TY - JOUR
AU - Liu, Tianming
AB - Abstract Literature studies have demonstrated the structural, connectional, and functional differences between cortical folding patterns in mammalian brains, such as convex and concave patterns. However, the molecular underpinning of such convex/concave differences remains largely unknown. Thanks to public access to a recently released set of marmoset whole-brain in situ hybridization data by RIKEN, Japan; this data’s accessibility empowers us to improve our understanding of the organization, regulation, and function of genes and their relation to macroscale metrics of brains. In this work, magnetic resonance imaging and diffusion tensor imaging macroscale neuroimaging data in this dataset were used to delineate convex/concave patterns in marmoset and to examine their structural features. Machine learning and visualization tools were employed to investigate the possible transcriptome difference between cortical convex and concave patterns. Experimental results demonstrated that a collection of genes is differentially expressed in convex and concave patterns, and their expression profiles can robustly characterize and differentiate the two folding patterns. More importantly, neuroscientific interpretations of these differentially expressed genes, as well as axonal guidance pathway analysis and gene enrichment analysis, offer novel understanding of structural and functional differences between cortical folding patterns in different regions from a molecular perspective. cortical folding, ISH, marmoset Introduction Elaborate convolution of the cerebral cortex is one of the most unique and prominent characteristics of the mammalian brain (Rakic 1988; Zilles et al. 1988; Fischl and Wald 2007; Honey et al. 2010; Striedter et al. 2015). In the literature, considerable efforts have been devoted into investigating the morphological (Smart 2002; Nordahl et al. 2007), structurally connectional (Chen et al. 2013; Zhang et al. 2014), and functional differences (Bullmore and Sporns 2009; van den Heuvel et al. 2009; Zamora-López et al. 2011; Zhang et al. 2018) between cortical folding patterns, such as convex and concave patterns, also known as gyri and sulci. Reports on this topic range from studying neural progenitor cells during the embryonic life at the cellular level (Kriegstein et al. 2006) to studying their functional difference by comparing functional magnetic resonance imaging (MRI) signals of human brains at macroscale (Liu et al. 2019). Meanwhile, various hypotheses regarding mechanical factors involved in cortical gyrification have been proposed, such as differential laminar growth (Richman et al. 1975; Razavi et al. 2017), differential progenitor proliferation (Kriegstein et al. 2006), axonal tension or compression (Van Essen 1997; Nie et al. 2012), and gliogenesis that was recently suggested to support the growth of cortical connections in white matter and to be a more relevant process than neuronal production, given the observation that the cortex exhibits an almost lissencephalic form at the time of cessation of neurogenesis (Rash et al. 2019). These hypotheses have also been suggested to fundamentally regulated by gene expressions (Beck et al. 1995; Rakic 2004; Barkovich et al. 2012; Stahl et al. 2013; Sun and Hevner 2014). For example, expression of Gene Trnp1 was discovered to control neuronal differentiation process and further regulate the radial and tangential growth of different cortical regions, resulting in the regionalized folding landscapes of the mammalian cerebral cortex (Stahl et al. 2013). Thus, the examination of gene expression patterns in different cortical folding patterns might open a new window to probe possible mechanisms under which genes control the specification and differentiation and to understand the genetic sources of the structural and functional differences between cortical folding patterns (Bartley et al. 1997; Stahl et al. 2013). Figure 1 Open in new tabDownload slide Cerebral cortical folding patterns of different species. Convex patterns are in dark gray and concave patterns are in light gray. For visualization, macaque, marmoset, and mouse brains are scaled to the size of human brain. Mouse cortical surface has no clear convex and concave patterns, whereas marmoset brains have convex and concave pairs around lateral and calcarine sulcus regions. Figure 1 Open in new tabDownload slide Cerebral cortical folding patterns of different species. Convex patterns are in dark gray and concave patterns are in light gray. For visualization, macaque, marmoset, and mouse brains are scaled to the size of human brain. Mouse cortical surface has no clear convex and concave patterns, whereas marmoset brains have convex and concave pairs around lateral and calcarine sulcus regions. Among a variety of methods that examine gene expressions at the transcriptome, in situ hybridization (ISH) has drawn an increasing attention. ISH is an image-based approach that preserves the spatial location of RNAs within a cell and the organization of cells within tissue. Therefore, it has been applied to animal studies to report transcripts mirroring the anatomy of the brain and displaying regional gene expression patterns (Lein et al. 2007; Li et al. 2017, 2018), which further facilitates the understanding of the molecular underpinning of cortical organization. For instance, an interesting study (Zeng et al. 2015) analyzed the ISH-derived gene expression patterns of cerebellum gyri and sulci of mice brains and reported significant variations in their expression profiles even though it is generally considered that rodents do not undergo cerebral cortical gyrification during the development. Compared with rodents, marmosets have pairs of obvious cortical concave and convex patterns, such as lateral fissure region and calcarine sulcus region (Solomon and Rosa 2014) (Fig. 1). Other advantages of using marmosets as animal models for cortical folding pattern analyses include their small brain size, the availability of transgenic capabilities, as well as their closer phylogenetical relationship to human (Paxinos et al. 2012) and relatively more human-like characteristics that are useful for cognitive behavioral research (Shimogori et al. 2018). Recently, Marmoset Gene Atlas (whole-brain ISH data) was released by RIKEN Brain Science Institute, Japan (Shimogori et al. 2018), which provides genome-wide, high-resolution gene expressions in neonatal marmoset brain with cellular resolution as well as anatomical MRI and diffusion tensor imaging (DTI) data. Till now, ISHs of over 1000 genes have been performed on marmoset brains and the data is free to download at https://gene-atlas.brainminds.riken.jp/. These ISH image series were digitized, stitched, and roughly registered to a common reference atlas so that a global comparison across regions and against the standard neuroanatomy is made possible and feasible, laying down a solid foundation for the study in this paper. In this work, we identified cortical convex and concave patterns from marmoset MRI data and then used ISH data on these regions to examine potentially different gene expression patterns via established machine learning and visualization approaches. Experimental results showed that a set of genes are differentially expressed between concave patterns and convex ones in these regions. The top 10% of those differently expressed genes were selected for gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis (Appendix, Ogata et al. 1999; Ashburner et al. 2000). In addition, these concave–convex folding pattern comparison analyses were performed on two different cortical regions such as lateral fissure and calcarine region. Different structural and anatomical patterns and gene expression patterns were related to possible different mechanisms of cortical folding on these two regions. Overall, our studies contributed to a better understanding of the genetic differences between different cortical folding patterns in different cortical regions. Figure 2 Open in new tabDownload slide Computational pipeline of analyzing and visualizing the transcript and macroscale image-based difference between convex and concave cortical patterns. It is noted that lateral fissure is used as an example in this figure and all the same processes were applied to calcarine sulcal regions. (a) Reconstruct cortical surfaces and compute fiber density and cortical thickness on the surfaces. (b) Extract gray matter and white matter boundaries from lateral fissure in reference atlas by selecting the boundary voxels and identify coordinates of convex voxels (red) and concave voxels (blue), which were mapped back to surfaces in (a). (c) Register gene expression images to the reference atlas and calculate gene expression feature maps of convex voxels and concave voxels. (d) Apply gene-level SVM models to classify voxels into different morphological patterns and identify the genes that are differently expressed. Input is gene expression feature maps of voxels. (e) Apply gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of the identified genes. (f) Statistically analyze expressions of genes related to axon guidance pathway. Figure 2 Open in new tabDownload slide Computational pipeline of analyzing and visualizing the transcript and macroscale image-based difference between convex and concave cortical patterns. It is noted that lateral fissure is used as an example in this figure and all the same processes were applied to calcarine sulcal regions. (a) Reconstruct cortical surfaces and compute fiber density and cortical thickness on the surfaces. (b) Extract gray matter and white matter boundaries from lateral fissure in reference atlas by selecting the boundary voxels and identify coordinates of convex voxels (red) and concave voxels (blue), which were mapped back to surfaces in (a). (c) Register gene expression images to the reference atlas and calculate gene expression feature maps of convex voxels and concave voxels. (d) Apply gene-level SVM models to classify voxels into different morphological patterns and identify the genes that are differently expressed. Input is gene expression feature maps of voxels. (e) Apply gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of the identified genes. (f) Statistically analyze expressions of genes related to axon guidance pathway. Materials and Methods Our computational pipeline consists of six steps. The first step is to reconstruct cortical surfaces and compute fiber density and cortical thickness from macroscale MRI data (Fig. 2a). The second step is to determine morphological convex voxels and concave voxels in gene expression atlas. This was done on two cortical regions such as lateral fissure region and calcarine sulcus region (lateral fissure region is shown as an example in Fig. 2b). These folding patterns were also transferred to cortical surfaces in Figure 2a. The third step is to register all gene expression images to gene expression atlas (Fig. 2c). The fourth step is to apply support vector machine (SVM) to classify convex/concave voxels based on their transcript difference (Fig. 2d). In the fifth step, we apply enrichment pathway analysis on the selected top accuracy gene sets (Fig. 2e). The sixth step is to statistically analyze expressions of genes related to axon guidance pathway (Fig. 2f). Material and Preprocessing Marmoset Anatomical Atlas and DTI Brain Images The marmoset T2-weighted MRI (T2 MRI for short) atlas was downloaded from Brain/MINDS 3D Marmoset Reference Atlas dataset (Shimogori et al. 2018). The dataset includes ex vivo T2 MRI data, brain region segmentation data, and gray, midcortical, and white matter boundary segmentation. Volume size of the MRI image is 678 × 1000 × 210, and resolution is 0.04 × 0.04 × 0.11 mm3. Diffusion weighted imaging (DWI) data were acquired using a 9.4 T small animal MRI scanner (Schaeffer et al. 2017), equipped with a 12 cm gradient coil set of 400 mT/m strength (Agilent Technologies). DTI data of four marmosets were acquired with the following parameters: number of shots = 1, repetition time = 5 s, echo time = 26 ms, diffusion time = 12.32 ms, diffusion-encoding directions = 256 (plus 9 b = 0 s/mm2 images), b-value = 1000 s/mm2, and number of averages = 2. Each volume comprised of 36 slices with an in-plane resolution of 0.6 × 0.6 mm2 and a slice thickness of 0.6 mm. The field of view was 48 × 48 mm2, and the corresponding matrix size was 80 × 80. It should be mentioned that DTI data were acquired on only four marmoset monkeys, but have a comparatively high spatial resolution (180 times higher than the human brain, Schaeffer et al. 2017). In their experiment, small-bore