1000 values were added to obtain T, heritability and always close to 75% when that was the and each SNP was weighted by w =t /T. The allele sub- i i desired heritability. stitution effect for each SNP i was then determined as From theoretical considerations we expected assorta- qffiffiffiffiffiffiffiffiffiffiffiffiffiffi w h tive mating to elevate the expected liability of pseudo- a ¼ . For each simulation 1000 families were 2pðÞ 1p i i controls and evaluated its impact by a simple experiment generated consisting of a father, mother, and one child using the simulation structure just described. Rather (AGP simplex) or two children (SSC simplex or AGP than randomly assign genotypes to mates, we first multiplex). Genotypes for the parents were assigned at randomly chose the paternal genotypes at the 1,000 random using the allele frequencies, while children liability SNPs, then assigned maternal genotypes on the received alleles from the parents using the rules of basis of the toss of a fair coin: heads the genotype was Mendelian inheritance. Likewise a pseudo-control was chosen at random, tails it was taken to be the father’s generated by comparing the genotype of the parents genotype. All simulations procedures were as described to that of the proband and assigning the un-transmitted above, except we conducted two simulations: for sim- allele of each parent as the alleles for the pseudo- ulation (a) the heritability of probands from simplex control’s genotype. After all genotypes in a family were families was taken to be 50% and ascertainment followed assigned, the genetic contribution to the underlying scenario 2 above; and for simulation (b) the heritability liability phenotype for each individual j in the family of probands from multiplex families was set to 75% and was determined by G = x a − μ in which x is j i =1 i i G i ascertainment followed scenario 3 above. the allele count for SNP i and μ = p (1 − p )a G i =1 i i i is the average genetic contribution over all genotypes. Robustness of results To simulate the environmental influence on the phe- To evaluate the robustness of the results, 1,986 indivi- notype of individual j, e , we drew a random number duals of European descent from the Neurogenetics from a normal distribution with mean 0 and variance Research Consortium [32] (NGRC) were available through (1-h ). The liability phenotype was then determined as dbGap [33] and used as a second control sample. For y = G + e . Affection status was then assigned based on j j j the NGRC study, genotypes were produced using the not affected when y < 2:326 affection status ¼ Illumina Infinium Human Omni2.5 microarray. There- affected when y ≥ 2:326 fore, to combine all four data sets, we performed QC representing a disease risk of 1% in the population. on 444,200 SNPs genotyped on all platforms, yielding Four different scenarios were simulated: 391,425 SNPs for analyses. 1. Primary child in the family is affected (proband), and father, mother, and designated sibling were Assessing the potential for experimental bias not-affected (SSC family); To explore the impact of different cohorts and genotyp- 2. Proband is affected, no restriction on the other ing protocols on estimated heritability, we conducted a individuals in the family (unscreened simplex family); series of contrasts between SSC and AGP samples of the Klei et al. Molecular Autism 2012, 3:9 Page 5 of 13 http://www.molecularautism.com/content/3/1/9 same relationship type – contrasting probands, mothers, Results and discussion fathers, and pseudo-controls – as well as HealthABC Estimates of heritability (h ) versus NGRC controls. Heritability of SSC probands, measured against HealthABC controls, was found to be 39.6% (Figure 1A, Table 1). SSC mothers, fathers and siblings, when contrasted to controls, Determining genomic coverage yielded an estimated heritability approximately half that of While 713,259 SNPs were used for primary analyses, probands (Figure 1A, Table 1), consistent with expected they constitute a small fraction of the SNPs in the values from theoretical analyses of an additive model human genome. Hence the heritability presented could (Figure 1A). We also generate a “pseudo-control” from the underestimate total heritability. On the other hand, alleles that parents did not transmit to their affected because genotypes of SNPs in close proximity tend to be offspring by using the program Plink [37]. When these correlated due to linkage disequilibrium, it does not pseudo-controls were contrasted to the unrelated control follow that the coverage of the genome by the SNPs sample they produce estimates roughly one-quarter of used here estimate only a small fraction of the heritabil- that identified in probands and close to the theoretical ity. To determine the shortfall in “genomic coverage” expectation, zero (Figure 1A), demonstrating that the and how it impacts estimates of heritability, we per- probands received the majority of risk alleles carried formed an experiment using data from the 1,000 by parents. Genomes project [34], under the assumption that cover- When heritability is estimated using AGP probands age of common variants in the 1,000 Genomes data (Figure 1B, Table 1), the point estimates are larger than is perfect. Assessing all SNPs genotyped in our data, those from SSC (h =55.2% versus 39.6%) although the as well as subsets thereof, we estimated heritability 95% confidence intervals overlap. Moreover the decline of liability. Using the same subsets, but in 1,000 Gen- in heritability for AGP parents relative to probands is omes subjects, we estimated levels of genomic coverage. 30% (55% for probands, 37% for parents), instead of the We can then relate estimated heritability to genomic 50% seen for SSC, and heritability estimated from coverage to develop a functional relationship between pseudo-controls is also higher (38%), consistent with the two. parental values (Figure 1B, Table 1). These results sug- We performed the experiment assessing “genomic gest that AGP parents carry a greater load of additive coverage” as follows. We assumed genomic coverage of risk variants than SSC parents and thus are, on average, SNPs with MAF > 0.1 would be essentially complete for closer to the threshold of being affected. the 379 European samples analyzed by the 1,000 A major difference between the SSC and AGP samples Genomes project. From these genomes we selected 50 was the ascertainment and assessment process. SSC par- 1Mb regions in which at least 500 SNPs in the 1,000 ents were systematically screened on two instruments to Genomes samples had MAF > 0.10. Coverage of these ensure they did not meet criteria for a spectrum diagno- regions by the 713,259 SNPs was calculated as a func- sis. Most parents from AGP families were not evaluated tion of the number of other SNPs with MAF > 0.1 that in this way, and a small fraction of those parents met were tagged by (correlated with) them; call the set of criteria for ASD [9,16]. While not as systematic as the M = 713,259 SNPs “tagSNP”. The tagging evaluation SSC phenotyping assessment, most AGP families did was implemented using Hclust [35]. Forcing