Bioinformatics

Suo, Chen; Hrydziuszko, Olga; Lee, Donghwan; Pramana, Setia; Saputra, Dhany; Joshi, Himanshu; Calza, Stefano; Pawitan, Yudi

doi: 10.1093/bioinformatics/btv164pmid: 25810432

Motivation: Genome and transcriptome analyses can be used to explore cancers comprehensively, and it is increasingly common to have multiple omics data measured from each individual. Furthermore, there are rich functional data such as predicted impact of mutations on protein coding and gene/protein networks. However, integration of the complex information across the different omics and functional data is still challenging. Clinical validation, particularly based on patient outcomes such as survival, is important for assessing the relevance of the integrated information and for comparing different procedures.Results: An analysis pipeline is built for integrating genomic and transcriptomic alterations from whole-exome and RNA sequence data and functional data from protein function prediction and gene interaction networks. The method accumulates evidence for the functional implications of mutated potential driver genes found within and across patients. A driver-gene score (DGscore) is developed to capture the cumulative effect of such genes. To contribute to the score, a gene has to be frequently mutated, with high or moderate mutational impact at protein level, exhibiting an extreme expression and functionally linked to many differentially expressed neighbors in the functional gene network. The pipeline is applied to 60 matched tumor and normal samples of the same patient from The Cancer Genome Atlas breast-cancer project. In clinical validation, patients with high DGscores have worse survival than those with low scores (P = 0.001). Furthermore, the DGscore outperforms the established expression-based signatures MammaPrint and PAM50 in predicting patient survival. In conclusion, integration of mutation, expression and functional data allows identification of clinically relevant potential driver genes in cancer.Availability and implementation: The documented pipeline including annotated sample scripts can be found in http://fafner.meb.ki.se/biostatwiki/driver-genes/.Contact: [email protected] information: Supplementary data are available at Bioinformatics online.

Leng, Ning; Li, Yuan; McIntosh, Brian E.; Nguyen, Bao Kim; Duffin, Bret; Tian, Shulan; Thomson, James A.; Dewey, Colin N.; Stewart, Ron; Kendziorski, Christina

doi: 10.1093/bioinformatics/btv193pmid: 25847007

Motivation: With improvements in next-generation sequencing technologies and reductions in price, ordered RNA-seq experiments are becoming common. Of primary interest in these experiments is identifying genes that are changing over time or space, for example, and then characterizing the specific expression changes. A number of robust statistical methods are available to identify genes showing differential expression among multiple conditions, but most assume conditions are exchangeable and thereby sacrifice power and precision when applied to ordered data.Results: We propose an empirical Bayes mixture modeling approach called EBSeq-HMM. In EBSeq-HMM, an auto-regressive hidden Markov model is implemented to accommodate dependence in gene expression across ordered conditions. As demonstrated in simulation and case studies, the output proves useful in identifying differentially expressed genes and in specifying gene-specific expression paths. EBSeq-HMM may also be used for inference regarding isoform expression.Availability and implementation: An R package containing examples and sample datasets is available at Bioconductor.Contact: [email protected] information: Supplementary data are available at Bioinformatics online.

Colombo, Nicoló; Vlassis, Nikos

doi: 10.1093/bioinformatics/btv208pmid: 25886979

Motivation: Sequence discovery tools play a central role in several fields of computational biology. In the framework of Transcription Factor binding studies, most of the existing motif finding algorithms are computationally demanding, and they may not be able to support the increasingly large datasets produced by modern high-throughput sequencing technologies.Results: We present FastMotif, a new motif discovery algorithm that is built on a recent machine learning technique referred to as Method of Moments. Based on spectral decompositions, our method is robust to model misspecifications and is not prone to locally optimal solutions. We obtain an algorithm that is extremely fast and designed for the analysis of big sequencing data. On HT-Selex data, FastMotif extracts motif profiles that match those computed by various state-of-the-art algorithms, but one order of magnitude faster. We provide a theoretical and numerical analysis of the algorithm’s robustness and discuss its sensitivity with respect to the free parameters.Availability and implementation: The Matlab code of FastMotif is available from http://lcsb-portal.uni.lu/bioinformatics.Contact: [email protected] information: Supplementary data are available at Bioinformatics online.