animal MRI scanners with purpose-built hardware to acquire higher field strengths and higher signal-to-noise ratios. The settings of scanners allow the use of multifiber/voxel modeling (Jeurissen et al. 2011) and thus the estimation of multiple intravoxel fiber populations. The preprocessing steps are described as follows. For DTI data, DSI-Studio (http://dsi-studio.labsolver.org) was applied to track fiber for DTI analysis. Skull stripping, eddy correction, manually drawn brain masking, and the generalized q-sampling imaging model were applied (Yeh et al. 2010). Then, the deterministic fiber tracking algorithm (Yeh et al. 2013) in DSI-Studio was applied to reconstruct 5000 fibers for each subject. The parameters are as follows: quantitative anisotropy (QA) = 0.02, angular threshold = 60, step size = 0.3 mm, minimal fiber length = 10 mm, and maximal fiber length = 100 mm. For T2-weighted MRI data, the white matter segmentation image from T2 MRI atlas was obtained to reconstruct the white matter and gray matter cortical surfaces (Fig. 2a) via FreeSurfer (Dale et al. 1999; Fischl et al. 1999, 2002, 2004; Fischl 2012). T2-weighted MRI data was nonlinearly warped to QA map of DTI data via FSL-FLIRT and FSL-FNIRT (Jenkinson et al. 2002; Andersson and Skare 2010; Jenkinson et al. 2012; Andersson and Sotiropoulos 2016) in sequence. The resulted linear transformation matrix and nonlinear warp field were used to register cortical surfaces into DTI space via Connectome Workbench (www.humanconnectome.org). Marmoset Brain Gene Expression Atlas The marmoset brain gene expression atlas contains a neonate marmoset reference atlas (Nissl-stained coronal slices) and over 1000 ISH gene expression profiles for marmosets of different ages. For each gene, a set of 2D images were generated by ISH and then collected and preprocessed by an informatics data processing pipeline (Shimogori et al. 2018). The resulting data is a 3D volume with quantified gene expression values at the grid voxel level. Each 3D volume is composed of 60 slices with matrix dimension of 7200 × 8400. It is noted that the ISH volumes were aligned to the anatomical reference atlas based on the selected landmarks. However, the landmarked-based registration is not ideal. For example, 59 slices were not well aligned to the corresponding slice in the anatomical reference atlas. Also, the landmark slice was selected based on anterior commissure. Since anterior commissure regions exist in several adjacent slices, it is hard to select the landmark slices across subjects. To resolve these issues, we picked several adjacent slices in each ISH image and warped them into slices in the Nissl-stained reference atlas through global linear image registration (Goshtasby 2005), as illustrated in Figures 2c and 3. Then, the pair of registered slices and reference slices with higher similarities was chosen. As the slice size of the MRI image is 678 × 1000, we downsampled slices in ISH images from 7200 × 8400 to 720 × 840. Figure 3 Open in new tabDownload slide An example for ISH volume registration. (a–c) show three adjacent slices of gene CLSTN2, slices #18, #19, and #20. Three slices are registered to the 18th, 19th, 20th, and 21st slices in the reference atlas. The purple background in second and third columns is the 18th and 19th slices in reference atlas, and in fourth and fifth columns are the 20th and 21st slices in reference atlas. (d–f) show the overlapping view of slices registered to the 18th slice in reference atlas. (g–i) show the overlapping view of slices registered to the 19th slice in reference atlas. (j–l) show the overlapping view of slices registered to the 20th slice in reference atlas. (m–o) show the overlapping view of slices registered to the 21st slice in reference atlas. The yellow boxes represent the registration results of slices of the same index in reference atlas and gene CLSTN2. The red boxes represent better registration results of slices of different indices between reference atlas and gene CLSTN2. Figure 3 Open in new tabDownload slide An example for ISH volume registration. (a–c) show three adjacent slices of gene CLSTN2, slices #18, #19, and #20. Three slices are registered to the 18th, 19th, 20th, and 21st slices in the reference atlas. The purple background in second and third columns is the 18th and 19th slices in reference atlas, and in fourth and fifth columns are the 20th and 21st slices in reference atlas. (d–f) show the overlapping view of slices registered to the 18th slice in reference atlas. (g–i) show the overlapping view of slices registered to the 19th slice in reference atlas. (j–l) show the overlapping view of slices registered to the 20th slice in reference atlas. (m–o) show the overlapping view of slices registered to the 21st slice in reference atlas. The yellow boxes represent the registration results of slices of the same index in reference atlas and gene CLSTN2. The red boxes represent better registration results of slices of different indices between reference atlas and gene CLSTN2. The gene expression value of a voxel was calculated on ISH images. Assuming that the coordinate of one voxel is |$(i,j)$| and its grayscale intensity is |$g(i,j)$| in the origin slices, the gene expression value |${v}_{g(i,j)}$| in the sampled slice was defined as |${v}_{g(i,j)}=(\sum_{m=-5}^5\sum_{n=-5}^5c(i\times 10+m,j\times 10+n)\ast g(i\times 10+m,j\times 10+n))$|⁠. Here, |$c(i,j)$| is equal to 1 if |$g(i,j)$| is greater than the threshold value and is 0, otherwise. It is noted that gene expression volumes of all 1129 genes in lateral fissure region have been aligned to the reference atlas, whereas only 1116 same genes among these 1129 genes in calcarine sulcus region have been aligned to the reference atlas. Thus, expression volumes of these shared 1116 genes were used in this work. Identification of Convex and Concave Patterns In the open Brain/MINDS 3D Marmoset Reference Atlas dataset, Nissl reference atlas has been registered to T2 volume data. As illustrated in Figure 2a,b, cortical surface was reconstructed from T2 volume data, and it is therefore in the same space of the Nissl reference. Regions of interest (lateral fissure and calcarine sulcus) were identified on the surface, and within this region we identified convex and concave voxels in these regions on Nissl reference atlas (Fig. 2b). More specifically, we performed an automatic extraction by selecting the white/gray matter boundary voxels with high curvatures in these regions (Zhang et al. 2018). We found clear boundaries between gray matter and white matter (Fig. 2b) in the reference atlas. The curvature of the boundary curves is an effective indicator for convex and concave patterns. In detail, we first converted these images into grayscale images and determined the gray matter with a manually set threshold. Next, we extracted the boundaries of gray matter and white matter. Accordingly, we computed the curvature of every voxel in the boundary curves. Those voxels with local maximum curvature values in convex parts were considered as the central voxels of convex patterns, and the central voxels of concave patterns in concave parts. To ensure selected voxels assigned to a specific convex and concave pattern, these voxels whose curvature values are higher than 90% of local maximal curvature values of concave pits belong to concave regions, and voxels whose curvature values higher than 90% of local maximal curvature values of convex peaks belong to convex regions. In total, 2800 voxels were identified as convex (1400) and concave (1400) voxels in lateral fissure region, 4400 voxels were identified as convex (2130) and concave (2270) voxels in calcarine sulcus region. Finally, these convex and concave regions were mapped back to the surfaces (red and blue regions in Fig. 2a,b) for image-based analyses. Since a surface facet is defined by three vertices, it is classified as a convex facet if it has more than one convex vertex or as a concave facet, otherwise. Structural Features on Cortical Surface Fiber density of a surface facet was defined as the number of fibers that penetrate this facet. Then, the average fiber density value of all convex facets and concave facets could be computed and compared. Cortical thickness of one vertex in white matter surface was acquired by calculating the smallest distance from this vertex in white matter surface to the gray matter surface. Thus, the average cortical thickness of all convex vertices and concave vertices were computed. Illustrations of these metrics are found in the rightmost column in Figure 2a. Classification of Convex and Concave Patterns Based on Gene Expression We trained a gene-level SVM classification model (Fig. 2d) to identify genes that are significantly differentially expressed in cortical convex and concave voxels (Cortes and Vapnik 1995). The SVM model was employed due to its reasonable performance in binary classification problem with a small number of samples. A linear kernel model was used. Sequential minimal optimization method was applied to find the separating hyperplane (Platt 1998). To single out the top contributing genes, we trained an SVM model at gene level (Fig. 2d). The 2800 voxel samples in lateral fissure region were split into 2240 for training and 560 for testing (a 5-fold cross-validation). The same 5-fold cross-validation operation was applied to voxels in calcarine sulcus region. Area under the curve (AUC) of these models was calculated to evaluate the classification capability of a gene. GO and KEGG Enrichment Analysis of Classification Candidate Gene Set We applied GO and KEGG enrichment analysis to understand biological functions of genes related to morphological difference (Ogata et al. 1999; Ashburner et al. 2000). Here, we selected three sets of genes. Set #1 has genes of the top 10% AUC values in both lateral fissure region and calcarine sulcus region. Set #2 has genes of the top 10% AUC in calcarine sulcus region. Set #3 has genes of the top 10% AUC in lateral fissure region. The enrichment analysis of biological process and molecular function of these sets were performed using GO database (http://geneontology.org/). After significance analysis, P < 0.1 was set to a significant threshold, and the results of gene annotation and enrichment analysis with statistical significance were obtained. Simultaneously, the signal transduction pathway enrichment analysis was performed on the gene sets using the KEGG database (http://www.kegg.jp/kegg/tool/map_pathway1.html), and the P-values with statistical differences were statistically analyzed by Fisher’s Exact Test. P < 0.05 represents the significance threshold and represents the statistically significant signal transmission path of the gene sets relative to the background. Statistical Analysis of Genes Related to Axon Guidance Pathway For most genes in the gene expression atlas, only gene expressions in one certain growth stage were collected, so that the comparison of different growth stages was not available here. Therefore, our work is limited within the comparison of gene expressions between convex and concave patterns. The above-mentioned convex and concave voxels in lateral fissure region and calcarine sulcus region were separately used as the observation of each gene. Gene expression of convex (vcvx) and concave (vccv) patterns were calculated as the average value of gene expressions in these two patterns, respectively. The contrast value of gene expression of a gene in convex pattern and concave pattern is defined as |$\mathrm{contrast}=\frac{v_{cvx}-{v}_{ccv}}{\max ({v}_{cvx},{v}_{ccv})}$|⁠. Among the 1116 genes in marmoset brain gene expression atlas, 27 genes were found in the axon guidance pathway. Then gene expression of convex patterns (vcvx), concave patterns (vccv), and contrast value of these 27 genes were calculated. The amounts (1089) of genes not directly related to axon guidance pathway are far more than the amounts (27) of genes directly related to the pathway. Thus, we randomly selected 27 genes out of the 1089 genes not directly related to axon pathway as a contrast group and calculated their expression of convex patterns (vcvx), concave patterns (vccv), and contrast value. We repeated selection and calculation for 5000 times to acquire the average values of vcvx, vccv, and contrast. Results Classification Performance of Gene-level SVM We separately trained the SVM model with expressions of 1116 genes in lateral fissure region and calcarine sulcus region. Genes with higher AUC value (the top 10%) in both regions as well as their AUC values and their average gene expressions in convex and concave patterns are listed in Table 1, sorted