tagSNP to have available information about simplex versus multi- be in the set of selected tag SNPs from the region, plex status. Consequently, we were able to compare Hclust evaluated how many more independent SNPs heritability of probands from AGP multiplex versus sim- N were required to cover the region when the min- plex families (Figure 1D, Table 2). The former was esti- imum linkage disequilibrium [36] r amongst tags could mated at 65.5% by comparison to HealthABC, whereas be no less than X,where X = {0.5, 0.7, and 0.9}. Then, probands for AGP simplex families it was 49.8%. Thus for each value of X, M/(M+N) estimates the coverage. estimates of heritability for AGP simplex probands Next we randomly sampled 50, 25 and 12.5% of the are somewhat closer to those from SSC probands 713,259 SNPs (356,630, 178,315, and 89,158 SNPs (Figure 1C) than to estimates for AGP multiplex pro- respectively) five times and each time estimated coverage bands. Moreover, for multiplex families and the mixed for these subsets. set of AGP families (simplex/multiplex/unknown), both the observed and expected heritability for first-degree relatives was higher than that seen in simplex families Human subjects research statement (Figure 1). These results comport with the literature The research described here is in compliance with the showing that unaffected relatives from multiplex families Helsinki Declaration, including appropriate informed tend to exhibit more features of the broader autism consent or assent [16,19,20,26,32,33]. phenotype than relatives in simplex families [38-40] (see Klei et al. Molecular Autism 2012, 3:9 Page 6 of 13 http://www.molecularautism.com/content/3/1/9 A B Pr Mo Fa Si Pc Pr Mo Fa Si Pc C D Pr Mo Fa Si Pc Pr Mo Fa Si Pc Figure 1 Estimated heritability for Autism Spectrum Disorders from ASD probands (Pr), as well as for their mothers (Mo), fathers (Fa), siblings (Si) and pseudo-controls (Pc). Blue dotted reference line is set to the estimated heritability from probands; the black line marks the expected heritability for first degree relatives; and the gray line marks the expected heritability from pseudo-controls. Expected values derived from simulations mimicking the recruitment strategy producing the samples for (A)-(D). (A) Simons Simplex Collection or SSC data; (B) Autism Genome Project or AGP data; (C) AGP data, only simplex families; (D) AGP data, only multiplex families. Additional file 2: Table S1 for estimates from combined estrogen/testosterone balance) in the face of a similar simplex samples). degree of genetic risk. A curious observation from AGP multiplex families Our results support either the first or second hypo- was that fathers generate larger heritability than theses but are not consistent with the third. The first mothers. We reasoned that this could be explained by hypothesis is impossible to rule out given the limited three plausible hypotheses: (1) the confidence intervals sample size. For the second hypothesis, if AGP fathers of the paternal and maternal estimates overlap, so there were simply carrying greater risk, some of those is no true difference; (2) the load of risk variants is, in additional risk alleles would be carried by the pseudo- fact, greater for AGP fathers; or (3) fathers carry a larger controls and the heritability obtained from the contrast number of both liability and protective alleles. The last of probands and pseudo-controls should be substantially of these requires some elaboration. Males are at much smaller than that observed from probands versus greater risk for ASD than females (4:1 or greater) and controls. Indeed the values are substantially smaller: parents carry additive risk factors, yet AGP fathers and 10.9% vs. 39.6% for SSC; 14.5% vs. 55.2% for all mothers are largely unaffected. It is possible, then, that AGP; 0.0% vs. 49.8% for simplex AGP, and 27.1% vs. the increased allele sharing in unaffected fathers is due 65.5% for multiplex AGP. Finally, if (3) were true, to a greater proportion of protective alleles, with females then contrasting probands to pseudo-controls would being resilient for some other reason (for example, produce substantial estimates of heritability because of Table 1 Heritability estimates and their standard errors (se) based on contrasts to HealthABC controls using genotypes from 713,259 SNPs SSC AGP Simplex All Simplex Multiplex Estimate se Estimate se Estimate se Estimate se Probands 0.396 0.082 0.552 0.068 0.498 0.118 0.655 0.139 Mothers 0.199 0.082 0.371 0.070 0.314 0.119 0.377 0.141 Fathers 0.196 0.084 0.370 0.070 0.352 0.119 0.666 0.143 Siblings 0.158 0.082 – – –– –– Pseudo controls 0.090 0.082 0.381 0.070 0.317 0.120 0.503 0.146 Heritability 0.0 0.4 0.8 0.0 0.4 0.8 0.0 0.4 0.8 0.0 0.4 0.8 Klei et al. Molecular Autism 2012, 3:9 Page 7 of 13 http://www.molecularautism.com/content/3/1/9 Table 2 Heritability estimates and their standard errors (se) based on contrasts to HealthABC and NGRC controls using genotypes from 391,425 SNPs SSC AGP HealthABC NGRC HealthABC NGRC Estimate se Estimate se Estimate se Estimate se Probands 0.395 0.082 0.378 0.073 0.553 0.068 0.586 0.063 Mothers 0.200 0.082 0.232 0.074 0.371 0.070 0.342 0.065 Fathers 0.196 0.084 0.153 0.073 0.373 0.070 0.518 0.063 Siblings 0.158 0.082 0.170 0.073 –– –– Pseudo controls 0.090 0.082 0.107 0.073 0.380 0.070 0.446 0.065 the differentiation induced by protective alleles, but this families, heritability estimated from X is less than that pre- is not observed. dicted by its size. This is noteworthy because chromosome X has been cited as a possible source of sex-differential li- Distribution of liability alleles in the genome ability for ASD [41]. Our results suggest that common var- If the additive variation for liability to ASD conforms iants affecting liability do not cluster on chromosome X. to the traditional polygenic or infinitesimal model, then liability variants should be distributed at random over the genome. The implication is that if heritability were Evaluating robustness of results estimated for each chromosome, the resulting estimates To evaluate the robustness of our results, we first should be correlated with the lengths of the chromo- contrasted the genotypes of SSC and AGP probands to somes. On the other hand, if the heritability traced to a a second large set of controls, 1,986 individuals from relatively small number of variants, even a few dozen, the Neurogenetics Research Consortium [32,33]. These such a correlation would be unlikely. In fact, we observe samples, genotyped on the Illumina Infinium Human significant correlation between per-chromosome herit- Omni2.5, were filtered and subjected to QC in an identi- ability and chromosome length (Figure 2), both for sim- cal fashion to the HealthABC control set. There was ex- plex (r = 0.46, P value = 0.028) and multiplex (r= 0.54, cellent agreement of heritability estimates for ASD from P value = 0.0075) families. the two control samples (Tables 2 and 3) despite differ- In Figure 2the deviationfrompredictionfor chromo- ences in ascertainment of the controls and the different some X is surprising. For both multiplex and simplex genotyping