Mandric, Igor; Zelikovsky, Alex

doi: 10.1093/bioinformatics/btv211pmid: 25890305

Motivation: Next-generation high-throughput sequencing has become a state-of-the-art technique in genome assembly. Scaffolding is one of the main stages of the assembly pipeline. During this stage, contigs assembled from the paired-end reads are merged into bigger chains called scaffolds. Because of a high level of statistical noise, chimeric reads, and genome repeats the problem of scaffolding is a challenging task. Current scaffolding software packages widely vary in their quality and are highly dependent on the read data quality and genome complexity. There are no clear winners and multiple opportunities for further improvements of the tools still exist.Results: This article presents an efficient scaffolding algorithm ScaffMatch that is able to handle reads with both short (<600 bp) and long (>35 000 bp) insert sizes producing high-quality scaffolds. We evaluate our scaffolding tool with the F score and other metrics (N50, corrected N50) on eight datasets comparing it with the most available packages. Our experiments show that ScaffMatch is the tool of preference for the most datasets.Availability and implementation: The source code is available at http://alan.cs.gsu.edu/NGS/?q=content/scaffmatch.Contact: [email protected] information: Supplementary data are available at Bioinformatics online.

Wang, Xiao; Zhang, Weiwei; Zhang, Qiuwen; Li, Guo-Zheng

doi: 10.1093/bioinformatics/btv212pmid: 25900916

Motivation: Identifying protein subchloroplast localization in chloroplast organelle is very helpful for understanding the function of chloroplast proteins. There have existed a few computational prediction methods for protein subchloroplast localization. However, these existing works have ignored proteins with multiple subchloroplast locations when constructing prediction models, so that they can predict only one of all subchloroplast locations of this kind of multilabel proteins.Results: To address this problem, through utilizing label-specific features and label correlations simultaneously, a novel multilabel classifier was developed for predicting protein subchloroplast location(s) with both single and multiple location sites. As an initial study, the overall accuracy of our proposed algorithm reaches 55.52%, which is quite high to be able to become a promising tool for further studies.Availability and implementation: An online web server for our proposed algorithm named MultiP-SChlo was developed, which are freely accessible at http://biomed.zzuli.edu.cn/bioinfo/multip-schlo/.Contact: [email protected] or [email protected] information: Supplementary data are available at Bioinformatics online.

Tang, Ke; Wong, Samuel W.K.; Liu, Jun S.; Zhang, Jinfeng; Liang, Jie

doi: 10.1093/bioinformatics/btv198pmid: 25861965

Motivation: Loops in proteins are often involved in biochemical functions. Their irregularity and flexibility make experimental structure determination and computational modeling challenging. Most current loop modeling methods focus on modeling single loops. In protein structure prediction, multiple loops often need to be modeled simultaneously. As interactions among loops in spatial proximity can be rather complex, sampling the conformations of multiple interacting loops is a challenging task.Results: In this study, we report a new method called multi-loop Distance-guided Sequential chain-Growth Monte Carlo (M-DiSGro) for prediction of the conformations of multiple interacting loops in proteins. Our method achieves an average RMSD of 1.93 Å for lowest energy conformations of 36 pairs of interacting protein loops with the total length ranging from 12 to 24 residues. We further constructed a data set containing proteins with 2, 3 and 4 interacting loops. For the most challenging target proteins with four loops, the average RMSD of the lowest energy conformations is 2.35 Å. Our method is also tested for predicting multiple loops in β-barrel membrane proteins. For outer-membrane protein G, the lowest energy conformation has a RMSD of 2.62 Å for the three extracellular interacting loops with a total length of 34 residues (12, 12 and 10 residues in each loop).Availability and implementation: The software is freely available at: tanto.bioe.uic.edu/m-DiSGro.Contact: [email protected] or [email protected] information: Supplementary data are available at Bioinformatics online.

Lee, Hui Sun; Jo, Sunhwan; Mukherjee, Srayanta; Park, Sang-Jun; Skolnick, Jeffrey; Lee, Jooyoung; Im, Wonpil

doi: 10.1093/bioinformatics/btv202pmid: 25857669

Motivation: Glycans play critical roles in many biological processes, and their structural diversity is key for specific protein-glycan recognition. Comparative structural studies of biological molecules provide useful insight into their biological relationships. However, most computational tools are designed for protein structure, and despite their importance, there is no currently available tool for comparing glycan structures in a sequence order- and size-independent manner.Results: A novel method, GS-align, is developed for glycan structure alignment and similarity measurement. GS-align generates possible alignments between two glycan structures through iterative maximum clique search and fragment superposition. The optimal alignment is then determined by the maximum structural similarity score, GS-score, which is size-independent. Benchmark tests against the Protein Data Bank (PDB) N-linked glycan library and PDB homologous/non-homologous N-glycoprotein sets indicate that GS-align is a robust computational tool to align glycan structures and quantify their structural similarity. GS-align is also applied to template-based glycan structure prediction and monosaccharide substitution matrix generation to illustrate its utility.Availability and implementation: http://www.glycanstructure.org/gsalign.Contact: [email protected] information: Supplementary data are available at Bioinformatics online.