by their AUC value rankings. Table 1 Classification performance of gene-level SVM models on selected transcripts Gene name . Calcarine AUC . Calcarine convex pattern . Calcarine concave pattern . Lateral AUC . Lateral convex pattern . Lateral concave pattern . TMEM130 0.80 0.18 0.35 0.67 0.35 0.49 GAMT 0.74 0.17 0.24 0.65 0.14 0.16 KCTD17 0.74 0.17 0.24 0.64 0.18 0.22 GRM7 0.74 0.20 0.30 0.65 0.24 0.28 CHL1 0.73 0.36 0.57 0.65 0.51 0.57 CHRNA3 0.71 0.11 0.12 0.70 0.11 0.12 GNRH1 0.71 0.17 0.18 0.67 0.17 0.18 LAYN 0.70 0.15 0.16 0.65 0.15 0.16 ROBO1 0.69 0.16 0.26 0.69 0.15 0.25 PGRMC1 0.69 0.16 0.20 0.73 0.17 0.24 GLRA3 0.68 0.18 0.21 0.65 0.19 0.22 ECE2 0.68 0.17 0.22 0.71 0.16 0.20 GRID2 0.68 0.17 0.19 0.64 0.17 0.18 HMP19 0.68 0.23 0.37 0.63 0.25 0.36 CBLN4 0.67 0.20 0.22 0.66 0.18 0.21 SLC1A5 0.67 0.19 0.22 0.76 0.20 0.25 Gene name . Calcarine AUC . Calcarine convex pattern . Calcarine concave pattern . Lateral AUC . Lateral convex pattern . Lateral concave pattern . TMEM130 0.80 0.18 0.35 0.67 0.35 0.49 GAMT 0.74 0.17 0.24 0.65 0.14 0.16 KCTD17 0.74 0.17 0.24 0.64 0.18 0.22 GRM7 0.74 0.20 0.30 0.65 0.24 0.28 CHL1 0.73 0.36 0.57 0.65 0.51 0.57 CHRNA3 0.71 0.11 0.12 0.70 0.11 0.12 GNRH1 0.71 0.17 0.18 0.67 0.17 0.18 LAYN 0.70 0.15 0.16 0.65 0.15 0.16 ROBO1 0.69 0.16 0.26 0.69 0.15 0.25 PGRMC1 0.69 0.16 0.20 0.73 0.17 0.24 GLRA3 0.68 0.18 0.21 0.65 0.19 0.22 ECE2 0.68 0.17 0.22 0.71 0.16 0.20 GRID2 0.68 0.17 0.19 0.64 0.17 0.18 HMP19 0.68 0.23 0.37 0.63 0.25 0.36 CBLN4 0.67 0.20 0.22 0.66 0.18 0.21 SLC1A5 0.67 0.19 0.22 0.76 0.20 0.25 These genes in this table have higher AUC values (the top 10%) in both lateral fissure and calcarine sulcus. Their average expressions are indicated in the last two columns. Open in new tab Table 1 Classification performance of gene-level SVM models on selected transcripts Gene name . Calcarine AUC . Calcarine convex pattern . Calcarine concave pattern . Lateral AUC . Lateral convex pattern . Lateral concave pattern . TMEM130 0.80 0.18 0.35 0.67 0.35 0.49 GAMT 0.74 0.17 0.24 0.65 0.14 0.16 KCTD17 0.74 0.17 0.24 0.64 0.18 0.22 GRM7 0.74 0.20 0.30 0.65 0.24 0.28 CHL1 0.73 0.36 0.57 0.65 0.51 0.57 CHRNA3 0.71 0.11 0.12 0.70 0.11 0.12 GNRH1 0.71 0.17 0.18 0.67 0.17 0.18 LAYN 0.70 0.15 0.16 0.65 0.15 0.16 ROBO1 0.69 0.16 0.26 0.69 0.15 0.25 PGRMC1 0.69 0.16 0.20 0.73 0.17 0.24 GLRA3 0.68 0.18 0.21 0.65 0.19 0.22 ECE2 0.68 0.17 0.22 0.71 0.16 0.20 GRID2 0.68 0.17 0.19 0.64 0.17 0.18 HMP19 0.68 0.23 0.37 0.63 0.25 0.36 CBLN4 0.67 0.20 0.22 0.66 0.18 0.21 SLC1A5 0.67 0.19 0.22 0.76 0.20 0.25 Gene name . Calcarine AUC . Calcarine convex pattern . Calcarine concave pattern . Lateral AUC . Lateral convex pattern . Lateral concave pattern . TMEM130 0.80 0.18 0.35 0.67 0.35 0.49 GAMT 0.74 0.17 0.24 0.65 0.14 0.16 KCTD17 0.74 0.17 0.24 0.64 0.18 0.22 GRM7 0.74 0.20 0.30 0.65 0.24 0.28 CHL1 0.73 0.36 0.57 0.65 0.51 0.57 CHRNA3 0.71 0.11 0.12 0.70 0.11 0.12 GNRH1 0.71 0.17 0.18 0.67 0.17 0.18 LAYN 0.70 0.15 0.16 0.65 0.15 0.16 ROBO1 0.69 0.16 0.26 0.69 0.15 0.25 PGRMC1 0.69 0.16 0.20 0.73 0.17 0.24 GLRA3 0.68 0.18 0.21 0.65 0.19 0.22 ECE2 0.68 0.17 0.22 0.71 0.16 0.20 GRID2 0.68 0.17 0.19 0.64 0.17 0.18 HMP19 0.68 0.23 0.37 0.63 0.25 0.36 CBLN4 0.67 0.20 0.22 0.66 0.18 0.21 SLC1A5 0.67 0.19 0.22 0.76 0.20 0.25 These genes in this table have higher AUC values (the top 10%) in both lateral fissure and calcarine sulcus. Their average expressions are indicated in the last two columns. Open in new tab In addition to genes in Table 1, we also identified other two sets of genes as introduced in section GO and KEGG Enrichment Analysis of Classification Candidate Gene Set. We show all three sets in Table 2, where set #1 is also listed for convenience. Table 2 Three gene sets where genes have top AUC values in both lateral fissure and calcarine sulcus regions or in only one of them Gene sets . Genes . Gene set #1: Top 10% AUC in both regions TMEM130, GAMT, KCTD17, GRM7, CHL1, CHRNA3, GNRH1, LAYN, ROBO1, PGRMC1, GLRA3, ECE2, GRID2, HMP19, CBLN4, SLC1A5 Gene set #2: Top 10% AUC in calcarine sulcus region NRIP3, KLHL1, TMEM108, PITPNC1, GLRB, CLIP2, NECAB3, SOSTDC1, RAB1A, KCTD1, CLASP2, GRIN2B, TSPYL2, CACNA1E, SEMA6D, KCNB1, APBA1, KCTD5, ECEL1, TMX4, RPS19, KCNN2, GJC1, DDX51, NFXL1, NETO1, GRIN2B, CHRNA2, RND3, SYT10, SEMA3A, KCTD2, RPS6KA6, BRINP1, SIM1, RAD54L2, HPS3. CCNDBP1. NEUROD6, HGF, ANO2, ACLY, KIF3C, ETV6, EIF4EBP1, DDC, PRKCG, PLXNA4, PODXL2, INPP4B, MAPK11, GCH1, FMN1, GRM4, TPBG, DYNC1H1, BCKDHB, KIAA0319, HTRA3, KCNH3, HLF, IK, DHX30, SNTB1, TMEM179, SPATS2L, IGFBP5, SHANK3, JKAMP, MSMO1, ADRA1A, SLC6A4, HERC4, HOMER3, CDH13, ATP11A, SMPD1, SLC1A1, FIGN, IGFBP6, SGSM1, ABCD2, UCN3, ADNP, NECAB2, CTTN, DCDC2, CIT, VWC2, NPAS2, NRN1, DOCK5, HES6, ACYP1, SHANK1, COL12A1 Gene set #3: Top 10% AUC in lateral fissure region RDH10, CACNA1A, SV2B, OPRK1, ANOS1, CAMKK2, NFIX, NXPH3, IFIT3, CRIM1, CADM1, CA12, ANKRD6, EMX1, NOV, GRM2, BDNF, FAM234A, AMZ2, GRM6, ADAMTS2, HTR2A, GALNT10, BBS2, NOG, PLCG2, NOSIP, ATP2A2, GRM1, CDKL5, PTPRA, EMX2, HCRTR2, GRIK3, NSDHL, PLS3, CLN8, FCHO2, NNAT, RGN, EFNB3, NGFR, SYNPR, DLK1, MPPED1, RGS14, ST6GALNAC5, ITGA6, SLC17A6, ARHGAP32, ZDHHC2, CYP19A1, PHOSPHO2, CAMKV, STMN4, EDEM3, ADAM33, FAM43A, HTR1A, SLC29A1, PJA2, CHRM5, RGS2, CDKN2C, FHDC1, DACT2, ALDH2, SULF1, HAPLN3, SCNN1A, PLEKHG1, TYRO3, ITGB1, NUP210, PRDX3, APOPT1, IRF2BPL, CEP128, LYPD6, PSMD6, PDE1A, NOTCH3, ENPP2, PLXNB2, TRIB3, NAPRT, MACROD2, RET, SHF, PLCL1, FAM196A, RERE, MAL2, AUTS2 Gene sets . Genes . Gene set #1: Top 10% AUC in both regions TMEM130, GAMT, KCTD17, GRM7, CHL1, CHRNA3, GNRH1, LAYN, ROBO1, PGRMC1, GLRA3, ECE2, GRID2, HMP19, CBLN4, SLC1A5 Gene set #2: Top 10% AUC in calcarine sulcus region NRIP3, KLHL1, TMEM108, PITPNC1, GLRB, CLIP2, NECAB3, SOSTDC1, RAB1A, KCTD1, CLASP2, GRIN2B, TSPYL2, CACNA1E, SEMA6D, KCNB1, APBA1, KCTD5, ECEL1, TMX4, RPS19, KCNN2, GJC1, DDX51, NFXL1, NETO1, GRIN2B, CHRNA2, RND3, SYT10, SEMA3A, KCTD2, RPS6KA6, BRINP1, SIM1, RAD54L2, HPS3. CCNDBP1. NEUROD6, HGF, ANO2, ACLY, KIF3C, ETV6, EIF4EBP1, DDC, PRKCG, PLXNA4, PODXL2, INPP4B, MAPK11, GCH1, FMN1, GRM4, TPBG, DYNC1H1, BCKDHB, KIAA0319, HTRA3, KCNH3, HLF, IK, DHX30, SNTB1, TMEM179, SPATS2L, IGFBP5, SHANK3, JKAMP, MSMO1, ADRA1A, SLC6A4, HERC4, HOMER3, CDH13, ATP11A, SMPD1, SLC1A1, FIGN, IGFBP6, SGSM1, ABCD2, UCN3, ADNP, NECAB2, CTTN, DCDC2, CIT, VWC2, NPAS2, NRN1, DOCK5, HES6, ACYP1, SHANK1, COL12A1 Gene set #3: Top 10% AUC in lateral fissure region RDH10, CACNA1A, SV2B, OPRK1, ANOS1, CAMKK2, NFIX, NXPH3, IFIT3, CRIM1, CADM1, CA12, ANKRD6, EMX1, NOV, GRM2, BDNF, FAM234A, AMZ2, GRM6, ADAMTS2, HTR2A, GALNT10, BBS2, NOG, PLCG2, NOSIP, ATP2A2, GRM1, CDKL5, PTPRA, EMX2, HCRTR2, GRIK3, NSDHL, PLS3, CLN8, FCHO2, NNAT, RGN, EFNB3, NGFR, SYNPR, DLK1, MPPED1, RGS14, ST6GALNAC5, ITGA6, SLC17A6, ARHGAP32, ZDHHC2, CYP19A1, PHOSPHO2, CAMKV, STMN4, EDEM3, ADAM33, FAM43A, HTR1A, SLC29A1, PJA2, CHRM5, RGS2, CDKN2C, FHDC1, DACT2, ALDH2, SULF1, HAPLN3, SCNN1A, PLEKHG1, TYRO3, ITGB1, NUP210, PRDX3, APOPT1, IRF2BPL, CEP128, LYPD6, PSMD6, PDE1A, NOTCH3, ENPP2, PLXNB2, TRIB3, NAPRT, MACROD2, RET, SHF, PLCL1, FAM196A, RERE, MAL2, AUTS2 Open in new tab Table 2 Three gene sets where genes have top AUC values in both lateral fissure and calcarine sulcus regions or in only one of them Gene sets . Genes . Gene set #1: Top 10% AUC in both regions TMEM130, GAMT, KCTD17, GRM7, CHL1, CHRNA3, GNRH1, LAYN, ROBO1, PGRMC1, GLRA3, ECE2, GRID2, HMP19, CBLN4, SLC1A5 Gene set #2: Top 10% AUC in calcarine sulcus region NRIP3, KLHL1, TMEM108, PITPNC1, GLRB, CLIP2, NECAB3, SOSTDC1, RAB1A, KCTD1, CLASP2, GRIN2B, TSPYL2, CACNA1E, SEMA6D, KCNB1, APBA1, KCTD5, ECEL1, TMX4, RPS19, KCNN2, GJC1, DDX51, NFXL1, NETO1, GRIN2B, CHRNA2, RND3, SYT10, SEMA3A, KCTD2, RPS6KA6, BRINP1, SIM1, RAD54L2, HPS3. CCNDBP1. NEUROD6, HGF, ANO2, ACLY, KIF3C, ETV6, EIF4EBP1, DDC, PRKCG, PLXNA4, PODXL2, INPP4B, MAPK11, GCH1, FMN1, GRM4, TPBG, DYNC1H1, BCKDHB, KIAA0319, HTRA3, KCNH3, HLF, IK, DHX30, SNTB1, TMEM179, SPATS2L, IGFBP5, SHANK3, JKAMP, MSMO1, ADRA1A, SLC6A4, HERC4, HOMER3, CDH13, ATP11A, SMPD1, SLC1A1, FIGN, IGFBP6, SGSM1, ABCD2, UCN3, ADNP, NECAB2, CTTN, DCDC2, CIT, VWC2, NPAS2, NRN1, DOCK5, HES6, ACYP1, SHANK1, COL12A1 Gene set #3: Top 10% AUC in lateral fissure region RDH10, CACNA1A, SV2B, OPRK1, ANOS1, CAMKK2, NFIX, NXPH3, IFIT3, CRIM1, CADM1, CA12, ANKRD6, EMX1, NOV, GRM2, BDNF, FAM234A, AMZ2, GRM6, ADAMTS2, HTR2A, GALNT10, BBS2, NOG, PLCG2, NOSIP, ATP2A2, GRM1, CDKL5, PTPRA, EMX2, HCRTR2, GRIK3, NSDHL, PLS3, CLN8, FCHO2, NNAT, RGN, EFNB3, NGFR, SYNPR, DLK1, MPPED1, RGS14, ST6GALNAC5, ITGA6, SLC17A6, ARHGAP32, ZDHHC2, CYP19A1, PHOSPHO2, CAMKV, STMN4, EDEM3, ADAM33, FAM43A, HTR1A, SLC29A1, PJA2, CHRM5, RGS2, CDKN2C, FHDC1, DACT2, ALDH2, SULF1, HAPLN3, SCNN1A, PLEKHG1, TYRO3, ITGB1, NUP210, PRDX3, APOPT1, IRF2BPL, CEP128, LYPD6, PSMD6, PDE1A, NOTCH3, ENPP2, PLXNB2, TRIB3, NAPRT, MACROD2, RET, SHF, PLCL1, FAM196A, RERE, MAL2, AUTS2 Gene sets . Genes . Gene set #1: Top 10% AUC in both regions TMEM130, GAMT, KCTD17, GRM7, CHL1, CHRNA3, GNRH1, LAYN, ROBO1, PGRMC1, GLRA3, ECE2, GRID2, HMP19, CBLN4, SLC1A5 Gene set #2: Top 10% AUC in calcarine sulcus region NRIP3, KLHL1, TMEM108, PITPNC1, GLRB, CLIP2, NECAB3, SOSTDC1, RAB1A, KCTD1, CLASP2, GRIN2B, TSPYL2, CACNA1E, SEMA6D, KCNB1, APBA1, KCTD5, ECEL1, TMX4, RPS19, KCNN2, GJC1, DDX51, NFXL1, NETO1, GRIN2B, CHRNA2, RND3, SYT10, SEMA3A, KCTD2, RPS6KA6, BRINP1, SIM1, RAD54L2, HPS3. CCNDBP1. NEUROD6, HGF, ANO2, ACLY, KIF3C, ETV6, EIF4EBP1, DDC, PRKCG, PLXNA4, PODXL2, INPP4B, MAPK11, GCH1, FMN1, GRM4, TPBG, DYNC1H1, BCKDHB, KIAA0319, HTRA3, KCNH3, HLF, IK, DHX30, SNTB1, TMEM179, SPATS2L, IGFBP5, SHANK3, JKAMP, MSMO1, ADRA1A, SLC6A4, HERC4, HOMER3, CDH13, ATP11A, SMPD1, SLC1A1, FIGN, IGFBP6, SGSM1, ABCD2, UCN3, ADNP, NECAB2, CTTN, DCDC2, CIT, VWC2, NPAS2, NRN1, DOCK5, HES6, ACYP1, SHANK1, COL12A1 Gene set #3: Top 10% AUC in lateral fissure region RDH10, CACNA1A, SV2B, OPRK1, ANOS1, CAMKK2, NFIX, NXPH3, IFIT3, CRIM1, CADM1, CA12, ANKRD6, EMX1, NOV, GRM2, BDNF, FAM234A, AMZ2, GRM6, ADAMTS2, HTR2A, GALNT10, BBS2, NOG, PLCG2, NOSIP, ATP2A2, GRM1, CDKL5, PTPRA, EMX2, HCRTR2, GRIK3, NSDHL, PLS3, CLN8, FCHO2, NNAT, RGN, EFNB3, NGFR, SYNPR, DLK1, MPPED1, RGS14, ST6GALNAC5, ITGA6, SLC17A6, ARHGAP32, ZDHHC2, CYP19A1, PHOSPHO2, CAMKV, STMN4, EDEM3, ADAM33, FAM43A, HTR1A, SLC29A1, PJA2, CHRM5, RGS2, CDKN2C, FHDC1, DACT2, ALDH2, SULF1, HAPLN3, SCNN1A, PLEKHG1, TYRO3, ITGB1, NUP210, PRDX3, APOPT1, IRF2BPL, CEP128, LYPD6, PSMD6, PDE1A, NOTCH3, ENPP2, PLXNB2, TRIB3, NAPRT, MACROD2, RET, SHF, PLCL1, FAM196A, RERE, MAL2, AUTS2 Open in new tab Finally, we performed experiments to compare the two pairs of the concave fundus and one of the convex crests to demonstrate that gene expression difference between two neighboring convex crests is not significant, whereas difference between either one of them and the sulcus is referred to Fig. S1 and Tables S1-2 in Supplementary Materials. We applied random tests to demonstrate that the identified genes are specific to lateral and calcarine regions other than other smooth regions. Also, we used different parameters and kernels of SVM to identified genes to demonstrate that the results are referred to Table S3 in Supplementary Materials. More details are referred in Supplementary Materials. Neuroscientific Interpretations of Genes with High Classification Performance In general, the neuroscientific interpretations of the differentially expressed genes in three gene sets in Table 2 are well related to the neuronal and axonal development. In set #1, we found some genes playing an essential role in axonal development. The ROBO1 gene is a protein-coding gene; its product encodes an integral membrane protein that functions in axon guidance and neuronal precursor cell migration, known to be involved in the decision by axons to cross the central nervous system midline (Kidd et al. 1999). The GRM7 gene is associated with L-glutamate, which is the major excitatory neurotransmitter in the central nervous system. This neurotransmission is involved in most aspects of normal brain function and can be perturbed in many neuropathologic conditions (Xia et al. 2015). The GRID2 gene encodes the protein, which is a member of the family of ionotropic glutamate receptors; these receptors are proven to be the predominant excitatory neurotransmitter receptors (Takayama et al. 1995). And this protein also plays a role in synapse organization between parallel fibers (Hills et al. 2013). The other genes such as CHL1 (its encoded protein is a member of the L1 gene family of neural cell adhesion molecules, which may be involved in signal transduction pathway; Kleene et al. 2015) and GLRA3 (related with ligand-gated ion channels, widely distributed