platforms. Multiplex Simplex 50 100 150 200 250 300 Chromosome length (cM) Figure 2 Estimated heritability per chromosome for simplex and muliplex families. In this figure chromosome X is marked distinctly, but each chromosome is mapped by its length. Heritability 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Klei et al. Molecular Autism 2012, 3:9 Page 8 of 13 http://www.molecularautism.com/content/3/1/9 Next, the impact of different cohorts and genotyping dimensional space of allele frequencies, phenotypes and platforms on estimates of heritability was explored by their interrelationships. Therefore even if two controls conducting a series of contrasts between SSC and AGP groups evoke similar estimates of ASD heritability from samples of the same relationship type: contrasting the same sample of probands, the controls themselves probands, mothers, fathers, and pseudo-controls. Note need not be close in the multidimensional space of allele that most SSC samples were genotyped on the Illumina frequencies. What generates the differentiation between 1Mv3 (duo) microarray (71.8%) while most AGP controls is unknown. It could arise from the different samples were genotyped on the Illumina Infinium genotyping platforms or from differences in ascertain- 1Mv1 microarray (98.7%). Contrasts between SSC and ment. In light of this difference, the fact that both con- AGP samples of the same relationship type (Additional trols sets give rise to nearly identical estimates of file 3: Table S2) produce estimates close to the difference heritability for all proband subsets is remarkable and between their control-based heritability. Indeed the suggests that the similarity amongst cases overwhelms estimates from direct contrasts were usually smaller than differences between the controls. the difference of control-based heritability (for probands, 0.08 vs. 0.15 ≈ 0.552-0.396 from Table 1; for mothers, Heritability of pseudo-controls 0.11 vs. 0.17; for fathers, 0.19 vs. 0.17; and for pseudo- There remains an unexplained feature of the results: controls, 0.22 vs. 0.29). Thus these results are not estimates of heritability for pseudo-controls tend to be consistent with effects attributable to genotyping plat- elevated over their theoretical values (Figure 1). Several form or ascertainment beyond multiplex/simplex status. genetic forces could be at play. The simulations to derive Implicit in these results is common genetic liability - the distribution of liability in families also produce esti- SSC and AGP probands must share many liability mates for pseudo-controls. Those results show (Figure 1) variants despite their differences in ascertainment. that while the expected heritability for simplex families Indeed when AGP multiplex probands are contrasted to is zero, multiplex status raises the expected value to SSC probands the resulting heritability is 0.23, quite 20%. It is not unreasonable to assume that the simplex similar to that expected by the difference in their esti- collections analyzed here contain families with unreal- mated heritability (0.66 - 0.40 = 0.26); and when AGP ized multiplex potential, and that might be especially simplex probands are contrasted to SSC probands, the true for AGP families that had ascertainment criteria less resulting estimated heritability, 0.0, is below that of the dif- stringent than those for SSC families. ference in their estimated heritability (0.50 - 0.40 = 0.10). A factor that will elevate the expected heritability in These results suggest that the difference between multi- pseudo-controls is positive assortative mating (hence- plex and simplex families is largely a matter of degree (see forth assortative mating). Assortative mating on pheno- also [42]), namely the number of liability alleles carried by types related to ASD liability has been previously parents, rather than a fundamental difference in the gen- reported [39]. When parents are genetically similar at etic architecture [4,43]. liability loci and they bear affected offspring, their Given the remarkable similarities of heritability esti- gametes will tend to be highly enriched for risk alleles, mates obtained for either set of control samples (Tables 2 even those that are not transmitted to affected offspring. and 3), one might anticipate there would be little, if Simple simulations mimicking assortative mating show any, difference between these controls. When we con- that it can exert an impact similar to the difference be- trasted these control samples, however, they produced a tween simplex and multiplex status. When simplex pro- heritability of 26.5% (Additional file 3: Table S2). Math- bands had heritability of 50% (that is, simulation a in ematically, estimates of heritability arise from a high Methods), the expected heritability of pseudo-controls Table 3 Heritability estimates and their standard errors (se) based on contrasts to HealthABC and NGRC controls using genotypes from 391,425 SNPs but separating the AGP data into multiplex and simplex families for estimation AGP multiplex AGP simplex HealthABC NGRC HealthABC NGRC Estimate se Estimate se Estimate se Estimate se Probands 0.650 0.139 0.710 0.140 0.503 0.117 0.494 0.114 Mothers 0.369 0.141 0.387 0.136 0.311 0.119 0.268 0.117 Fathers 0.664 0.143 0.693 0.140 0.359 0.119 0.520 0.113 Pseudo controls 0.497 0.146 0.524 0.140 0.323 0.120 0.438 0.117 Klei et al. Molecular Autism 2012, 3:9 Page 9 of 13 http://www.molecularautism.com/content/3/1/9 was 11.3% – versus 0% without assortative mating. direct estimate of the proportion of liability attribut- When multiplex probands had heritability of 75%, the able to additive genetic effects, whereas twin studies ob- expected heritability of pseudo-controls was 42.8% – tain their estimates by relying on assumptions that are versus 20.2% without assortative mating. These simple approximations. For example, Zuk et al. [45] point out experiments were not intended to cover the range of that non-additive genetic effects are almost surely a plausible scenarios for assortative mating relevant to component of the genetic architecture of any trait, but ASD, which would be impossible, but rather to demon- these effects cannot be captured by twin designs. Yet strate the effect of such mating on the nature of pseudo- for autism and other psychiatric disorders non-additive controls. Thus assortative mating could be an important genetic effects could be an integral component [46-48]. and salient source of enrichment. Whether these forces Twin designs also fail to capture other features, explain all of the elevated heritability for pseudo- such as maternal effects [49] and de novo mutations, controls will require further data and analyses. which are an important component of ASD genetic architecture [4-11]. Impact of genome coverage A recent ASD twin study [12] estimates 38% of ASD Because the set of SNPs used for primary analyses con- liability traces to additive genetic effects while 55% stitute a small fraction of the SNPs in the human gen- traces to common environment. Our point estimates ome, estimates of heritability (Figure 1) could be biased would be close to theirs if ascertainment