Ding, Liang; Xue, Xingran; LaMarca, Sal; Mohebbi, Mohammad; Samad, Abdul; Malmberg, Russell L.; Cai, Liming

doi: 10.1093/bioinformatics/btv210pmid: 25886978

Motivation: Given the importance of non-coding RNAs to cellular regulatory functions, it would be highly desirable to have accurate computational prediction of RNA 3D structure, a task which remains challenging. Even for a short RNA sequence, the space of tertiary conformations is immense; existing methods to identify native-like conformations mostly resort to random sampling of conformations to achieve computational feasibility. However, native conformations may not be examined and prediction accuracy may be compromised due to sampling. State-of-the-art methods have yet to deliver satisfactory predictions for RNAs of length beyond 50 nucleotides.Results: This paper presents a method to tackle a key step in the RNA 3D structure prediction problem, the prediction of the nucleotide interactions that constitute the desired 3D structure. The research is based on a novel graph model, called a backbone k-tree, to tightly constrain the nucleotide interaction relationships considered for RNA 3D structures. It is shown that the new model makes it possible to efficiently predict the optimal set of nucleotide interactions (including the non-canonical interactions in all recently revealed families) from the query sequence along with known or predicted canonical basepairs. The preliminary results indicate that in most cases the new method can predict with a high accuracy the nucleotide interactions that constitute the 3D structure of the query sequence. It thus provides a useful tool for the accurate prediction of RNA 3D structure.Availability and Implementation: The source package for BkTree is available at http://rna-informatics.uga.edu/index.php?f=software&p=BkTree.Contact: [email protected] or [email protected] information: Supplementary data are available at Bioinformatics online.

Tang, Yin; Bouvier, Emil; Kwok, Chun Kit; Ding, Yiliang; Nekrutenko, Anton; Bevilacqua, Philip C.; Assmann, Sarah M.

doi: 10.1093/bioinformatics/btv213pmid: 25886980

Motivation: RNAs fold into complex structures that are integral to the diverse mechanisms underlying RNA regulation of gene expression. Recent development of transcriptome-wide RNA structure profiling through the application of structure-probing enzymes or chemicals combined with high-throughput sequencing has opened a new field that greatly expands the amount of in vitro and in vivo RNA structural information available. The resultant datasets provide the opportunity to investigate RNA structural information on a global scale. However, the analysis of high-throughput RNA structure profiling data requires considerable computational effort and expertise.Results: We present a new platform, StructureFold, that provides an integrated computational solution designed specifically for large-scale RNA structure mapping and reconstruction across any transcriptome. StructureFold automates the processing and analysis of raw high-throughput RNA structure profiling data, allowing the seamless incorporation of wet-bench structural information from chemical probes and/or ribonucleases to restrain RNA secondary structure prediction via the RNAstructure and ViennaRNA package algorithms. StructureFold performs reads mapping and alignment, normalization and reactivity derivation, and RNA structure prediction in a single user-friendly web interface or via local installation. The variation in transcript abundance and length that prevails in living cells and consequently causes variation in the counts of structure-probing events between transcripts is accounted for. Accordingly, StructureFold is applicable to RNA structural profiling data obtained in vivo as well as to in vitro or in silico datasets. StructureFold is deployed via the Galaxy platform.Availability and Implementation: StructureFold is freely available as a component of Galaxy available at: https://usegalaxy.org/.Contact: [email protected] or [email protected] information: Supplementary data are available at Bioinformatics online.

Ghadie, Mohamed A.; Japkowicz, Nathalie; Perkins, Theodore J.

doi: 10.1093/bioinformatics/btv192pmid: 25847008

Motivation: Stem cell differentiation is largely guided by master transcriptional regulators, but it also depends on the expression of other types of genes, such as cell cycle genes, signaling genes, metabolic genes, trafficking genes, etc. Traditional approaches to understanding gene expression patterns across multiple conditions, such as principal components analysis or K-means clustering, can group cell types based on gene expression, but they do so without knowledge of the differentiation hierarchy. Hierarchical clustering can organize cell types into a tree, but in general this tree is different from the differentiation hierarchy itself.Methods: Given the differentiation hierarchy and gene expression data at each node, we construct a weighted Euclidean distance metric such that the minimum spanning tree with respect to that metric is precisely the given differentiation hierarchy. We provide a set of linear constraints that are provably sufficient for the desired construction and a linear programming approach to identify sparse sets of weights, effectively identifying genes that are most relevant for discriminating different parts of the tree.Results: We apply our method to microarray gene expression data describing 38 cell types in the hematopoiesis hierarchy, constructing a weighted Euclidean metric that uses just 175 genes. However, we find that there are many alternative sets of weights that satisfy the linear constraints. Thus, in the style of random-forest training, we also construct metrics based on random subsets of the genes and compare them to the metric of 175 genes. We then report on the selected genes and their biological functions. Our approach offers a new way to identify genes that may have important roles in stem cell differentiation.Contact: [email protected] information: Supplementary data are available at Bioinformatics online.