throughout the central nervous system; Nikolic et al. 1998) were also meaningful in neural development. In set #2, several genes are related to axon guidance pathway, including PLXNA4, SEMA6D, SEMA3A, MAPK11, NEUROD6, which play essential roles in the developing nervous system. It should be mentioned that PLXNA4 is a family member of receptors for transmembrane, secreted, and glycosylphosphatidylinositol-anchored semaphorins in vertebrates as well as a receptor for secreted Semaphorin 3A (SEMA3A) and SEMA6 proteins (Tamagnone et al. 1999). PLXNA4 plays an important role in semaphorin signaling and axon guidance (Suto et al. 2005). MAPK11 encodes one kind of protein kinases involved in the integration of biochemical signals for a wide variety of cellular processes, including cell proliferation, differentiation, and transcriptional regulation (Yosimichi et al. 2001). The protein produced by NERUOD6 is involved in the development and differentiation of the nervous system (Uittenbogaard et al. 2010). CLASP2 (regulates neuronal polarity and synaptic function; Beffert et al. 2012), SHANK1 (regulates the development and functioning of neuronal synapses; Sato et al. 2012), HOMER3 (encodes a member of the homer family of dendritic proteins; Ango et al. 2000), SLC1A1 (encodes a member of the high-affinity glutamate transporters; Arnold et al. 2006), and PRKCG (this kind protein kinase is expressed solely in the brain and spinal cord and its localization is restricted to neurons; Brandt et al. 1987) are related to brain development, too. In set #3, NNAT encodes the protein neuronatin, a proteolipid that controls ion channels during brain development (Numata et al. 2012). NNAT starts to differentiate pluripotent stem cells into neural cells by lifting the calcium levels and its relevant expression in neural tissues throughout the brain, contributing to the development of the nervous system (Joseph 2014). The exact mechanisms of how NNAT guides the differentiation of pluripotent stem cells into neuronal cells on convex and concave patterns are to be investigated biologically in the future. The BDNF gene shows higher significance in concave patterns (average expression = 0.14) than in convex patterns (average expression = 0.01). The BDNF gene provides instructions for making a protein called brain-derived neurotrophic factor, which promotes the survival of neurons by playing a role in the growth, differentiation, and maintenance of these neurons (Lindsay et al. 1985). In brains, the BDNF protein is active at the connections between synapses. Therefore, the differential expression of BDNF between convex and concave patterns could be related to the difference of axonal fiber connections between convex and concave patterns, which was reported in Van Essen (1997) and Nie et al. (2012). Other differentially expressed genes involved in various stages of brain developments include NXPH3 (signal molecules that resemble neuropeptides; Craig and Kang 2007), SYNPR (involved in transporter activity; Knaus et al. 1990), EFNB3 (important in brain development and its maintenance; Pohlkamp et al. 2016), AUTS2 (implicated in neurodevelopment; Oksenberg and Ahituv 2013), ANOS1 (plays a key role in the migration of neurons; de Castro et al. 2016), and GRIK3 (associated with predominant excitatory neurotransmitter receptors in the mammalian brain and activated in variety of normal neurophysiologic processes; Hollmann and Heinemann 1994). In addition, we performed GO and KEGG enrichment analysis of genes in these three gene sets. As shown in Table 3, it was found that genes in set #1 were related to axon guidance, cell junction, postsynaptic membrane, and integral component of membrane. Functions of genes in set #2 mainly focus on nervous system development, regulation of cell growth, chemical synaptic transmission, dendritic, cell junction, adenosine 5′-triphosphate (ATP) binding, and postsynaptic membrane. Functions of genes in set #3 mainly focus on integral component of membrane, dendrite morphogenesis, transmission of nerve impulse, forebrain cell migration, dendritic spine, dendritic, locomotory behavior, regulation of axonogenesis, etc. Table 3 Enrichment analysis of three gene sets based on GO database GO_ID/GO_Term . Count/Percentage (%) . P-value (×10−2) . Genes . Gene set 1: Top 10% AUC in both regions GO:0007411 ~ axon guidance 2/12.5 6.90 ROBO1,CHL1 GO:0045211 ~ postsynaptic membrane 3/18.75 0.24 GRID2, CHRNA3, GLRA3 GO:0016021 ~ integral component of membrane 11/68.75 0.33 TMEM130, GRM7, CHRNA3, LAYN, ROBO1, PGRMC1, ECE2, GRID2, HMP19, SLC1A5, CHL1 GO:0030054 ~ cell junction 3/18.75 0.85 GRID2, CHRNA3, GLRA3 Gene set 2: Top 10% AUC in only calcarine sulcus region 0051260/prote in homooligomerization 5/4.55 0.14 KCTD5, KCTD1, KCTD17, SLC1A1, KCNB1 0007268/chemical synaptic transmission 3/2.73 0.29 APBA1, SYT10, PRKCG 0007399/nervous system development 3/2.73 0.36 GLRB, HES6, NEUROD6 0001558/regulation of cell growth 3/2.73 2.36 IGFBP6, IGFBP5, HTRA3 0030514/negative regulation of BMP signaling pathway 3/2.73 2.49 VWC2, SOSTDC1, HTRA3 1 902 287/semaphorin–plexin signaling pathway involved in axon guidance 2/1.82 3.46 PLXNA4, SEMA3A 0021637/trigeminal nerve structural organization 2/1.82 3.46 PLXNA4, SEMA3A 0014912/negative regulation of smooth muscle cell migration 2/1.82 4.81 IGFBP5, SEMA6D 0045211/postsynaptic membrane 3/2.73 0.01 CHRNA2, GLRB, GRIN2B 0030425/dendrite 4/3.64 0.45 PRKCG, CHL1, KCNB1, KLHL1 0030054/cell junction 2/1.82 1.18 CHRNA2, GRIN2B 0016594/glycine binding 2/1.82 0.16 GLRB, GRIN2B 0005524/ATP binding 13/11.82 4.18 ABCD2, KIF3C, MAPK11, DYNC1H1, CIT, ACLY, FIGN, DDX51, PRKCG, ATP11A, RPS6KA6, RAD54L2, DHX30 0021612/facial nerve structural organization 2/1.82 6.14 PLXNA4, SEMA3A 0045987/positive regulation of smooth muscle contraction 2/1.82 7.45 CTTN, ADRA1A 0005615/extracellular space 8/7.28 8.62 NRN1, VWC2, UCN3, SMPD1, COL12A1, CDH13, ADNP, SOSTDC1 Gene set 3: Top 10% AUC in only lateral fissure region 0019226 ~ transmission of nerve impulse 2/1.77 0.40 CACNA1A, CHRM5 0043588 ~ skin development 3/2.65 1.67 ADAMTS2, ITGA6, DACT2 0048813 ~ dendrite morphogenesis 3/2.65 1.79 CACNA1A, ITGB1, RERE 0007628 ~ adult walking behavior 3/2.65 2.04 CACANA1, EFNB3, CLN8 0051216 ~ cartilage development 3/2.65 2.17 NOG, SULF1, BBS2 0021885 ~ forebrain cell migration 2/1.77 2.20 TYRO3, EMX2 0007216 ~ G-protein-coupled glutamate receptor signaling pathway 2/1.77 2.92 GRM1, GRM6 0046883 ~ regulation of hormone secretion 2/1.77 2.92 HTR1A, HTR2A 0005887 ~ integral component of plasma membrane 11/9.73 0.00 RET, HTR1A, CHRM5, PLXNB2, SLC29A1, GRM6, GRIK, SCNN1A, OPRK1, ZDHHC2, HTR2A 0016021 ~ integral component of membrane 33/29.20 0.03 ST6GALNAC5, GALNT10, MAL2, SYNPR, NOTCH3, PTPRA, NSDHL, DLK1, SLC17A6, EFNB3, PLXNB2, GRIK3, OPRK1, ZDHHC2, NUP210, CRIM1, BDNF, NNAT, NGFR, FAM234A, ATP2A2, CYP19A1, HCRTR2, TYRO3, SV2B, CADM1, CA12, GRM1, ADAM33, GRM2, CLN8, SCNN1A, RDH10 0045211/postsynaptic membrane 2/1.77 0.11 CHRM5, GRIK3 0030054/cell junction 2/1.77 1.05 CHRM5, GRIK3 0043197/dendritic spine 2/1.77 2.00 ITGB1, RGS14 0043235/receptor complex 2/1.77 2.50 CACNA1A, GRM1 0030425/dendrite 2/1.77 2.72 CACNA1A, GRM1 0005509/calcium ion binding 10/8.85 0.69 RET, RGN, EDEM3, NOTCH3, ATP2A2, PLS3, CAMKK2, SULF1, ENPP2, DLK1 0008066/glutamate receptor activity 2/1.77 2.57 GRM1, GRM6 0004993/G-protein-coupled serotonin receptor activity 2/1.77 6.90 HTR1A, HTR2A 0021796/cerebral cortex regionalization 2/1.77 5.05 EMX2, EMX1 0033627/cell adhesion mediated by integrin 2/1.77 5.05 ITGA6, NOV 0072001/renal system development 2/1.77 5.75 EMX2, ITGA6 0050795/regulation of behavior 2/1.77 6.44 HTR1A, HTR2A 0042493/response to drug 3/2.65 7.01 EMX2, EMX1, HTR2A 0009395/phospholipid catabolic process 2/1.77 7.82 PLCG2, ENPP2 0007626/locomotory behavior 3/2.65 7.86 GRM1, GRM6, OPRK1 0050770/regulation of axonogenesis 2/1.77 8.50 RET, CACNA1A 0007283/spermatogenesis 3/2.65 9.15 ADAMTS2, TYRO3, RGS2 0051209/release of sequestered calcium ion into cytosol 2/1.77 9.18 PLCG2, HTR2A GO_ID/GO_Term . Count/Percentage (%) . P-value (×10−2) . Genes . Gene set 1: Top 10% AUC in both regions GO:0007411 ~ axon guidance 2/12.5 6.90 ROBO1,CHL1 GO:0045211 ~ postsynaptic membrane 3/18.75 0.24 GRID2, CHRNA3, GLRA3 GO:0016021 ~ integral component of membrane 11/68.75 0.33 TMEM130, GRM7, CHRNA3, LAYN, ROBO1, PGRMC1, ECE2, GRID2, HMP19, SLC1A5, CHL1 GO:0030054 ~ cell junction 3/18.75 0.85 GRID2, CHRNA3, GLRA3 Gene set 2: Top 10% AUC in only calcarine sulcus region 0051260/prote in homooligomerization 5/4.55 0.14 KCTD5, KCTD1, KCTD17, SLC1A1, KCNB1 0007268/chemical synaptic transmission 3/2.73 0.29 APBA1, SYT10, PRKCG 0007399/nervous system development 3/2.73 0.36 GLRB, HES6, NEUROD6 0001558/regulation of cell growth 3/2.73 2.36 IGFBP6, IGFBP5, HTRA3 0030514/negative regulation of BMP signaling pathway 3/2.73 2.49 VWC2, SOSTDC1, HTRA3 1 902 287/semaphorin–plexin signaling pathway involved in axon guidance 2/1.82 3.46 PLXNA4, SEMA3A 0021637/trigeminal nerve structural organization 2/1.82 3.46 PLXNA4, SEMA3A 0014912/negative regulation of smooth muscle cell migration 2/1.82 4.81 IGFBP5, SEMA6D 0045211/postsynaptic membrane 3/2.73 0.01 CHRNA2, GLRB, GRIN2B 0030425/dendrite 4/3.64 0.45 PRKCG, CHL1, KCNB1, KLHL1 0030054/cell junction 2/1.82 1.18 CHRNA2, GRIN2B 0016594/glycine binding 2/1.82 0.16 GLRB, GRIN2B 0005524/ATP binding 13/11.82 4.18 ABCD2, KIF3C, MAPK11, DYNC1H1, CIT, ACLY, FIGN, DDX51, PRKCG, ATP11A, RPS6KA6, RAD54L2, DHX30 0021612/facial nerve structural organization 2/1.82 6.14 PLXNA4, SEMA3A 0045987/positive regulation of smooth muscle contraction 2/1.82 7.45 CTTN, ADRA1A 0005615/extracellular space 8/7.28 8.62 NRN1, VWC2, UCN3, SMPD1, COL12A1, CDH13, ADNP, SOSTDC1 Gene set 3: Top 10% AUC in only lateral fissure region 0019226 ~ transmission of nerve impulse 2/1.77 0.40 CACNA1A, CHRM5 0043588 ~ skin development 3/2.65 1.67 ADAMTS2, ITGA6, DACT2 0048813 ~ dendrite morphogenesis 3/2.65 1.79 CACNA1A, ITGB1, RERE 0007628 ~ adult walking behavior 3/2.65 2.04 CACANA1, EFNB3, CLN8 0051216 ~ cartilage development 3/2.65 2.17 NOG, SULF1, BBS2 0021885 ~ forebrain cell migration 2/1.77 2.20 TYRO3, EMX2 0007216 ~ G-protein-coupled glutamate receptor signaling pathway 2/1.77 2.92 GRM1, GRM6 0046883 ~ regulation of hormone secretion 2/1.77 2.92 HTR1A, HTR2A 0005887 ~ integral component of plasma membrane 11/9.73 0.00 RET, HTR1A, CHRM5, PLXNB2, SLC29A1, GRM6, GRIK, SCNN1A, OPRK1, ZDHHC2, HTR2A 0016021 ~ integral component of membrane 33/29.20 0.03 ST6GALNAC5, GALNT10, MAL2, SYNPR, NOTCH3, PTPRA, NSDHL, DLK1, SLC17A6, EFNB3, PLXNB2, GRIK3, OPRK1, ZDHHC2, NUP210, CRIM1, BDNF, NNAT, NGFR, FAM234A, ATP2A2, CYP19A1, HCRTR2, TYRO3, SV2B, CADM1, CA12, GRM1, ADAM33, GRM2, CLN8, SCNN1A, RDH10 0045211/postsynaptic membrane 2/1.77 0.11 CHRM5, GRIK3 0030054/cell junction 2/1.77 1.05 CHRM5, GRIK3 0043197/dendritic spine 2/1.77 2.00 ITGB1, RGS14 0043235/receptor complex 2/1.77 2.50 CACNA1A, GRM1 0030425/dendrite 2/1.77 2.72 CACNA1A, GRM1 0005509/calcium ion binding 10/8.85 0.69 RET, RGN, EDEM3, NOTCH3, ATP2A2, PLS3, CAMKK2, SULF1, ENPP2, DLK1 0008066/glutamate receptor activity 2/1.77 2.57 GRM1, GRM6 0004993/G-protein-coupled serotonin receptor activity 2/1.77 6.90 HTR1A, HTR2A 0021796/cerebral cortex regionalization 2/1.77 5.05 EMX2, EMX1 0033627/cell adhesion mediated by integrin 2/1.77 5.05 ITGA6, NOV 0072001/renal system development 2/1.77 5.75 EMX2, ITGA6 0050795/regulation of behavior 2/1.77 6.44 HTR1A, HTR2A 0042493/response to drug 3/2.65 7.01 EMX2, EMX1, HTR2A 0009395/phospholipid catabolic process 2/1.77 7.82 PLCG2, ENPP2 0007626/locomotory behavior 3/2.65 7.86 GRM1, GRM6, OPRK1 0050770/regulation of axonogenesis 2/1.77 8.50 RET, CACNA1A 0007283/spermatogenesis 3/2.65 9.15 ADAMTS2, TYRO3, RGS2 0051209/release of sequestered calcium ion into cytosol 2/1.77 9.18 PLCG2, HTR2A Open in new tab Table 3 Enrichment analysis of three gene sets based on GO database GO_ID/GO_Term . Count/Percentage (%) . P-value (×10−2) . Genes . Gene set 1: Top 10% AUC in both regions GO:0007411 ~ axon guidance 2/12.5 6.90 ROBO1,CHL1 GO:0045211 ~ postsynaptic membrane 3/18.75 0.24 GRID2, CHRNA3, GLRA3 GO:0016021 ~ integral component of membrane 11/68.75 0.33 TMEM130, GRM7, CHRNA3, LAYN, ROBO1, PGRMC1, ECE2, GRID2, HMP19, SLC1A5, CHL1 GO:0030054 ~ cell junction 3/18.75 0.85 GRID2, CHRNA3, GLRA3 Gene set 2: Top 10% AUC in only calcarine sulcus region 0051260/prote in homooligomerization 5/4.55 0.14 KCTD5, KCTD1, KCTD17, SLC1A1, KCNB1 0007268/chemical synaptic transmission 3/2.73 0.29 APBA1, SYT10, PRKCG 0007399/nervous system development 3/2.73 0.36 GLRB, HES6, NEUROD6 0001558/regulation of cell growth 3/2.73 2.36 