of their families downward. Still, due to linkage disequilibrium, the was like that for SSC families, but not like that for AGP degree of bias is not trivial to estimate. Therefore families. A substantial fraction of their dizygotic twins, we performed an experiment to evaluate the shortfall however, are multiplex for ASD. Thus their point esti- in genomic coverage and how it impacts estimates of mate for heritability from additive genetic effects is low heritability. Results from the experiment are shown relative to ours. If rare inherited variation contributes in Additional file 4: Figure S2, in which estimated herit- substantially to liability for ASD, this makes the 38% ability was plotted against estimated coverage. These estimate seem lower still because twin studies should results suggest that heritability estimates from probands, capture these effects whereas our estimates cannot. as shown in Figure 1, are good approximations. They Genomewide association studies [18,50-52] have represent only slight underestimates of what would be detected only a handful of SNPs, all of small effect and obtained had the entire genome been sampled. none replicating reliably. Teaming this observation with In total our results demonstrate that a substantial our estimates of heritability (Figure 1) and the fact that portion of ASD liability arises from inherited variation these studies are underpowered to detect genetic var- acting additively. This pattern holds both for simplex iants of small effect size, but are otherwise well powered and multiplex families, with the burden of liability [15], we conclude there must be thousands of SNPs scat- greater in multiplex families, consistent with theoretical tered across the genome with common liability alleles. and empirical [38-40] results. The modeling reported Analyses of chromosome-specific heritability support here does not differentiate between additive effects due this conclusion (Figure 2). Employing analyses like those to common versus rare variation. Nonetheless it is rea- proposed by Stahl et al. [53] could estimate this distribu- sonable to assume that most of the estimated heritability tion of effects. traces to common variants because linkage disequilib- Because these loci have small effect, samples far larger rium between the common variants analyzed and rare than exist today will be required to identify a substantial liability variants should, on average, be small [44]. Thus fraction of them using standard genome-wide asso- the additive contribution of rare variants to ASD liability ciation methodology. Hence, for the immediate future, is likely underestimated. Imperfect coverage must also ample “missing heritability” for ASD will remain. Ingeni- ous designs will be required in the near term [54] have an impact, but our analyses suggest its impact is not large (Additional file 4: Figure S2). to identify SNPs affecting liability. In the longer Our analyses cannot address other features of the gen- term GWAS of a large number of ASD subjects, at least on the order of that performed for schizophrenia etic architecture of ASD, including non-additive genetic effects, which add to ASD’s broad-sense heritability [45], [55-57], should be one of the priorities for the field of and de novo mutations. In addition, because they under- ASD genetics. One way forward is to exploit shared liability across estimate the impact of rare inherited variation, they dif- fer from family-based estimates, such as from twin psychiatric disorders, taking advantage of larger samples studies, that do capture these effects. Still our findings of [58] afforded by cross-disorder meta-analysis. There is now sound evidence for common variants affecting substantial heritability are consistent with the majority of twin studies [1,2] and are richer in some ways because liability for schizophrenia [55-57], including a study the analytic technique [17,18] used here provides a similar to ours [46]. Given the documented sharing of Klei et al. Molecular Autism 2012, 3:9 Page 10 of 13 http://www.molecularautism.com/content/3/1/9 rare variants affecting risk for both disorders (for ex- study that reveals a plausible biological link to ASD ample [59]), it would not be surprising to find that some liability [67]. common variants affect liability to both schizophrenia The genetic architecture of ASD has numerous compo- and ASD. nents: additive, non-additive and de novo genetic effects, as The estimated heritability for schizophrenia using well as gene-gene and gene-environment interactions. The methods similar to ours is 23% [46]; for bipolar disorder results shown here are relevant to only one of these compo- and similar methods it is 40% [60]; and for major de- nents. Other components, such as de novo events, are also pression it is 32% [61]. None of these studies separate known to make a substantial contribution to liability [4-11], out simplex and multiplex families, so in that sense whileothersremain tobethoroughlyinvestigated[45]. they are most comparable to the estimate obtained over Already analyses of rare variation of major effect has all AGP families, 55%, although the representation of revealed a substantial number of genes affecting liability multiplex families in the AGP sample is likely larger [8-11,68-70]; it is reasonable to predict that common var- than for the other samples. Regardless of the differences iants regulating expression of those ASD genes could also in simplex/multiplex representation, these estimates affect liability [71]. We hypothesize that the interplay of are stochastically similar, in view of their standard rare and common variants is critical not only to liability it- errors, emphasizing that common variants affect liability self, but to the expression of ASD or other relevant psychi- for most if not all psychiatric disorders. Moreover their atric and developmental disorders. The dynamics of this impact appears to be similar in magnitude across dis- interplay will likely be an important area for future autism orders, as measured by heritability estimated from research. common variants. That ASD shows the largest estimated heritability is Conclusions notable and could reflect the fact that the sibling recur- Common genetic polymorphisms exert substantial additive rence risk is, on average, higher for siblings of an ASD genetic effects on ASD liability and their impact differs by proband than for siblings of probands diagnosed with ascertainment strategies used to recruit families. For sim- schizophrenia, bipolar disorder or major depression. Sib- plex families, who have only a single affected individual in ling recurrence risk is a ratio, defined as the probability multiple generations, approximately 40% of liability traces of a sibling being affected, given that the proband is to additive effects whereas this narrow-sense heritability affected, divided by the prevalence of the disorder in the exceeds 60% for ASD individuals from multiplex families. general population. Recent studies put this recurrence Data for simplex ASD families follow the expectation for risk at almost 20 for ASD [62], whereas for schizophre- additive