IGFBP6, IGFBP5, HTRA3 0030514/negative regulation of BMP signaling pathway 3/2.73 2.49 VWC2, SOSTDC1, HTRA3 1 902 287/semaphorin–plexin signaling pathway involved in axon guidance 2/1.82 3.46 PLXNA4, SEMA3A 0021637/trigeminal nerve structural organization 2/1.82 3.46 PLXNA4, SEMA3A 0014912/negative regulation of smooth muscle cell migration 2/1.82 4.81 IGFBP5, SEMA6D 0045211/postsynaptic membrane 3/2.73 0.01 CHRNA2, GLRB, GRIN2B 0030425/dendrite 4/3.64 0.45 PRKCG, CHL1, KCNB1, KLHL1 0030054/cell junction 2/1.82 1.18 CHRNA2, GRIN2B 0016594/glycine binding 2/1.82 0.16 GLRB, GRIN2B 0005524/ATP binding 13/11.82 4.18 ABCD2, KIF3C, MAPK11, DYNC1H1, CIT, ACLY, FIGN, DDX51, PRKCG, ATP11A, RPS6KA6, RAD54L2, DHX30 0021612/facial nerve structural organization 2/1.82 6.14 PLXNA4, SEMA3A 0045987/positive regulation of smooth muscle contraction 2/1.82 7.45 CTTN, ADRA1A 0005615/extracellular space 8/7.28 8.62 NRN1, VWC2, UCN3, SMPD1, COL12A1, CDH13, ADNP, SOSTDC1 Gene set 3: Top 10% AUC in only lateral fissure region 0019226 ~ transmission of nerve impulse 2/1.77 0.40 CACNA1A, CHRM5 0043588 ~ skin development 3/2.65 1.67 ADAMTS2, ITGA6, DACT2 0048813 ~ dendrite morphogenesis 3/2.65 1.79 CACNA1A, ITGB1, RERE 0007628 ~ adult walking behavior 3/2.65 2.04 CACANA1, EFNB3, CLN8 0051216 ~ cartilage development 3/2.65 2.17 NOG, SULF1, BBS2 0021885 ~ forebrain cell migration 2/1.77 2.20 TYRO3, EMX2 0007216 ~ G-protein-coupled glutamate receptor signaling pathway 2/1.77 2.92 GRM1, GRM6 0046883 ~ regulation of hormone secretion 2/1.77 2.92 HTR1A, HTR2A 0005887 ~ integral component of plasma membrane 11/9.73 0.00 RET, HTR1A, CHRM5, PLXNB2, SLC29A1, GRM6, GRIK, SCNN1A, OPRK1, ZDHHC2, HTR2A 0016021 ~ integral component of membrane 33/29.20 0.03 ST6GALNAC5, GALNT10, MAL2, SYNPR, NOTCH3, PTPRA, NSDHL, DLK1, SLC17A6, EFNB3, PLXNB2, GRIK3, OPRK1, ZDHHC2, NUP210, CRIM1, BDNF, NNAT, NGFR, FAM234A, ATP2A2, CYP19A1, HCRTR2, TYRO3, SV2B, CADM1, CA12, GRM1, ADAM33, GRM2, CLN8, SCNN1A, RDH10 0045211/postsynaptic membrane 2/1.77 0.11 CHRM5, GRIK3 0030054/cell junction 2/1.77 1.05 CHRM5, GRIK3 0043197/dendritic spine 2/1.77 2.00 ITGB1, RGS14 0043235/receptor complex 2/1.77 2.50 CACNA1A, GRM1 0030425/dendrite 2/1.77 2.72 CACNA1A, GRM1 0005509/calcium ion binding 10/8.85 0.69 RET, RGN, EDEM3, NOTCH3, ATP2A2, PLS3, CAMKK2, SULF1, ENPP2, DLK1 0008066/glutamate receptor activity 2/1.77 2.57 GRM1, GRM6 0004993/G-protein-coupled serotonin receptor activity 2/1.77 6.90 HTR1A, HTR2A 0021796/cerebral cortex regionalization 2/1.77 5.05 EMX2, EMX1 0033627/cell adhesion mediated by integrin 2/1.77 5.05 ITGA6, NOV 0072001/renal system development 2/1.77 5.75 EMX2, ITGA6 0050795/regulation of behavior 2/1.77 6.44 HTR1A, HTR2A 0042493/response to drug 3/2.65 7.01 EMX2, EMX1, HTR2A 0009395/phospholipid catabolic process 2/1.77 7.82 PLCG2, ENPP2 0007626/locomotory behavior 3/2.65 7.86 GRM1, GRM6, OPRK1 0050770/regulation of axonogenesis 2/1.77 8.50 RET, CACNA1A 0007283/spermatogenesis 3/2.65 9.15 ADAMTS2, TYRO3, RGS2 0051209/release of sequestered calcium ion into cytosol 2/1.77 9.18 PLCG2, HTR2A GO_ID/GO_Term . Count/Percentage (%) . P-value (×10−2) . Genes . Gene set 1: Top 10% AUC in both regions GO:0007411 ~ axon guidance 2/12.5 6.90 ROBO1,CHL1 GO:0045211 ~ postsynaptic membrane 3/18.75 0.24 GRID2, CHRNA3, GLRA3 GO:0016021 ~ integral component of membrane 11/68.75 0.33 TMEM130, GRM7, CHRNA3, LAYN, ROBO1, PGRMC1, ECE2, GRID2, HMP19, SLC1A5, CHL1 GO:0030054 ~ cell junction 3/18.75 0.85 GRID2, CHRNA3, GLRA3 Gene set 2: Top 10% AUC in only calcarine sulcus region 0051260/prote in homooligomerization 5/4.55 0.14 KCTD5, KCTD1, KCTD17, SLC1A1, KCNB1 0007268/chemical synaptic transmission 3/2.73 0.29 APBA1, SYT10, PRKCG 0007399/nervous system development 3/2.73 0.36 GLRB, HES6, NEUROD6 0001558/regulation of cell growth 3/2.73 2.36 IGFBP6, IGFBP5, HTRA3 0030514/negative regulation of BMP signaling pathway 3/2.73 2.49 VWC2, SOSTDC1, HTRA3 1 902 287/semaphorin–plexin signaling pathway involved in axon guidance 2/1.82 3.46 PLXNA4, SEMA3A 0021637/trigeminal nerve structural organization 2/1.82 3.46 PLXNA4, SEMA3A 0014912/negative regulation of smooth muscle cell migration 2/1.82 4.81 IGFBP5, SEMA6D 0045211/postsynaptic membrane 3/2.73 0.01 CHRNA2, GLRB, GRIN2B 0030425/dendrite 4/3.64 0.45 PRKCG, CHL1, KCNB1, KLHL1 0030054/cell junction 2/1.82 1.18 CHRNA2, GRIN2B 0016594/glycine binding 2/1.82 0.16 GLRB, GRIN2B 0005524/ATP binding 13/11.82 4.18 ABCD2, KIF3C, MAPK11, DYNC1H1, CIT, ACLY, FIGN, DDX51, PRKCG, ATP11A, RPS6KA6, RAD54L2, DHX30 0021612/facial nerve structural organization 2/1.82 6.14 PLXNA4, SEMA3A 0045987/positive regulation of smooth muscle contraction 2/1.82 7.45 CTTN, ADRA1A 0005615/extracellular space 8/7.28 8.62 NRN1, VWC2, UCN3, SMPD1, COL12A1, CDH13, ADNP, SOSTDC1 Gene set 3: Top 10% AUC in only lateral fissure region 0019226 ~ transmission of nerve impulse 2/1.77 0.40 CACNA1A, CHRM5 0043588 ~ skin development 3/2.65 1.67 ADAMTS2, ITGA6, DACT2 0048813 ~ dendrite morphogenesis 3/2.65 1.79 CACNA1A, ITGB1, RERE 0007628 ~ adult walking behavior 3/2.65 2.04 CACANA1, EFNB3, CLN8 0051216 ~ cartilage development 3/2.65 2.17 NOG, SULF1, BBS2 0021885 ~ forebrain cell migration 2/1.77 2.20 TYRO3, EMX2 0007216 ~ G-protein-coupled glutamate receptor signaling pathway 2/1.77 2.92 GRM1, GRM6 0046883 ~ regulation of hormone secretion 2/1.77 2.92 HTR1A, HTR2A 0005887 ~ integral component of plasma membrane 11/9.73 0.00 RET, HTR1A, CHRM5, PLXNB2, SLC29A1, GRM6, GRIK, SCNN1A, OPRK1, ZDHHC2, HTR2A 0016021 ~ integral component of membrane 33/29.20 0.03 ST6GALNAC5, GALNT10, MAL2, SYNPR, NOTCH3, PTPRA, NSDHL, DLK1, SLC17A6, EFNB3, PLXNB2, GRIK3, OPRK1, ZDHHC2, NUP210, CRIM1, BDNF, NNAT, NGFR, FAM234A, ATP2A2, CYP19A1, HCRTR2, TYRO3, SV2B, CADM1, CA12, GRM1, ADAM33, GRM2, CLN8, SCNN1A, RDH10 0045211/postsynaptic membrane 2/1.77 0.11 CHRM5, GRIK3 0030054/cell junction 2/1.77 1.05 CHRM5, GRIK3 0043197/dendritic spine 2/1.77 2.00 ITGB1, RGS14 0043235/receptor complex 2/1.77 2.50 CACNA1A, GRM1 0030425/dendrite 2/1.77 2.72 CACNA1A, GRM1 0005509/calcium ion binding 10/8.85 0.69 RET, RGN, EDEM3, NOTCH3, ATP2A2, PLS3, CAMKK2, SULF1, ENPP2, DLK1 0008066/glutamate receptor activity 2/1.77 2.57 GRM1, GRM6 0004993/G-protein-coupled serotonin receptor activity 2/1.77 6.90 HTR1A, HTR2A 0021796/cerebral cortex regionalization 2/1.77 5.05 EMX2, EMX1 0033627/cell adhesion mediated by integrin 2/1.77 5.05 ITGA6, NOV 0072001/renal system development 2/1.77 5.75 EMX2, ITGA6 0050795/regulation of behavior 2/1.77 6.44 HTR1A, HTR2A 0042493/response to drug 3/2.65 7.01 EMX2, EMX1, HTR2A 0009395/phospholipid catabolic process 2/1.77 7.82 PLCG2, ENPP2 0007626/locomotory behavior 3/2.65 7.86 GRM1, GRM6, OPRK1 0050770/regulation of axonogenesis 2/1.77 8.50 RET, CACNA1A 0007283/spermatogenesis 3/2.65 9.15 ADAMTS2, TYRO3, RGS2 0051209/release of sequestered calcium ion into cytosol 2/1.77 9.18 PLCG2, HTR2A Open in new tab To further identify which genes above participate in cell signaling pathways, the KEGG classic pathway database was used to perform signal pathway enrichment analysis on these three gene sets in Table 2. According to KEGG classic pathway database, genes in all three sets are found to be involved in neuroactive ligand–receptor interaction pathway. Genes in both set #2 and set #3 are found to be involved in axon guidance pathway and other pathways related to synapse. Genes in set #2 are found to be related to more bioactivities of synapse, such as serotonergic synapse, dopaminergic synapse pathways, etc. Details are found in Table 4. Table 4 Result of pathway analysis of three gene sets based on KEGG classic pathway database Pathway . Count/Percentage (%) . P-value . Genes . Gene set 1: Top 10% AUC in both regions cjc04080:Neuroactive ligand–receptor interaction 425 1.50E−04 GRM7, GRID2, CHRN3, GLRA3 Gene set 2: Top 10% AUC in only calcarine sulcus region cjc04724:Glutamatergic synapse 6/5.46 5.30E−06 PRKCG, GRIN2B, HOMER3, SLC1A1, SHANK1, SHANK3 cjc04080:Neuroactive ligand–receptor interaction 4/3.64 3.00E−04 ADRA1A, CHRNA2, GLRB, GRIN2B cjc04726:Serotonergic synapse 4/3.64 3.90E−03 PRKCG, SLC6A4, DDC, KCNN2 cjc04150:mTOR signaling pathway 3/2.73 5.90E−03 PRKCG, EIF4EBP1, RPS6KA6 cjc04728:Dopaminergic synapse 4/3.64 7.60E−03 MAPK11, PRKCG, GRIN2B, DDC cjc04360:Axon guidance 3/2.73 7.60E−03 PLXNA4, SEMA3A, SEMA6D cjc05031:Amphetamine addiction 3/2.73 8.00E−03 PRKCG, GRIN2B, DDC cjc04720:Long-term potentiation 3/2.73 8.70E−03 PRKCG, GRIN2B, RPS6KA6 cjc04974:Protein digestion and absorption 2/1.82 1.60E−02 COL12A1, SLC1A1 Gene set 3: Top 10% AUC in only lateral fissure region cjc04080:Neuroactive ligand–receptor interaction 9/7.96 1.30E−07 GRM6, HCRTR2, HTR1A, CHRM5, GRM1, GRIK3, GRM2, OPRK1, HTR2A cjc04724:Glutamatergic synapse 6/5.31 2.10E−05 CACNA1A, SLC17A6, GRM1, GRM6, GRM2, GRIK3 cjc04020:Calcium signaling pathway 6/5.31 5.10E−04 PLCG2, CACNA1A, CHRM5, GRM1, ATP2A2, HTR2A cjc04730:Long-term depression 2/1.77 1.1E−02 CACNA1A, GRM1 cjc04360:Axon guidance 3/2.65 1.5E−02 EFNB3, PLXNB2, ITGB1 Pathway . Count/Percentage (%) . P-value . Genes . Gene set 1: Top 10% AUC in both regions cjc04080:Neuroactive ligand–receptor interaction 425 1.50E−04 GRM7, GRID2, CHRN3, GLRA3 Gene set 2: Top 10% AUC in only calcarine sulcus region cjc04724:Glutamatergic synapse 6/5.46 5.30E−06 PRKCG, GRIN2B, HOMER3, SLC1A1, SHANK1, SHANK3 cjc04080:Neuroactive ligand–receptor interaction 4/3.64 3.00E−04 ADRA1A, CHRNA2, GLRB, GRIN2B cjc04726:Serotonergic synapse 4/3.64 3.90E−03 PRKCG, SLC6A4, DDC, KCNN2 cjc04150:mTOR signaling pathway 3/2.73 5.90E−03 PRKCG, EIF4EBP1, RPS6KA6 cjc04728:Dopaminergic synapse 4/3.64 7.60E−03 MAPK11, PRKCG, GRIN2B, DDC cjc04360:Axon guidance 3/2.73 7.60E−03 PLXNA4, SEMA3A, SEMA6D cjc05031:Amphetamine addiction 3/2.73 8.00E−03 PRKCG, GRIN2B, DDC cjc04720:Long-term potentiation 3/2.73 8.70E−03 PRKCG, GRIN2B, RPS6KA6 cjc04974:Protein digestion and absorption 2/1.82 1.60E−02 COL12A1, SLC1A1 Gene set 3: Top 10% AUC in only lateral fissure region cjc04080:Neuroactive ligand–receptor interaction 9/7.96 1.30E−07 GRM6, HCRTR2, HTR1A, CHRM5, GRM1, GRIK3, GRM2, OPRK1, HTR2A cjc04724:Glutamatergic synapse 6/5.31 2.10E−05 CACNA1A, SLC17A6, GRM1, GRM6, GRM2, GRIK3 cjc04020:Calcium signaling pathway 6/5.31 5.10E−04 PLCG2, CACNA1A, CHRM5, GRM1, ATP2A2, HTR2A cjc04730:Long-term depression 2/1.77 1.1E−02 CACNA1A, GRM1 cjc04360:Axon guidance 3/2.65 1.5E−02 EFNB3, PLXNB2, ITGB1 Open in new tab Table 4 Result of pathway analysis of three gene sets based on KEGG classic pathway database Pathway . Count/Percentage (%) . P-value . Genes . Gene set 1: Top 10% AUC in both regions cjc04080:Neuroactive ligand–receptor interaction 425 1.50E−04 GRM7, GRID2, CHRN3, GLRA3 Gene set 2: Top 10% AUC in only calcarine sulcus region cjc04724:Glutamatergic synapse 6/5.46 5.30E−06 PRKCG, GRIN2B, HOMER3, SLC1A1, SHANK1, SHANK3 cjc04080:Neuroactive ligand–receptor interaction 4/3.64 3.00E−04 ADRA1A, CHRNA2, GLRB, GRIN2B cjc04726:Serotonergic synapse 4/3.64 3.90E−03 PRKCG, SLC6A4, DDC, KCNN2 cjc04150:mTOR signaling pathway 3/2.73 5.90E−03 PRKCG, EIF4EBP1, RPS6KA6 cjc04728:Dopaminergic synapse 4/3.64 7.60E−03 MAPK11, PRKCG, GRIN2B, DDC cjc04360:Axon guidance 3/2.73 7.60E−03 PLXNA4, SEMA3A, SEMA6D cjc05031:Amphetamine addiction 3/2.73 8.00E−03 PRKCG, GRIN2B, DDC cjc04720:Long-term potentiation 3/2.73 8.70E−03 PRKCG, GRIN2B, RPS6KA6 cjc04974:Protein digestion and absorption 2/1.82 1.60E−02 COL12A1, SLC1A1 Gene set 3: Top 10% AUC in only lateral fissure region cjc04080:Neuroactive ligand–receptor interaction 9/7.96 1.30E−07 GRM6, HCRTR2, HTR1A, CHRM5, GRM1, GRIK3, GRM2, OPRK1, HTR2A cjc04724:Glutamatergic synapse 6/5.31 2.10E−05 CACNA1A, SLC17A6, GRM1, GRM6, GRM2, GRIK3 cjc04020:Calcium signaling pathway 6/5.31 5.10E−04 PLCG2, CACNA1A, CHRM5, GRM1, ATP2A2, HTR2A cjc04730:Long-term depression 2/1.77 1.1E−02 CACNA1A, GRM1 cjc04360:Axon guidance 3/2.65 1.5E−02 EFNB3, PLXNB2, ITGB1 Pathway . Count/Percentage (%) . P-value . Genes . Gene set 1: Top 10% AUC in both regions cjc04080:Neuroactive ligand–receptor interaction 425 1.50E−04 GRM7, GRID2, CHRN3, GLRA3 Gene set 2: Top 10% AUC in only calcarine sulcus region cjc04724:Glutamatergic synapse 6/5.46 5.30E−06 PRKCG, GRIN2B, HOMER3, SLC1A1, SHANK1, SHANK3 cjc04080:Neuroactive ligand–receptor interaction 4/3.64 3.00E−04 ADRA1A, CHRNA2, GLRB, GRIN2B cjc04726:Serotonergic synapse 4/3.64 3.90E−03 PRKCG, SLC6A4, DDC, KCNN2 cjc04150:mTOR signaling pathway 3/2.73 5.90E−03 PRKCG, EIF4EBP1, RPS6KA6 cjc04728:Dopaminergic synapse 4/3.64 7.60E−03 MAPK11, PRKCG, GRIN2B, DDC cjc04360:Axon guidance 3/2.73 7.60E−03 PLXNA4, SEMA3A, SEMA6D cjc05031:Amphetamine addiction 3/2.73 8.00E−03 PRKCG, GRIN2B, DDC cjc04720:Long-term potentiation 3/2.73 8.70E−03 PRKCG, GRIN2B, RPS6KA6 cjc04974:Protein digestion and absorption 2/1.82 1.60E−02 COL12A1, SLC1A1 Gene set 3: Top 10% AUC in only lateral fissure region cjc04080:Neuroactive ligand–receptor interaction 9/7.96 1.30E−07 GRM6, HCRTR2, HTR1A, CHRM5, GRM1, GRIK3, GRM2, OPRK1, HTR2A cjc04724:Glutamatergic synapse 6/5.31 2.10E−05 CACNA1A, SLC17A6, GRM1, GRM6, GRM2, GRIK3 cjc04020:Calcium signaling pathway 6/5.31 5.10E−04 PLCG2, CACNA1A, CHRM5, GRM1, ATP2A2, HTR2A cjc04730:Long-term depression 2/1.77 1.1E−02 CACNA1A, GRM1 cjc04360:Axon guidance 3/2.65 1.5E−02 EFNB3, PLXNB2, ITGB1 Open in new tab Figure 4 Open in new tabDownload slide Simple description of the pathway-related genes in the axon guidance pathway. Genes in dark gray boxes represent that gene expression is more intensive in gyri than in sulci. Genes in light gray boxes represent that gene expression in sulci is more intensive. Figure 4 Open in new tabDownload slide Simple description of the pathway-related genes in the axon guidance pathway. Genes in dark gray boxes represent that gene expression is more intensive in gyri than in sulci. Genes in light gray boxes represent that gene expression in sulci is more intensive. In summary, genes related to axonal development, regulation of dendritic, glutamatergic synapse, and other neuronal signal can be found in all three gene sets, emphasizing their important relation to cortical morphology. More importantly, by comparing gene sets #2 and #3, we found that more genes in lateral fissure region are related to cortex regionalization and cell junction and adhesion, whereas more genes in calcarine sulcus region are related to nervous system development and axon guidance. Genes Related to Axon Guidance Pathway In the previous sections, a data-driven approach was used to identify the genes, through which the expression is significantly different between convex and concave patterns. In this section, we particularly focus on genes that relate to axons. A total of 27 genes out of all the 1116 genes were found in axon guidance pathway, and they were then related to convex–concave difference. The 27 genes and their interaction in axon guidance pathway are demonstrated in Figure 4. Figure 5 Open in new tabDownload slide (a) Average expression of genes related to axon guidance pathway in convex and concave patterns in lateral fissure region. (b) Average gene expression in convex and concave patterns in calcarine sulcus region. The left y-axis represents average gene expression value. The right y-axis represents the AUC values of these genes in gene-level SVM models. Genes are sorted by AUC values in a descending order in both subfigures. Figure 5 Open in new tabDownload slide (a) Average expression of genes related to axon guidance pathway in convex and concave patterns in lateral fissure region. (b) Average gene expression in convex and concave patterns in calcarine sulcus region. The left y-axis represents average gene expression value. The right y-axis represents the AUC values of these genes in gene-level SVM models. Genes are sorted by AUC values in a descending order in both subfigures. In lateral fissure region, average expression value of the 27 genes in convex patterns is 0.14 ± 0.08 and the average gene expression value in concave ones is 0.18 ± 0.10. The average contrast value of gene expression in convex and concave patterns of these genes is −0.14. In calcarine sulcus region, average expression value of these genes in convex patterns is 0.22 ± 0.12 and the average gene expression value in concave ones is 0.17 ± 0.11. The average contrast value of gene expression is 0.23. As a comparison, another 27 genes are randomly selected from other genes for 5000 times. The average gene expression in convex ones is 0.05 ± 0.03 and is 0.06 ± 0.03 in concave ones, whereas the average contrast of gene expression in convex and concave patterns of randomly selected 27 genes is −0.02. Average gene expression value on the cortical surface of the 27 genes is much higher than other genes for both convex and concave patterns tested by low P-value (≪0.001) via a one-sample paired t-test. We also investigate the performance of these 27 genes in the gene-level SVM models. AUC values and gene expression in these regions of 27 genes are shown in Figure 5. ROBO1, EFNB3, SEMA6D, PLXNA4, MAPK11 among these 27 genes could be found in the top 10% AUC values in either lateral or calcarine sulcus regions. This result suggests that genes related to axon guidance pathway show significantly different expressions between convex and concave patterns compared with other genes. Finally, we found that more genes show better classification performance in calcarine sulcus region than in lateral fissure region, suggesting that the formation of convex–concave patterns in calcarine sulcus region could be more related to axon growth and regulation than that in lateral fissure. Image-based Structural Features of Convex and Concave Patterns The fiber density and cortical thickness on convex and concave patterns in lateral fissure and calcarine sulcus regions are reported in Table 5, as well as the convex–concave contrasts of the two structural features. It is found that the convex–concave contrast of cortical thickness on calcarine region is not as pronounced as the one on lateral region, whereas the contrast of fiber density on calcarine region is more pronounced than the one on lateral region. Table 5 Cortical thickness and fiber density of concave and convex patterns in lateral fissure region and calcarine sulcus region . Calcarine convex . Calcarine concave . Lateral convex . Lateral concave . Cortical thickness (mm) 1.43 ± 0.37 1.34 ± 0.32 1.68 ± 0.45 1.35 ± 0.50 Contrast (conv.–conc.) 0.09 ± 0.01 0.33 ± 0.14 Fiber density (No. per are ×10−2) 2.96 ± 2.45 1.73 ± 1.55 3.16 ± 0.58 2.20 ± 0.90 Contrast (conv.–conc.) 1.21 ± 1.03 0.95 ± 0.69 . Calcarine convex . Calcarine concave . Lateral convex . Lateral concave . Cortical thickness (mm) 1.43 ± 0.37 1.34 ± 0.32 1.68 ± 0.45 1.35 ± 0.50 Contrast (conv.–conc.) 0.09 ± 0.01 0.33 ± 0.14 Fiber density (No. per are ×10−2) 2.96 ± 2.45 1.73 ± 1.55 3.16 ± 0.58 2.20 ± 0.90 Contrast (conv.–conc.) 1.21 ± 1.03 0.95 ± 0.69 The convex–concave contrasts are also reported. Open in new tab Table 5 Cortical thickness and fiber density of concave and convex patterns in lateral fissure region and calcarine sulcus region . Calcarine convex . Calcarine concave . Lateral convex . Lateral concave . Cortical thickness (mm) 1.43 ± 0.37 1.34 ± 0.32 1.68 ± 0.45 1.35 ± 0.50 Contrast (conv.–conc.) 0.09 ± 0.01 0.33 ± 0.14 Fiber density (No. per are ×10−2) 2.96 ± 2.45 1.73 ± 1.55 3.16 ± 0.58 2.20 ± 0.90 Contrast (conv.–conc.) 1.21 ± 1.03 0.95 ± 0.69 . Calcarine convex . Calcarine concave . Lateral convex . Lateral concave . Cortical thickness (mm) 1.43 ± 0.37 1.34 ± 0.32 1.68 ± 0.45 1.35 ± 0.50 Contrast (conv.–conc.) 0.09 ± 0.01 0.33 ± 0.14 Fiber density (No. per are ×10−2) 2.96 ± 2.45 1.73 ± 1.55 3.16 ± 0.58 2.20 ± 0.90 Contrast (conv.–conc.) 1.21 ± 1.03 0.95 ± 0.69 The convex–concave contrasts are also reported. Open in new tab Conclusion and Discussion In this paper, we applied SVM models, pathway analysis, and image-based analysis to investigate the difference between convex and concave cortical folding patterns of marmoset brains. Expression value of 1116 genes were automatically computed and applied to classify the two folding patterns via SVM models. By means of GO enrichment analysis and KEGG enrichment analysis, functions of these genes with top 10% AUC values in classifying convex and concave patterns in either lateral fissure region or calcarine sulcus region are found to be closely associated with synapse and cell growth, axon guidance, and neural system development, in accordance with the cortical folding patterns hypotheses, which are related to axons gene regulation (Beck et al. 1995; Rakic 2004; Barkovich et al. 2012; Stahl et al. 2013). On the other perspective, genes selected from calcarine sulcus region tend to be more associated with axon and nervous system development, whereas those from lateral fissure region seem to be more associated with cell interaction and cerebral cortex regionalization. This result is in line with MRI- and DTI-derived results that the convex–concave contrast of fiber density is more pronounced in calcarine region, whereas the one of cortical thickness is more pronounced in lateral region. In addition to the concordance between the image-based results and the GO and KEGG enrichment results, we also found disparity across different regions. The difference between gene sets #2 (calcarine sulcus) and #3 (lateral sulcus) could be related to different mechanisms of cortex formation. Cortex development was suggested to occur via two sequential processes. One is that the folding (including the lateral fissure) suggested to be conserved across primates, whereas the other is the evolved folding (including the calcarine sulcus) (Namba et al. 2019). Contribution of neuron progenitor cell proliferation and neuron production is greater to lateral fissure than calcarine sulcus, which is in line with the observation that lateral region develops before the onset of cortical gyrification, whereas the contribution of neuronal migration possibly induced by axonogenesis and gliogenesis could be greater to calcarine sulcus than lateral fissure. These conclusions gain partial supports from our postnatal analyses (the summary in sections Neuroscientific Interpretations of Genes with High Classification Performance and Genes Related to Axon Guidance Pathway and Table 5). As for the inference to structural wiring and functioning patterns from this work, although axon growth and placement were not the only factor to cortical convolution, they are indeed suggested to strongly correlated to the landscape of cortex and brain functioning patterns. In particular, many studies found that long-range axons are radially distributed in the convex folds while circumferentially course along the deep boundaries of the concave folds (Xu et al. 2010; Nie et al. 2012; Budde and Annese 2013; Chen et al. 2013; Zhang et al. 2014). It appears that convex folds are far from each other, making the dense connection between them against the parsimonious principle of wiring cost. However, wiring diagram of cortex was demonstrated to emphasize the extra importance of some regions, which are usually recognized as hubs with higher density of connections (Bullmore and Sporns 2012). In this sense, the convex cortex could be suggested to serve as hubs than other morphological patterns (Zhang et al. 2020). As for associating the genes with brain functions, given that calcarine sulcus was accepted as one of the unimodal cortical regions (Huntenburg et al. 2018) and lateral fissure is related to a variety of multimodal functions, such as audio–visual integration (Bushara et al. 2003), emotion (Phan et al. 2002), salience (Taylor et al. 2009), and etc., we might make suggestions as follows: If genes are identified to distinguish convex patterns from concave ones, we can always find some of them to be related to axonal development; if these genes are found on unimodal cortex, they will be more likely to be related to nervous system development and axon guidance; if these genes are found on multimodal cortex, they will be more likely to be related to cortex regionalization and cell junction and adhesion. There are some limitations and caveats in this work. A dataset to study gyrification is expected to include longitudinal (or at least cross-sectional) samples around a time when the cortex starts to fold. Because the ISH data here is acquired in the same growth period, we limited our interests in how gene expression contrasts between cortical folding patterns, as well as how this contrast is associated with cortical anatomy, white matter connectivity, and brain functions. These results are expected to provide clue to cortical maturation. DTI-derived fiber density is an effective measurement to quantify macroscale long-range structural connectivity strength. Fibers were proved to be much denser in convex patterns than concave patterns among different species (Nie et al. 2012; Chen et al. 2013; Li et al. 2015). However, it should be mentioned that gyral bias problem could be induced by using DTI to estimate white matter axons and their exit or entrance to gray matter (Maier-Hein et al. 2017; Schilling et al. 2018). This is one of the intrinsic limitations of diffusion MRI (dMRI) technique and cannot be overcome. Other limitations of dMRI could also bias our observation, such as its low spatial resolution, which results in uncertainty in voxels where fiber