models closely. Data from multiplex families nia it is 6 to 10 fold [63], for bipolar disorder it is 4 to 10 deviate somewhat from an additive model. This result is fold [64], and for major depression it is roughly twofold consistent with what would beexpectedfrom positiveas- [64]. The larger heritability could also trace to differences sortative mating, but our data do not prove such a among studies. It is possible that our estimates of herit- pattern of mating occurred. In light of results from ability are inflated by unknown differences between our genome-wideassociation studies, theremust bemanycom- case and control samples, including ascertainment biases mon variants of very small effect affecting liability to ASD. and genotype quality. Regarding the latter, we selected case and control samples genotyped on the same geno- Availability of supporting data typing platform to minimize differences and we did not The data sets supporting the results of this article detect any large differences in allele frequencies, but we are available in the repositories: Simons Foundation cannot rule out subtle differences in quality. Autism Research Initiative, SFARI [http://sfari.org/sfari- Regarding identification of common variants affecting initiatives/simons-simplex-collection]; and the National liability, our results suggest that the contrast of case and Institutes of Health database of Genotypes and Pheno- pseudo-control genotypes, the “family-based” analysis, is types, dbGaP [http://www.ncbi.nlm.nih.gov/gap]. not optimal. In many samples pseudo-controls carry a substantial burden of risk variants and their presence Additional files degrades the power of family-based analysis to detect risk SNPs (see also [30,31]). Instead it appears that Additional file 1: Figure S1. Ancestry projects for principal component 1 (PC.1) versus principal component 2 (PC.2) for the samples used in the population-based controls contrasted with ASD cases analysis of heritability. Red dots represent subjects with an ASD diagnosis would be a more powerful design [65], even after and blue are controls. HealthABC=HABC. adjusting for ancestry [66]. In this regard it is intriguing Additional file 2: Table S1. Heritability estimates and their standard that the earliest GWAS of ASD [50] used population- errors (se) using 391,425 SNP when AGP and SSC simplex family data are combined or only multiplex AGP families are analyzed. Analyses include based controls to identify a single locus at 5p14.1, and all HealthABC and NGRC control samples. this result has since garnered support from a functional Klei et al. Molecular Autism 2012, 3:9 Page 11 of 13 http://www.molecularautism.com/content/3/1/9 Swedish Science Council; The Centre for Applied Genomics, Canada; Utah Additional file 3: Table S2. Heritability estimates and their standard Autism Foundation, USA; Core award 075491/Z/04. Wellcome Trust, UK. errors (se) obtained when contrasting AGP and SSC samples of the same Genotype and phenotype data were obtained from dbGap, as provided by relationship type, as well as contrasting HealthABC versus NGRC controls. AGP Study Investigators. Additional file 4: Figure S2. Heritability for ASD probands as a function HealthABC: These controls were obtained from Database for Genotypes and of estimated “genomic coverage” for varying levels of r . Coverage is Phenotypes (dbGap) at http://www.ncbi.nlm.nih.gov/gap. Funding support for the estimated as the fraction of all known SNPs identified by 1000 Genomes “CIDR Visceral Adiposity Study” (Study accession number: phs000169.v1.p1) was with minor allele frequency > 0.1 tagged by the set of SNPs used to provided through the Division of Aging Biology and the Division of Geriatrics and estimate heritability for probands; see Methods for more details. From Clinical Gerontology, NIA. The CIDR Visceral Adiposity Study includes a genome- the left points map onto 12.5%, 25%, 50%, and 100% of the SNPs used to wide association study funded as part of the Division of Aging Biology and the estimate heritability. Top line is for probands from multiplex families, Division of Geriatrics and Clinical Gerontology, NIA. Assistance with phenotype bottom for probands from simplex families. harmonization and genotype cleaning, as well as with general study coordination, was provided by Heath ABC Study Investigators. NGRC: We also used the NINDS dbGaP database from the CIDR: NGRC Abbreviations Parkinson’s Disease Study (dbGap accession number phs000196.v2.p1). The AGP: Autism Genome Project; ASD: Autism Spectrum Disorders; CNVs: Copy genetic arm of the study has been funded by NIH since 1998 (R01 NS36960, Number Variants; GCTA: Genome-Wide Complex Trait Analysis, Software used Haydeh Payami, PI). In 2004, the consortium was formalized as a Michael J to estimate heritability, amongst others; GRM: Genetic Relationship Matrices; Fox Foundation Funded Global Genetic Consortium, and an epidemiologic HealthABC: A sample of subjects used as controls and genotyped on the ® arm was implemented. Genotype and phenotype data were obtained from Illumina Infinium 1Mv3 (duo) array; MAF: Minor Allele Frequency; dbGap, as provided by NGRC Parkinson’s Disease Study Investigators. NGRC: Neurogenetics Research Consortium, a sample of subjects used as ® For both the HealthABC and NGRC studies, genotyping services were controls and genotyped on the Illumina Infinium Human Omni2.5 provided by the Center for Inherited Disease Research (CIDR). CIDR is funded microarray; QC: Quality Control; SNPs: Single Nucleotide Polymorphisms; through a federal contract from the National Institutes of Health to The SSC: Simons Simplex Collection. Johns Hopkins University, contract number HHSN268200782096C and HHSN268201100011I. Competing interests The authors declare no competing financial interests. Author details Authors’ contributions Department of Psychiatry, University of Pittsburgh School of Medicine, MWS supervised the overall project, EHC its phenotypic portions; LK, KR and Pittsburgh, Pennsylvania, USA. Program on Neurogenetics, Yale University BD conceived of the analyses; LK implemented the analyses; EHC, KR, MWS, School of Medicine, New Haven, Connecticut, USA. Child Study Center, Yale SJS, and BD wrote the first draft of the manuscript; all others authors University School of Medicine, New Haven, Connecticut, USA. Department commented on and refined it. Most authors recruited families, produced or of Psychiatry, Yale University School of Medicine, New Haven, Connecticut, evaluated data and commented on the manuscript. All authors read and USA. Department of Genetics, Yale University School of Medicine, New approve the final manuscript. Haven, Connecticut, USA. Department of Psychology, University of Michigan, Ann Arbor, MI, USA. Neurogenetics Program, Department of Neurology and Center for Autism Research and Treatment, Semel Institute, David Geffen Acknowledgments School of Medicine, University of California Los Angeles, Los Angeles, Research supported by grants from the Simons Foundation and MH057881. California, USA. Department of Human Genetics, Emory University School of SSC: We are grateful to all of