fanning, crossing, and kissing are pronounced, and its incapability in identifying the short-range axonal pathways within the cortex and those in superficial white matters (Reveley et al. 2015). Therefore, more micro-scale imaging analyses will be needed to further validate the DTI-based fiber density analyses. In this work, we only considered these genes directly related to axon guidance pathway, whereas genes in other pathways can also have an impact on axon growth and formation of white matter fibers. For example, genes in regulation of actin cytoskeleton pathway, MAPK signaling pathway, WNT signaling pathway, and other pathways can influence the interaction of genes in axon guidance pathway and indirectly influence the process of axon growth (Cobb 1999; Maro et al. 2009; Nix et al. 2011). In summary, our works not only identified the significantly differentially expressed genes on convex and concave patterns, but also quantified and visualized their different expression patterns and provided a systematic and comprehensive viewpoint of the molecular and cellular differences between different cortical folding patterns and across cortical regions, which warrants broader and deeper neurobiological studies in the future. The results in this work show that the methods could be applied to other regions to detect the ability of genes of classifying convex and concave patterns and to a variety of other scenarios. As an example, in the future, we can apply this method in the same regions on normal and disease samples, and the AUC values and other metrics of SVM models could suggest a relationship between genes and the disease. We will also pay more attention to the expression patterns in different growth stage by studying longitudinal growth and variation of gene expression and its relation to cortical gyrification. Notes The authors thank David Schaeffer for sharing the marmoset DTI data used in this study. Conflict of Interest: The authors declare that there are no conflicts of interest. Funding National Natural Science Foundation of China (31971288, U1801265, 31671005 to T.Z.); National Natural Science Foundation of China (61936007 to L.G.). References Ango F , Pin JP, Tu JC, Xiao B, Worley PF, Bockaert J, Fagni L. 2000 . Dendritic and axonal targeting of type 5 metabotropic glutamate receptor is regulated by Homer1 proteins and neuronal excitation . J Neurosci . 20 : 8710 – 8716 . Google Scholar Crossref Search ADS PubMed WorldCat Andersson JL , Skare S. 2010 . Image distortion and its correction in diffusion MRI . In: Jones DK. Diffusion mri . New York: Oxford University Press. p 285 – 302 . Google Scholar OpenURL Placeholder Text WorldCat Andersson JL , Sotiropoulos SN. 2016 . An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging . Neuroimage . 125 : 1063 – 1078 . Google Scholar Crossref Search ADS PubMed WorldCat Arnold PD , Sicard T, Burroughs E, Richter MA, Kennedy JL. 2006 . Glutamate transporter gene SLC1A1 associated with obsessive-compulsive disorder . Arch Gen Psychiatry . 63 ( 7 ): 769 – 776 . Google Scholar Crossref Search ADS PubMed WorldCat Ashburner M , Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT et al. 2000 . Gene ontology: tool for the unification of biology . Nat Genet . 25 : 25 – 29 . Google Scholar Crossref Search ADS PubMed WorldCat Barkovich AJ , Guerrini R, Kuzniecky RI, Jackson GD, Dobyns WB. 2012 . A developmental and genetic classification for malformations of cortical development: update 2012 . Brain . 135 : 1348 – 1369 . Google Scholar Crossref Search ADS PubMed WorldCat Bartley AJ , Jones DW, Weinberger DR. 1997 . Genetic variability of human brain size and cortical gyral patterns . Brain . 120 : 257 – 269 . Google Scholar Crossref Search ADS PubMed WorldCat Beck KD , Powell-Braxtont L, Widmer HR, Valverde J, Hefti F. 1995 . Igf1 gene disruption results in reduced brain size, CNS hypomyelination, and loss of hippocampal granule and striatal parvalbumin-containing neurons . Neuron . 14 : 717 – 730 . Google Scholar Crossref Search ADS PubMed WorldCat Beffert U , Dillon GM, Sullivan JM, Stuart CE, Gilbert JP, Kambouris JA, Ho A. 2012 . Microtubule plus-end tracking protein CLASP2 regulates neuronal polarity and synaptic function . J Neurosci . 32 : 13906 – 13916 . Google Scholar Crossref Search ADS PubMed WorldCat Brandt SJ , Niedel JE, Bell RM, Young III WS. 1987 . Distinct patterns of expression of different protein kinase C mRNAs in rat tissues . Cell . 49 ( 1 ): 57 – 63 . Google Scholar Crossref Search ADS PubMed WorldCat Budde MD , Annese J. 2013 . Quantification of anisotropy and fiber orientation in human brain histological sections . Front Integr Neurosci . 7 : 3 . Google Scholar Crossref Search ADS PubMed WorldCat Bullmore E , Sporns O. 2009 . Complex brain networks: graph theoretical analysis of structural and functional systems . Nat Rev Neurosci . 10 ( 3 ): 186 . Google Scholar Crossref Search ADS PubMed WorldCat Bullmore E , Sporns O. 2012 . The economy of brain network organization . Nat Rev Neurosci . 13 ( 5 ): 336 – 349 . Google Scholar Crossref Search ADS PubMed WorldCat Bushara KO , Hanakawa T, Immisch I, Toma K, Kansaku K, Hallett M. 2003 . Neural correlates of cross-modal binding . Nat Neurosci . 6 ( 2 ): 190 – 195 . Google Scholar Crossref Search ADS PubMed WorldCat Chen H , Zhang T, Guo L, Li K, Yu X, Li L, Hu X, Han J, Hu X, Liu T. 2013 . Coevolution of Gyral folding and structural connection patterns in primate brains . Cereb Cortex . 23 : 1208 – 1217 . Google Scholar Crossref Search ADS PubMed WorldCat Cortes C , Vapnik V. 1995 . Support-vector networks . Mach Learn . 20 : 273 – 297 . Google Scholar OpenURL Placeholder Text WorldCat Cobb MH . 1999 . MAP kinase pathways . Prog Biophys Mol Biol . 71 ( 3–4 ): 479 – 500 . Google Scholar Crossref Search ADS PubMed WorldCat Craig AM , Kang Y. 2007 . Neurexin–neuroligin signaling in synapse development . Curr Opin Neurobiol . 17 : 43 – 52 . Google Scholar Crossref Search ADS PubMed WorldCat Dale AM , Fischl B, Sereno MI. 1999 . Cortical surface-based analysis: I. segmentation and surface reconstruction . Neuroimage . 9 ( 2 ): 179 – 194 . Google Scholar Crossref Search ADS PubMed WorldCat de Castro F , Seal R, Maggi R, on behalf of Group of HGNC consultants for KAL1 nomenclature. 2016 . ANOS1: a unified nomenclature for Kallmann syndrome 1 gene (KAL1) and anosmin-1 . Brief Funct Genomics . 16 : 205 – 210 . Google Scholar Crossref Search ADS WorldCat Dickson BJ . 2002 . Molecular mechanisms of axon guidance . Science (80−) . 298 : 1959 . Google Scholar Crossref Search ADS WorldCat Fischl B . 2012 . FreeSurfer . Neuroimage . 62 : 774 – 781 . Google Scholar Crossref Search ADS PubMed WorldCat Fischl B , Wald LL. 2007 . Phase maps reveal cortical architecture . Proc Natl Acad Sci . 104 ( 28 ): 11513 – 11514 . Google Scholar Crossref Search ADS PubMed WorldCat Fischl B , Sereno MI, Dale AM. 1999 . Cortical surface-based analysis: II: inflation, flattening, and a surface-based coordinate system . Neuroimage . 9 : 195 – 207 . Google Scholar Crossref Search ADS PubMed WorldCat Fischl B , Salat DH, Busa E, Albert M, Dieterich M, Haselgrove C, van der Kouwe AJ, Killiany R, Kennedy D, Klaveness S et al. 2002 . Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain . Neuron . 33 : 341 – 355 . Google Scholar Crossref Search ADS PubMed WorldCat Fischl B , Salat DH, van der Kouwe AJ, Makris N, Ségonne F, Quinn BT, Dale AM. 2004 . Sequence-independent segmentation of magnetic resonance images . Neuroimage . 23 : S69 – S84 . Google Scholar Crossref Search ADS PubMed WorldCat García-Campos MA , Espinal-Enríquez J, Hernández-Lemus E. 2015 . Pathway analysis: state of the art . Front Physiol . 6 : 383 . Google Scholar Crossref Search ADS PubMed WorldCat Gomez TM , Zheng JQ. 2006 . The molecular basis for calcium-dependent axon pathfinding . Nat Rev Neurosci . 7 : 115 – 125 . Google Scholar Crossref Search ADS PubMed WorldCat Goshtasby AA . 2005 . 2-D and 3-D image registration: for medical, remote sensing, and industrial applications . New Jersey : John Wiley & Sons . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hills LB , Masri A, Konno K, Kakegawa W, Lam AT, Lim-Melia E, Chandy N, Hill RS, Partlow JN, Al-Saffar M et al. 2013 . Deletions in GRID2 lead to a recessive syndrome of cerebellar ataxia and tonic upgaze in humans . Neurology . 81 ( 16 ): 1378 – 1386 . Google Scholar Crossref Search ADS PubMed WorldCat Hollmann M , Heinemann S. 1994 . Cloned glutamate receptors . Annu Rev Neurosci . 17 ( 1 ): 31 – 108 . Google Scholar Crossref Search ADS PubMed WorldCat Honey CJ , Thivierge JP, Sporns O. 2010 . Can structure predict function in the human brain? Neuroimage . 52 ( 3 ): 766 – 776 . Google Scholar Crossref Search ADS PubMed WorldCat Huntenburg JM , Bazin PL, Margulies DS. 2018 . Large-scale gradients in human cortical organization . Trends Cogn Sci . 22 ( 1 ): 21 – 31 . Google Scholar Crossref Search ADS PubMed WorldCat Jenkinson M , Bannister P, Brady M, Smith S. 2002 . Improved optimization for the robust and accurate linear registration and motion correction of brain images . Neuroimage . 17 ( 2 ): 825 – 841 . Google Scholar Crossref Search ADS PubMed WorldCat Jenkinson M , Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. 2012 . FSL . Neuroimage . 62 : 782 – 790 . Google Scholar Crossref Search ADS PubMed WorldCat Jeurissen B , Leemans A, Jones DK, Tournier JD, Sijbers J. 2011 . Probabilistic fiber tracking using the residual bootstrap with constrained spherical deconvolution . Hum Brain Mapp . 32 ( 3 ): 461 – 479 . Google Scholar Crossref Search ADS PubMed WorldCat Joseph RM . 2014 . Neuronatin gene: imprinted and misfolded: studies in Lafora disease, diabetes and cancer may implicate NNAT-aggregates as a common downstream participant in neuronal loss . Genomics . 103 : 183 – 188 . Google Scholar Crossref Search ADS PubMed WorldCat Kidd T , Bland KS, Goodman CS. 1999 . Slit is the midline repellent for the robo receptor in Drosophila . Cell . 96 ( 6 ): 785 – 794 . Google Scholar Crossref Search ADS PubMed WorldCat Kleene R , Chaudhary H, Karl N, Katic J, Kotarska A, Guitart K, Schachner M. 2015 . Interaction between CHL1 and serotonin receptor 2c regulates signal transduction and behavior in mice . J Cell Sci . 128 ( 24 ): 4642 – 4652 . Google Scholar Crossref Search ADS PubMed WorldCat Knaus P , Marquèze-Pouey B, Scherer H, Betzt H. 1990 . Synaptoporin, a novel putative channel protein of synaptic vesicles . Neuron . 5 : 453 – 462 . Google Scholar Crossref Search ADS PubMed WorldCat Kriegstein A , Noctor S, Martínez-cerdeño V. 2006 . Patterns of neural stem and progenitor cell division may underlie evolutionary cortical expansion . Nat Rev Neurosci . 7 : 883 – 890 . Google Scholar Crossref Search ADS PubMed WorldCat Lein ES , Hawrylycz MJ, Ao N, Ayres M, Bensinger A, Bernard A, Boe AF, Boguski MS, Brockway KS, Byrnes EJ et al. 2007 . Genome-wide atlas of gene expression in the adult mouse brain . Nature . 445 : 168 – 176 . Google Scholar Crossref Search ADS PubMed WorldCat Li G , Liu T, Ni D, Lin W, Gilmore JH, Shen D. 2015 . Spatiotemporal patterns of cortical fiber density in developing infants, and their relationship with cortical thickness . Hum Brain Mapp . 36 ( 12 ): 5183 – 5195 . Google Scholar Crossref Search ADS PubMed WorldCat Li Y , Chen H, Jiang X, Li X, Lv J, Li M, Peng H, Tsien JZ, Liu T. 2017 . Transcriptome architecture of adult mouse brain revealed by sparse coding of genome-wide in situ hybridization images . Neuroinformatics . 15 : 285 – 295 . Google Scholar Crossref Search ADS PubMed WorldCat Li Y , Huang H, Chen H, Liu T. 2018 . Deep neural networks for exploration of transcriptome of adult mouse brain . IEEE ACM Trans Comput Bi . 17 ( 2 ): 536 – 546 . Google Scholar OpenURL Placeholder Text WorldCat Lindsay RM , Thoenen H, Barde YA. 1985 . Placode and neural crest-derived sensory neurons are responsive at early developmental stages to brain-derived neurotrophic factor . Dev Biol . 112 : 319 – 328 . Google Scholar Crossref Search ADS PubMed WorldCat Liu H , Zhang S, Jiang X, Zhang T, Huang H, Ge F, Zhao L, Li X, Hu X, Han J et al. 2019 . The cerebral cortex is bisectionally segregated into two fundamentally different functional units of gyri and sulci . Cereb Cortex . 29 ( 10 ): 4238 – 4252 . Google Scholar Crossref Search ADS PubMed WorldCat Maier-Hein KH , Neher PF, Houde JC, Côté MA, Garyfallidis E, Zhong J, Chamberland M, Yeh FC, Lin YC, Ji Q et al. 2017 . The challenge of mapping the human connectome based on diffusion tractography . Nat Commun . 8 ( 1 ): 1 – 3 . Google Scholar Crossref Search ADS PubMed WorldCat Maro GS , Klassen MP, Shen K. 2009 . A β-catenin-dependent Wnt pathway mediates anteroposterior axon guidance in C. elegans motor neurons . PLoS One . 