the families participating in the Simons Medicine, Atlanta, Georgia, USA. Division of Genetics, Children's Hospital Foundation Autism Research Initiative (SFARI) Simplex Collection (SSC). This Boston, Harvard Medical School, Boston, Massachusetts, USA. Department work was supported by a grant from the Simons Foundation. We wish to of Psychiatry, McGill University, Montreal Children's Hospital, Montreal, QC thank the SSC principal investigators A.L. Beaudet, R. Bernier, J. Constantino, H3Z 1P2, Canada. Department of Psychiatry, Mount Sinai School of E.H. Cook, Jr., E. Fombonne, D. Geschwind, D.E. Grice, A. Klin, D.H. Ledbetter, Medicine, New York, New York, USA. Geisinger Health System, Danville, C. Lord, C.L. Martin, D.M. Martin, R. Maxim, J. Miles, O. Ousley, B. Peterson, Pennsylvania, USA. Center for Autism and the Developing Brain, Weill J. Piggot, C. Saulnier, M.W. State, W. Stone, J.S. Sutcliffe, C.A. Walsh, and Cornell Medical College, White Plains, New York, USA. Yale Center for E. Wijsman; the coordinators and staff at the SSC sites; the SFARI staff, in Genome Analysis, Orange, Connecticut, USA. Departments of Pediatrics and particular M. Benedetti; Prometheus Research; the Yale Center of Genomic Human Genetics, The University of Michigan Medical Center, Ann Arbor, Analysis staff, in particular M. Mahajan, S. Umlauf, I. Tikhonova and A. Lopez; Michigan, USA. Department of Molecular Biology, Cell Biology and T. Brooks-Boone, N. Wright-Davis and M. Wojciechowski for their help in Biochemistry, Brown University, Providence, Rhode Island, USA. Department administering the project at Yale; I. Hart for support; and G.D. Fischbach, of Psychiatry and Human Behavior, Brown University, Providence, Rhode A. Packer, J. Spiro, M. Benedetti and M. Carlson for their helpful suggestions Island, USA. Howard Hughes Medical Institute and Division of Genetics, throughout. Approved researchers can obtain the SSC population data set Children's Hospital Boston, and Neurology and Pediatrics, Harvard Medical described in this study by applying at https://base.sfari.org. School Center for Life Sciences, Boston, Massachusetts, USA. Department of AGP: We used data from the Autism Genome Project (AGP) Consortium - Molecular Physiology & Biophysics, Center for Molecular Neuroscience, Whole Genome Association and Copy Number Variation Study of over 1,500 Vanderbilt University, Nashville, Tennessee, USA. Institute for Juvenile Parent-Offspring Trios - Stage I (dbGaP Study Accession: phs000267.v1.p1). Research, Department of Psychiatry, University of Illinois at Chicago, Chicago, Funding for AGP was provided from National Institutes of Health (HD055751, Illinois, USA. Department of Statistics, Carnegie Mellon University, HD055782, HD055784, HD35465, MH52708, MH55284, MH57881, MH061009, Pittsburgh, Pennsylvania, USA. MH06359, MH066673, MH080647, MH081754, MH66766, NS026630, NS042165, NS049261); The Canadian Institutes for Health Research (CIHR); Received: 20 August 2012 Accepted: 4 October 2012 Assistance Publique - Hôpitaux de Paris, France; Autism Speaks UK; Canada Published: 15 October 2012 Foundation for Innovation/Ontario Innovation Trust; Grant: Po 255/17-4. Deutsche Forschungsgemeinschaft, Germany; EC Sixth FP AUTISM MOLGEN; Fundação Calouste Gulbenkian, Portugal; Fondation de France; Fondation References FondaMental, France; Fondation Orange, France; Fondation pour la 1. Bailey A, Le Couteur A, Gottesman I, Bolton P, Simonoff E, Yuzda E, Recherche Médicale, France; Fundação para a Ciência e Tecnologia, Portugal; Rutter M: Autism as a strongly genetic disorder: evidence from a British The Hospital for Sick Children Foundation and University of Toronto, Canada; twin study. Psychol Med 1995, 25:63–77. INSERM, France; Institut Pasteur, France; Convention 181 of 19.10.2001. Italian 2. Devlin B, Scherer SW: Genetic architecture in autism spectrum disorder. Ministry of Health; John P Hussman Foundation, USA; McLaughlin Centre, Curr Opin Genet Dev 2012, 22:229–237. Canada; Rubicon 825.06.031. Netherlands Organization for Scientific Research; 3. Risch N, Spiker D, Lotspeich L, Nouri N, Hinds D, Hallmayer J, Kalaydjieva L, TMF/DA/5801. Royal Netherlands Academy of Arts and Sciences; Ontario McCague P, Dimiceli S, Pitts T, Nguyen L, Yang J, Harper C, Thorpe D, Ministry of Research and Innovation, Canada; Seaver Foundation, USA; Vermeer S, Young H, Hebert J, Lin A, Ferguson J, Chiotti C, Wiese-Slater S, Klei et al. Molecular Autism 2012, 3:9 Page 12 of 13 http://www.molecularautism.com/content/3/1/9 Rogers T, Salmon B, Nicholas P, Petersen PB, Pingree C, McMahon W, copy number variation in autism spectrum disorders. Nature 2010, Wong DL, Cavalli-Sforza LL, Kraemer HC, et al: A genomic screen of 466:368–372. autism: evidence for a multilocus etiology. Am J Hum Genet 1999, 17. Yang J, Benyamin B, McEvoy BP, Gordon S, Henders AK, Nyholt DR, 65:493–507. Madden PA, Heath AC, Martin NG, Montgomery GW, Goddard ME, 4. Sebat J, Lakshmi B, Malhotra D, Troge J, Lese-Martin C, Walsh T, Yamrom B, Visscher PM: Common SNPs explain a large proportion of the heritability Yoon S, Krasnitz A, Kendall J, Leotta A, Pai D, Zhang R, Lee YH, Hicks J, for human height. Nat Genet 2010, 42:565–569. Spence SJ, Lee AT, Puura K, Lehtimäki T, Ledbetter D, Gregersen PK, 18. Lee SH, Wray NR, Goddard ME, Visscher PM: Estimating missing heritability Bregman J, Sutcliffe JS, Jobanputra V, Chung W, Warburton D, King MC, for disease from genome-wide association studies. Am J Hum Genet 2011, Skuse D, Geschwind DH, Gilliam TC, Ye K, et al: Strong association of de 88:294–305. novo copy number mutations with autism. Science 2007, 316:445–449. 19. Fischbach GD, Lord C: The Simons simplex collection: a resource 5. Marshall CR, Noor A, Vincent JB, Lionel AC, Feuk L, Skaug J, Shago M, for identification of autism genetic risk factors. Neuron 2010, Moessner R, Pinto D, Ren Y, Thiruvahindrapduram B, Fiebig A, Schreiber S, 68:192–195. Friedman J, Ketelaars CE, Vos YJ, Ficicioglu C, Kirkpatrick S, Nicolson R, 20. Anney R, Klei L, Pinto D, Regan R, Conroy J, Magalhaes TR, Correia C, Sloman L, Summers A, Gibbons CA, Teebi A, Chitayat D, Weksberg R, Abrahams BS, Sykes N, Pagnamenta AT, Almeida J, Bacchelli E, Bailey AJ, Thompson A, Vardy C, Crosbie V, Luscombe S, Baatjes R, et al: Structural Baird G, Battaglia A, Berney T, Bolshakova N, Bölte S, Bolton PF, Bourgeron T, variation of chromosomes in autism spectrum disorder. Am J Hum Genet Brennan S, Brian J, Carson AR, Casallo G, Casey J, Chu SH, Cochrane L, 2008, 82:477–488. Corsello C, Crawford EL, Crossett A, et al: A genome-wide scan for 6. Levy D, Ronemus M, Yamrom B, Lee YH, Leotta A, Kendall J, Marks S, common alleles affecting risk for autism. Hum Mol Genet 2010, Lakshmi B, Pai D, Ye K, Buja A, Krieger A, Yoon S, Troge J, Rodgers L, 19:4072–4082. Iossifov I, Wigler M: Rare de novo and transmitted copy-number variation 21. Falconer DS: Introduction to Quantitative Genetics. London: Longman; 1981. in autistic spectrum disorders. Neuron 2011, 70:886–897. 