4 ( 3 ):e4690. Google Scholar OpenURL Placeholder Text WorldCat Namba T , Vaid S, Huttner WB. 2019 . Primate neocortex development and evolution: conserved versus evolved folding . J Comp Neurol . 527 ( 10 ): 1621 – 1632 . Google Scholar Crossref Search ADS PubMed WorldCat Nie J , Guo L, Li K, Wang Y, Chen G, Li L, Chen H, Deng F, Jiang X, Zhang T et al. 2012 . Axonal fiber terminations concentrate on gyri . Cereb Cortex . 22 : 2831 – 2839 . Google Scholar Crossref Search ADS PubMed WorldCat Nikolic Z , Laube B, Weber RG, Lichter P, Kioschis P, Poustka A, Becker CM. 1998 . The human glycine receptor subunit α3 Glra3 gene structure, chromosomal localization, and functional characterization of alternative transcripts . J Biol Chem . 273 ( 31 ): 19708 – 19714 . Google Scholar Crossref Search ADS PubMed WorldCat Nix P , Hisamoto N, Matsumoto K, Bastiani M. 2011 . Axon regeneration requires coordinate activation of p38 and JNK MAPK pathways . Proc Natl Acad Sci . 108 ( 26 ): 10738 – 10743 . Google Scholar Crossref Search ADS PubMed WorldCat Nordahl CW , Dierker D, Mostafavi I, Schumann CM, Rivera SM, Amaral DG, Van Essen DC. 2007 . Cortical folding abnormalities in autism revealed by surface-based morphometry . J Neurosci . 27 ( 43 ): 11725 – 11735 . Google Scholar Crossref Search ADS PubMed WorldCat Numata S , Ye T, Hyde TM, Guitart-Navarro X, Tao R, Wininger M, Colantuoni C, Weinberger DR, Kleinman JE, Lipska BK. 2012 . DNA methylation signatures in development and aging of the human prefrontal cortex . Am J Hum Genet . 90 : 260 – 272 . Google Scholar Crossref Search ADS PubMed WorldCat Ogata H , Goto S, Sato K, Fujibuchi W, Bono H, Kanehisa M. 1999 . KEGG: Kyoto Encyclopedia of genes and genomes . Nucleic Acids Res . 27 : 29 – 34 . Google Scholar Crossref Search ADS PubMed WorldCat Oksenberg N , Ahituv N. 2013 . The role of AUTS2 in neurodevelopment and human evolution . Trends Genet . 29 : 600 – 608 . Google Scholar Crossref Search ADS PubMed WorldCat Paxinos G , Watson C, Petrides M, Rosa M, Tokuno H. 2012 . The marmoset brain in stereotaxic coordinates . Cambridge : Elsevier Academic Press , pp. 1 – 59 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Phan KL , Wager T, Taylor SF, Liberzon I. 2002 . Functional neuroanatomy of emotion: a meta-analysis of emotion activation studies in PET and fMRI . Neuroimage . 16 ( 2 ): 331 – 348 . Google Scholar Crossref Search ADS PubMed WorldCat Platt J. 1998 . Sequential minimal optimization: a fast algorithm for training support vector machines . MSR-TR-98-14. Pohlkamp T , Xiao L, Sultana R, Bepari A, Bock HH, Henkemeyer M, Herz J. 2016 . Ephrin Bs and canonical Reelin signalling . Nature . 539 : E4 – E6 . Google Scholar Crossref Search ADS PubMed WorldCat Rakic P . 1988 . Specification of cerebral cortical areas . Science . 241 : 170 – 176 . Google Scholar Crossref Search ADS PubMed WorldCat Rakic P . 2004 . Genetic control of cortical convolutions . Science (80−) . 303 : 1983 – 1984 . Google Scholar Crossref Search ADS WorldCat Rash BG , Duque A, Morozov YM, Arellano JI, Micali N, Rakic P. 2019 . Gliogenesis in the outer subventricular zone promotes enlargement and gyrification of the primate cerebrum . Proc Natl Acad Sci . 116 ( 14 ): 7089 – 7094 . Google Scholar Crossref Search ADS PubMed WorldCat Razavi MJ , Zhang T, Chen H, Li Y, Platt S, Zhao Y, Guo L, Hu X, Wang X, Liu T. 2017 . Radial structure scaffolds convolution patterns of developing cerebral cortex . Front Comput Neurosci . 11 : 76 . Google Scholar Crossref Search ADS PubMed WorldCat Reveley C , Seth AK, Pierpaoli C, Silva AC, Yu D, Saunders RC, Leopold DA, Ye FQ. 2015 . Superficial white matter fiber systems impede detection of long-range cortical connections in diffusion MR tractography . Proc Natl Acad Sci . 112 ( 21 ): 2820 – 2828 . Google Scholar Crossref Search ADS WorldCat Richman DP , Stewart RM, Hutchinson JW, Caviness VS. 1975 . Mechanical model of brain convolutional development . Science . 189 : 18 – 21 . Google Scholar Crossref Search ADS PubMed WorldCat Sato D , Lionel AC, Leblond CS, Prasad A, Pinto D, Walker S, O’Connor I, Russell C, Drmic IE, Hamdan FF et al. 2012 . SHANK1 deletions in males with autism Spectrum disorder . Am J Hum Genet . 90 : 879 – 887 . Google Scholar Crossref Search ADS PubMed WorldCat Schaeffer DJ , Adam R, Gilbert KM, Gati JS, Li AX, Menon RS, Everling S. 2017 . Diffusion-weighted tractography in the common marmoset monkey at 9.4 T . J Neurophysiol . 118 ( 2 ): 1344 – 1354 . Google Scholar Crossref Search ADS PubMed WorldCat Schilling K , Gao Y, Janve V, Stepniewska I, Landman BA, Anderson AW. 2018 . Confirmation of a gyral bias in diffusion MRI fiber tractography . Hum Brain Mapp . 39 ( 3 ): 1449 – 1466 . Google Scholar Crossref Search ADS PubMed WorldCat Shimogori T , Abe A, Go Y, Hashikawa T, Kishi N, Kikuchi SS, Kita Y, Niimi K, Nishibe H, Yoda T et al. 2018 . Digital gene atlas of neonate common marmoset brain . Neurosci Res . 128 : 1 – 13 . Google Scholar Crossref Search ADS PubMed WorldCat Smart IHM . 2002 . Unique morphological features of the proliferative zones and Postmitotic compartments of the neural epithelium giving rise to striate and extrastriate cortex in the monkey . Cereb Cortex . 12 : 37 – 53 . Google Scholar Crossref Search ADS PubMed WorldCat Solomon SG , Rosa MG. 2014 . A simpler primate brain: the visual system of the marmoset monkey . Front Neural Circuit . 8 : 96 . Google Scholar OpenURL Placeholder Text WorldCat Stahl R , Walcher T, De Juan RC, Pilz GA, Cappello S, Irmler M, Sanz-Aquela JM, Beckers J, Blum R, Borrell V et al. 2013 . Trnp1 regulates expansion and folding of the mammalian cerebral cortex by control of radial glial fate . Cell . 153 : 535 – 549 . Google Scholar Crossref Search ADS PubMed WorldCat Striedter GF , Srinivasan S, Monuki ES. 2015 . Cortical folding: when, where, how, and why? Annu Rev Neurosci . 38 : 291 – 307 . Google Scholar Crossref Search ADS PubMed WorldCat Sun T , Hevner RF. 2014 . Growth and folding of the mammalian cerebral cortex: from molecules to malformations . Nat Rev Neurosci . 15 : 217 – 232 . Google Scholar Crossref Search ADS PubMed WorldCat Suto F , Ito K, Uemura M, Shimizu M, Shinkawa Y, Sanbo M, Fujisawa H. 2005 . Plexin-a4 mediates axon-repulsive activities of both secreted and transmembrane semaphorins and plays roles in nerve fiber guidance . J Neurosci . 25 ( 14 ): 3628 – 3637 . Google Scholar Crossref Search ADS PubMed WorldCat Takayama C , Nakagawa S, Watanabe M, Mishina M, Inoue Y. 1995 . Light- and electron-microscopic localization of the glutamate receptor channel delta 2 subunit in the mouse Purkinje cell . Neurosci Lett . 188 ( 2 ): 89 – 92 . Google Scholar Crossref Search ADS PubMed WorldCat Tamagnone L , Artigiani S, Chen H, He Z, Ming GL, Song HJ, Tessier-Lavigne M. 1999 . Plexins are a large family of receptors for transmembrane, secreted, and GPI-anchored semaphorins in vertebrates . Cell . 99 ( 1 ): 71 – 80 . Google Scholar Crossref Search ADS PubMed WorldCat Taylor KS , Seminowicz DA, Davis KD. 2009 . Two systems of resting state connectivity between the insula and cingulate cortex . Hum Brain Mapp . 30 ( 9 ): 2731 – 2745 . Google Scholar Crossref Search ADS PubMed WorldCat Uittenbogaard M , Baxter KK, Chiaramello A. 2010 . NeuroD6 genomic signature bridging neuronal differentiation to survival via the molecular chaperone network . J Neurosci Res . 88 ( 1 ): 33 – 54 . Google Scholar Crossref Search ADS PubMed WorldCat van den Heuvel MP , Mandl RC, Kahn RS, Hulshoff Pol HE. 2009 . Functionally linked resting-state networks reflect the underlying structural connectivity architecture of the human brain . Hum Brain Mapp . 30 ( 10 ): 3127 – 3141 . Google Scholar Crossref Search ADS PubMed WorldCat Van Essen DC . 1997 . A tension-based theory of morphogenesis and compact wiring in the central nervous system . Nature . 385 : 313 . Google Scholar Crossref Search ADS PubMed WorldCat Xia W , Liu Y, Jiao J. 2015 . GRM7 regulates embryonic neurogenesis via CREB and YAP . Stem Cell Rep . 4 ( 5 ): 795 – 810 . Google Scholar Crossref Search ADS WorldCat Xu G , Knutsen AK, Dikranian K, Kroenke CD, Bayly PV, Taber LA. 2010 . Axons pull on the brain, but tension does not drive cortical folding . J Biomech Eng . 132 ( 7 ): 071013 . Google Scholar Crossref Search ADS PubMed WorldCat Yeh F , Wedeen VJ, Tseng WI. 2010 . Generalized q-sampling imaging . IEEE Trans Med Imaging . 29 : 1626 – 1635 . Google Scholar Crossref Search ADS PubMed WorldCat Yeh F , Verstynen TD, Wang Y, Fernández-Miranda JC, Tseng WI. 2013 . Deterministic diffusion fiber tracking improved by quantitative anisotropy . PLoS One . 8 ( 11 ):e80713. Google Scholar OpenURL Placeholder Text WorldCat Yosimichi G , Nakanishi T, Nishida T, Hattori T, Takano-Yamamoto T, Takigawa M. 2001 . CTGF/Hcs24 induces chondrocyte differentiation through a p38 mitogen-activated protein kinase (p38MAPK), and proliferation through a p44/42 MAPK/extracellular-signal regulated kinase (ERK) . Eur J Biochem . 268 ( 23 ): 6058 – 6065 . Google Scholar Crossref Search ADS PubMed WorldCat Zamora-López G , Zhou C, Kurths J. 2011 . Exploring brain function from anatomical connectivity . Front Neurosci . 5 : 83 . Google Scholar Crossref Search ADS PubMed WorldCat Zeng T , Chen H, Fakhry A, Hu X, Liu T, Ji S. 2015 . Allen mouse brain atlases reveal different neural connection and gene expression patterns in cerebellum gyri and sulci . Brain Struct Funct . 220 : 2691 – 2703 . Google Scholar Crossref Search ADS PubMed WorldCat Zhang T , Chen H, Guo L, Li K, Li L, Zhang S, Shen D, Hu X, Liu T. 2014 . Characterization of U-shape streamline fibers: methods and applications . Med Image Anal . 18 : 795 – 807 . Google Scholar Crossref Search ADS PubMed WorldCat Zhang T , Li X, Jiang X, Ge F, Zhang S, Zhao L, Liu H, Huang Y, Wang X, Yang J et al. 2020 . Cortical 3-hinges could serve as hubs in cortico-cortical connective network . Brain Imaging Behav . in press. Google Scholar OpenURL Placeholder Text WorldCat Zhang S, Liu H, Huang H, Zhao Y, Jiang X, Bowers B, Guo L, Hu X, Sanches M, Liu T. 2018 . Deep learning models unveiled functional difference between cortical gyri and sulci . IEEE Trans on Biomed Eng . 66 ( 5 ): 1297 – 1308 . OpenURL Placeholder Text WorldCat Zilles K , Armstrong E, Schleicher A, Kretschmann HJ. 1988 . The human pattern of gyrification in the cerebral cortex . Anat Embryol . 179 : 173 – 179 . Google Scholar Crossref Search ADS WorldCat Appendix: Gene Enrichment Analysis and Pathway Analysis In the field of identifying sets of genes with similar function profiles, gene enrichment analysis and pathway analysis are widely used. GO enrichment analysis is widely applied in understanding how genes contribute to the biology of an organism at the molecular, cellular, and organism levels (Ashburner et al. 2000). Pathway analysis makes it possible to detect gene interactions and changes of gene expression, predict downstream in the pathway, and look for mechanisms that can explain causal inference in experiments (Ogata et al. 1999; García-Campos et al. 2015). A pathway is a complex model to describe a process, mechanism, or phenomenon with a graph that contains nodes and edges. Nodes and edges represent genes and their interactions, respectively. One specifically interesting pathway is the axon guidance pathway. This pathway is remarkable because axonal tension or compression is considered as one mechanism factor of gyrification. Axon guidance pathway is a description of genes and their interactions on how axons are guided towards their targets to form synaptic connections (Dickson 2002; Gomez and Zheng 2006). Growth cone, which exists at the end of an axon senses guidance cues through guidance cue receptors, influences cytoskeletal changes in this axon and further determines the axon growth, repulsion, and attraction. Genes in this pathway could possibly interact with each other and their products, via which the genes control guidance cues, guidance cue receptors, growth cone, and axon growth, leading to the differentiation of white matter fibers growth. Author notes Xiao Li and Tao Liu equally contribute to this work © The Author(s) 2020. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permission@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Marmoset Brain ISH Data Revealed Molecular Difference Between Cortical Folding Patterns
JF - Cerebral Cortex
DO - 10.1093/cercor/bhaa317
DA - 2021-02-05
UR - https://www.deepdyve.com/lp/oxford-university-press/marmoset-brain-ish-data-revealed-molecular-difference-between-cortical-3kXhg0JUlf
SP - 1660
EP - 1674
VL - 31
IS - 3
DP - DeepDyve
ER -