22. Lee AB, Luca D, Klei L, Devlin B, Roeder K: Discovering genetic ancestry 7. Sanders SJ, Ercan-Sencicek AG, Hus V, Luo R, Murtha MT, Moreno-De-Luca D, using spectral graph theory. Genet Epidemiol 2009, 34:51–59. Chu SH, Moreau MP, Gupta AR, Thomson SA, Mason CE, Bilguvar K, 23. Klei L, Kent BP, Melhem N, Devlin B, Roeder K: GemTools: a fast and Celestino-Soper PB, Choi M, Crawford EL, Davis L, Wright NR, Dhodapkar RM, efficient approach to estimating genetic ancestry; 2011. http://arxiv.org/pdf/ DiCola M, DiLullo NM, Fernandez TV, Fielding-Singh V, Fishman DO, Frahm S, 1104.1162.pdf. Garagaloyan R, Goh GS, Kammela S, Klei L, Lowe JK, Lund SC, et al: Multiple 24. Hurley RS, Losh M, Parlier M, Reznick JS, Piven J: The broad autism recurrent de novo CNVs, including duplications of the 7q11.23 Williams phenotype questionnaire. J Autism Develop Dis 2007, 37:1679–1690. syndrome region, are strongly associated with autism. Neuron 2011, 25. Constantino JN, Gruber CP: The Social Responsiveness Scale manual. 70:863–885. Los Angeles, CA: Western Psychological Services; 2005. 8. Sanders SJ, Murtha MT, Gupta AR, Murdoch JD, Raubeson MJ, Willsey AJ, 26. HealthABC data. http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study. Ercan-Sencicek AG, DiLullo NM, Parikshak NN, Stein JL, Walker MF, Ober GT, cgi?study_id=phs000169.v1.p1. Teran NA, Song Y, El-Fishawy P, Murtha RC, Choi M, Overton JD, 27. Autism and Developmental Disabilities Monitoring Network Surveillance Bjornson RD, Carriero NJ, Meyer KA, Bilguvar K, Mane SM, Sestan N, Year 2008 Principal Investigators; Centers for Disease Control and Lifton RP, Günel M, Roeder K, Geschwind DH, Devlin B, State MW: De novo Prevention: Prevalence of autism spectrum disorders--Autism and mutations revealed by whole-exome sequencing are strongly associated Developmental Disabilities Monitoring Network, 14 sites, United States, with autism. Nature 2012, 485:237–241. 2008. MMWR Surveill Summ 2012, 61:1–19. 9. Neale BM, Kou Y, Liu L, Ma'ayan A, Samocha KE, Sabo A, Lin CF, Stevens C, 28. Yang J, Lee SH, Goddard ME, Visscher PM: GCTA: a tool for genome-wide Wang LS, Makarov V, Polak P, Yoon S, Maguire J, Crawford EL, Campbell NG, complex trait analysis. Am J Hum Genet 2011, 88:76–82. Geller ET, Valladares O, Schafer C, Liu H, Zhao T, Cai G, Lihm J, Dannenfelser R, 29. http://digitalcommons.unl.edu/cgi/viewcontent.cgi? Jabado O, Peralta Z, Nagaswamy U, Muzny D, Reid JG, Newsham I, Wu Y, et al: article=1425&context=animalscifacpub. Patterns and rates of exonic de novo mutations in autism spectrum 30. Risch N: Implications of multilocus inheritance for gene-disease disorders. Nature 2012, 485:242. association studies. Theor Popul Biol 2001, 60:215–220. 10. O'Roak BJ, Vives L, Girirajan S, Karakoc E, Krumm N, Coe BP, Levy R, Ko A, 31. Ferreira MA, Sham P, Daly MJ, Purcell S: Ascertainment through family Lee C, Smith JD, Turner EH, Stanaway IB, Vernot B, Malig M, Baker C, Reilly B, history of disease often decreases the power of family-based association Akey JM, Borenstein E, Rieder MJ, Nickerson DA, Bernier R, Shendure J, studies. Behav Genet 2007, 37:631–636. Eichler EE: Sporadic autism exomes reveal a highly interconnected 32. Hamza TH, Zabetian CP, Tenesa A, Laederach A, Montimurro J, Yearout D, protein network of de novo mutations. Nature 2012, 485:246–250. Kay DM, Doheny KF, Paschall J, Pugh E, Kusel VI, Collura R, Roberts J, 11. Iossifov I, Ronemus M, Levy D, Wang Z, Hakker I, Rosenbaum J, Yamrom B, Griffith A, Samii A, Scott WK, Nutt J, Factor SA, Payami H: Common genetic Lee YH, Narzisi G, Leotta A, Kendall J, Grabowska E, Ma B, Marks S, variation in the HLA region is associated with late-onset sporadic Rodgers L, Stepansky A, Troge J, Andrews P, Bekritsky M, Pradhan K, Parkinson's disease. Nat Genet 2010, 42:781–785. Ghiban E, Kramer M, Parla J, Demeter R, Fulton LL, Fulton RS, Magrini VJ, 33. Neurogenetics Research Consortium data. http://www.ncbi.nlm.nih.gov/ Ye K, Darnell JC, Darnell RB, et al: De novo gene disruptions in children projects/gap/cgi-bin/study.cgi?study_id=phs000196.v2.p1. on the autistic spectrum. Neuron 2012, 74:285–299. 34. Clarke L, Zheng-Bradley X, Smith R, Kulesha E, Xiao C, Toneva I, Vaughan B, 12. Hallmayer J, Cleveland S, Torres A, Phillips J, Cohen B, Torigoe T, Miller J, Preuss D, Leinonen R, Shumway M, Sherry S, Flicek P, 1000 Genomes Project Fedele A, Collins J, Smith K, Lotspeich L, Croen LA, Ozonoff S, Lajonchere C, Consortium: The 1000 Genomes Project: data management and Grether JK, Risch N: Genetic heritability and shared environmental factors community access. Nat Methods 2012, 9:459–462. among twin pairs with autism. Arch Gen Psychiatry 2011, 68:1095–1102. 35. Rinaldo A, Bacanu SA, Devlin B, Sonpar V, Wasserman L, Roeder K: 13. Ronald A, Hoekstra RA: Autism spectrum disorders and autistic traits: a Characterization of multilocus linkage disequilibrium. Genet Epidemiol decade of new twin studies. Am J Med Genet B Neuropsychiatr Genet 2011, 2005, 28:193–206. 156B:255–274. 36. Devlin B, Risch N: A comparison of linkage disequilibrium measures for 14. Taniai H, Nishiyama T, Miyachi T, Imaeda M, Sumi S: Genetic influences on fine-scale mapping. Genomics 1995, 29:311–322. the broad spectrum of autism: study of proband-ascertained twins. 37. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, Am J Med Genet B Neuropsychiatr Genet 2008, 147B:844–849. Maller J, Sklar P, de Bakker PI, Daly MJ, Sham PC: PLINK: a tool set for 15. Devlin B, Melhem N, Roeder K: Do common variants play a role in whole-genome association and population-based linkage analyses. risk for autism? Evidence and theoretical musings. Brain Res 2011, Am J Hum Genet 2007, 81:559–575. 1380:78–84. 38. Bernier R, Gerdts J, Munson J, Dawson G, Estes A: Evidence for broader autism phenotype characteristics in parents from multiple-incidence 16. Pinto D, Pagnamenta AT, Klei L, Anney R, Merico D, Regan R, Conroy J, autism families. Autism Res 2012, 5:13–20. Magalhaes TR, Correia C, Abrahams BS, Almeida J, Bacchelli E, Bader GD, Bailey AJ, Baird G, Battaglia A, Berney T, Bolshakova N, Bölte S, Bolton PF, 39. Virkud YV, Todd RD, Abbacchi AM, Zhang Y, Constantino JN: Familial Bourgeron T, Brennan S, Brian J, Bryson SE, Carson AR, Casallo G, Casey J, aggregation of quantitative autistic traits in multiplex versus simplex Chung BH, Cochrane L, Corsello C, et al: Functional impact of global rare autism. Am J Med Genet B Neuropsychiatr Genet 2009, 150B:328–334. Klei et al. Molecular Autism 2012, 3:9 Page 13 of 13 http://www.molecularautism.com/content/3/1/9 40. Szatmari P, MacLean JE, Jones MB, Bryson SE, Zwaigenbaum L, Bartolucci G, 59. Dolcetti A, Silversides CK, Marshall CR, Lionel AC, Stavropoulos DJ, Mahoney WJ, Tuff L: The familial aggregation of the lesser variant in Scherer SW, Bassett AS: 1q21.1 Microduplication expression in adults. biological and nonbiological relatives of PDD probands: a family history Genet Med 2012, in press. study. J Child Psychol Psychiatry 2000, 41:579–586. 60. Lee SH, DeCandia TR, Ripke S, Yang J, Schizophrenia Psychiatric 41. Marco EJ, Skuse DH: Autism-lessons from the X chromosome. Soc Cogn Genome-Wide Association Study Consortium (PGC-SCZ); International Affect Neurosci 2006, 1:183–193. Schizophrenia Consortium (ISC); Molecular Genetics of Schizophrenia 42. Spiker D, Lotspeich LJ, Dimiceli S, Myers RM, Risch N: Behavioral Collaboration (MGS), Sullivan PF, Goddard ME, Keller MC, Visscher PM, phenotypic variation in autism multiplex families: evidence for a Wray NR: Estimating the proportion of variation in susceptibility to continuous severity gradient. Am J Med Genet 2002, 114:129–136. schizophrenia captured by common SNPs. Nat Genet 2012, 44:247–250. 43. Zhao X, Leotta A, Kustanovich V, Lajonchere C, Geschwind DH, Law K, 61. Lubke GH, Hottenga JJ, Walters R, Laurin C, de Geus EJ, Willemsen G, Law P, Qiu S, Lord C, Sebat J, Ye K, Wigler M: A unified genetic theory Smit JH, Middeldorp CM, Penninx BW, Vink JM, Boomsma DI: Estimating for sporadic and inherited autism. Proc Natl Acad Sci USA 2007, the genetic variance of major depressive disorder due to all single 104:12831–12836. nucleotide polymorphisms. Biol Psychiatry 2012, 72:707–709. 62. Ozonoff S, Young GS, Carter A, Messinger D, Yirmiya N, Zwaigenbaum L, 44. Sun X, Namkung J, Zhu X, Elston RC: Capability of common SNPs to tag Bryson S, Carver LJ, Constantino JN, Dobkins K, Hutman T, Iverson JM, rare variants. BMC Proc 2011, 5(Suppl 9):S88. Landa R, Rogers SJ, Sigman M, Stone WL: Recurrence risk for autism 45. Zuk O, Hechter E, Sunyaev SR, Lander ES: The mystery of missing spectrum disorders: a Baby Siblings Research Consortium study. heritability: genetic interactions create phantom heritability. Proc Natl Pediatrics 2011, 128:e488–e495. Acad Sci USA 2012, 109:1193–1198. 63. Kendler KS, Diehl SR: The genetics of schizophrenia: a current, 46. Risch N: Linkage strategies for genetically complex traits. I. Multilocus genetic-epidemiologic perspective. Schizophr Bull 1993, 19:261–285. models. Am J Hum Genet 1990, 46:222–228. 64. Smoller JW, Finn CT: Family, twin, and adoption studies of bipolar 47. Sanders AR, Duan J, Gejman PV: Complexities in psychiatric genetics. disorder. Am J Med Genet C Semin Med Genet 2003, 123C:48–58. Int Rev Psychiatry 2004, 16:284–293. 65. Bacanu S-A, Devlin B, Roeder K: The power of genomic control. Am J Hum 48. Slatkin M: Exchangeable models of complex inherited diseases. Genetics Genet 2000, 66:933–944. 2008, 179:2253–2261. 66. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D: 49. Devlin B, Daniels M, Roeder K: The heritability of IQ. Nature 1997, 388:468–471. Principal components analysis corrects for stratification in genome-wide 50. Wang K, Zhang H, Ma D, Bucan M, Glessner JT, Abrahams BS, Salyakina D, association studies. Nat Genet 2006, 38:904–909. Imielinski M, Bradfield JP, Sleiman PM, Kim CE, Hou C, Frackelton E, 67. Kerin T, Ramanathan A, Rivas K, Grepo N, Coetzee GA, Campbell DB: Chiavacci R, Takahashi N, Sakurai T, Rappaport E, Lajonchere CM, Munson J, A noncoding RNA antisense to moesin at 5p14.1 in autism. Sci Transl Estes A, Korvatska O, Piven J, Sonnenblick LI, Alvarez Retuerto AI, Herman EI, Med 2012, 4:128ra40. Dong H, Hutman T, Sigman M, Ozonoff S, Klin A, et al: Common genetic 68. Berkel S, Marshall CR, Weiss B, Howe J, Roeth R, Moog U, Endris V, Roberts variants on 5p14.1 associate with autism spectrum disorders. Nature W, Szatmari P, Pinto D, Bonin M, Riess A, Engels H, Sprengel R, Scherer SW, 2009, 459:528–533. Rappold GA: Mutations in the SHANK2 synaptic scaffolding gene in 51. Weiss LA, Arking DE, Daly MJ, Chakravarti A: A genome-wide linkage and autism spectrum disorder and mental retardation. Nat Genet 2010, association scan reveals novel loci for autism. Nature 2009, 461:802–808. 42:489–491. 52. Anney R, Klei L, Pinto D, Almeida J, Bacchelli E, Baird G, Bolshakova N, 69. Vaags AK, Lionel AC, Sato D, Goodenberger M, Stein QP, Curran S, Ogilvie C, Bölte S, Bolton PF, Bourgeron T, Brennan S, Brian J, Casey J, Conroy J, Ahn JW, Drmic I, Senman L, Chrysler C, Thompson A, Russell C, Prasad A, Correia C, Corsello C, Crawford EL, de Jonge M, Delorme R, Duketis E, Walker S, Pinto D, Marshall CR, Stavropoulos DJ, Zwaigenbaum L, Duque F, Estes A, Farrar P, Fernandez BA, Folstein SE, Fombonne E, Gilbert J, Fernandez BA, Fombonne E, Bolton PF, Collier DA, Hodge JC, Roberts W, Gillberg C, Glessner JT, Green A, et al: Individual common variants exert Szatmari P, Scherer SW: Rare deletions at the neurexin 3 locus in autism weak effects on risk for autism spectrum disorders. Hum Mol Genet 2012, spectrum disorder. Am J Hum Genet 2012, 90:133–141. in press. 70. Sato D, Lionel AC, Leblond CS, Prasad A, Pinto D, Walker S, O'Connor I, 53. Stahl EA, Wegmann D, Trynka G, Gutierrez-Achury J, Do R, Voight BF, Kraft P, Russell C, Drmic IE, Hamdan FF, Michaud JL, Endris V, Roeth R, Delorme R, Chen R, Kallberg HJ, Kurreeman FA, Diabetes Genetics Replication and Huguet G, Leboyer M, Rastam M, Gillberg C, Lathrop M, Stavropoulos DJ, Meta-analysis Consortium; Myocardial Infarction Genetics Consortium, Anagnostou E, Weksberg R, Fombonne E, Zwaigenbaum L, Fernandez BA, Kathiresan S, Wijmenga C, Gregersen PK, Alfredsson L, Siminovitch KA, Roberts W, Rappold GA, Marshall CR, Bourgeron T, Szatmari P, Scherer SW: Worthington J, de Bakker PI, Raychaudhuri S, Plenge RM: Bayesian SHANK1 deletions in males with autism spectrum disorder. Am J Hum inference analyses of the polygenic architecture of rheumatoid arthritis. Genet 2012, 90:879–887. Nat Genet 2012, 44:483–489. 71. Davis LK, Gamazon ER, Kistner-Griffin E, Badner JA, Liu C, Cook EH, Sutcliffe 54. Melhem N, Devlin B: Shedding new light on genetic dark matter. Genome JS, Cox NJ: Loci nominally associated with autism from genome-wide Med 2010, 2:79. analysis show enrichment of brain expression quantitative trait loci but 55. Lee SH, Decandia TR, Ripke S, Yang J, Schizophrenia Psychiatric not lymphoblastoid cell line expression quantitative trait loci. Mol Autism Genome-Wide Association Study Consortium (PGC-SCZ), The International 2012, 3:3. Schizophrenia Consortium (ISC), The Molecular Genetics of Schizophrenia Collaboration (MGS), Sullivan PF, Goddard ME, Keller MC, Visscher PM, doi:10.1186/2040-2392-3-9 Wray NR: Estimating the proportion of variation in susceptibility to Cite this article as: Klei et al.: Common genetic variants, acting Schizophrenia captured by common SNPs. Nat Genet 2012, 44:831. additively, are a major source of risk for autism. Molecular Autism 2012 56. International Schizophrenia Consortium, Purcell SM, Wray NR, Stone JL, 3:9. Visscher PM, O'Donovan MC, Sullivan PF, Sklar P: Common polygenic variation contributes to risk of Schizophrenia and bipolar disorder. Nature 2009, 460:748. 57. Ripke S, Sanders AR, Kendler KS, Levinson DF, Sklar P, Holmans PA, Lin DY, Duan J, Ophoff RA, Andreassen OA, Scolnick E, Cichon S, St Clair D, Corvin A, Gurling H, Werge T, Rujescu D, Blackwood DH, Pato CN, Malhotra AK, Purcell S, Dudbridge F, Neale BM, Rossin L, Visscher PM, Posthuma D, Ruderfer DM, Fanous A, Stefansson H, Steinberg S, et al: Genome-wide association study identifies five new Schizophrenia loci. Nat Genet 2011, 43:969–976. 58. Sullivan PF: The psychiatric GWAS consortium: big science comes to psychiatry. Neuron 2010, 68:182–186.

Journal

Molecular Autism – Springer Journals

Published: Oct 15, 2012

There are no references for this article.