3DIV: A 3D-genome Interaction Viewer and database

Sung

et al. 2019

Preprint

Major progress in disease genetics has been made through genome-wide association studies (GWASs). One of the key tasks for post-GWAS analyses is to identify causal noncoding variants with regulatory function. Here, on the basis of > 2,000 functional features, we developed a convolutional neural network framework for combinatorial, nonlinear modeling of complex patterns shared by risk variants scattered among multiple associated loci. When applied for major psychiatric disorders and autoimmune diseases, neural and immune features, respectively, exhibited high explanatory power while reflecting the pathophysiology of the relevant disease. The predicted causal variants were concentrated in active regulatory regions of relevant cell types and tended to be in physical contact with transcription factors while residing in evolutionarily conserved regions and resulting in expression changes of genes related to the given disease. We demonstrate some examples of novel candidate causal variants and associated genes. Our method is expected to contribute to the identification and functional interpretation of causal noncoding variants in post-GWAS analyses.

Section: Examples Of Novel Candidate Causal Variantsmentioning

confidence: 99%

Convolutional neural network model to predict causal risk factors that share complex regulatory features

Sung

et al. 2019

Preprint

“…Neighboring chromatin interactions were investigated in intergenic loci using high-210 throughput chromatin conformation capture (Hi-C) data from the 3DIV database (42). 211…”

Section: Chromatin Interactions 209mentioning

confidence: 99%

Phenotypic clustering reveals distinct subtypes of polycystic ovary syndrome with novel genetic associations

Dapas¹,

Lin²,

Nadkarni

et al. 2019

Preprint

AD) 20 ghayes@northwestern.edu (MGH) 21 22 DISCLOSURE STATEMENT: The authors have nothing to disclose. 23 metabolic subtype (KCNH7/FIGN, P=1.0×10 -8 ). We have previously reported that rare 47 variants in DENND1A, a gene regulating androgen biosynthesis, were associated with PCOS 48 quantitative traits in a family-based whole genome sequencing analysis. We classified the 49 reproductive and metabolic subtypes in this family-based PCOS cohort and found that the 50 subtypes tended to cluster in families and that carriers of rare DENND1A variants were 51 significantly more likely to have the reproductive subtype of PCOS. Limitations of our study 52 were that only PCOS cases of European ancestry diagnosed by NIH criteria were included, 53 the sample sizes for the subtype GWAS were small, and the GWAS findings were not 54 replicated. 55 56 Conclusions 57In conclusion, we have found stable reproductive and metabolic subtypes of PCOS. Further, 58 these subtypes were associated with novel susceptibility loci. Our results suggest that these 59 subtypes are biologically relevant since they have distinct genetic architectures. This study 60 demonstrates how precise phenotypic delineation can be more powerful than increases in 61 sample size for genetic association studies. 62 63Understanding the genetic architecture of complex diseases is a central challenge in 64 human genetics (1-3). Often defined according to arbitrary diagnostic criteria, complex 65 diseases can represent the phenotypic convergence of numerous genetic etiologies (4)(5)(6)(7)(8). 66Recent studies in type 2 diabetes support the concept that there are disease subtypes with 67 distinct genetic architecture (7,8). Identifying and addressing genetic heterogeneity in 68 complex diseases could increase power to detect causal variants and improve treatment 69 efficacy (9). 70Polycystic ovary syndrome (PCOS) is a highly heritable, complex genetic disorder 71 affecting up to 15% of reproductive-age women worldwide, depending on the diagnostic 72 criteria applied (10). It is characterized by a variable constellation of reproductive and 73 metabolic abnormalities (Fig 1). It is the leading cause of anovulatory infertility and a major 74 risk factor for type 2 diabetes (T2D) in women (11). Despite these substantial morbidities, 75 the etiology(ies) of PCOS remains unknown (12). Accordingly, the commonly-used 76 diagnostic criteria for PCOS, the National Institutes of Health (NIH) criteria (13) and the 77 Rotterdam criteria (14,15), are based on expert opinion, rather than mechanistic insights, and 78 are designed to account for the diverse phenotypic presentations of PCOS. The NIH criteria 79 require the presence of hyperandrogenism (HA) and chronic oligo/anovulation or ovarian 80 dysfunction (OD) (13). The Rotterdam criteria include polycystic ovarian morphology 81 (PCOM) and require the presence of at least two of these three key reproductive traits, 82 resulting in three different affected phenotypes: HA and OD with or without PCOM, also 83 known as NIH PCOS, as w...

“…If a given DNM was located in the distal or proximal DHS of a gene, the gene was assigned as an 'NCD gene'. To verify the chromatin interactions putatively affecting the NCD genes, we obtained Hi-C dataset of human neuronal progenitor cell from an independent study 11 . Distance normalized interaction frequency matrix in 5kb resolution was used to plot the baseline Hi-C distribution.…”

Section: Defining Ncd Genes Using Chromatin Interactionsmentioning

confidence: 99%

“…In the heatmap, red, orange, and grey boxes respectively indicate transcription start site-related states, transcribed states, and repressed states, which were annotated by ChromHMM 72 . H3K4me3 peaks were independently adopted from 3D-genome Interaction Viewer and database11 . DLPFC, dorsolateral prefrontal cortex; fBrain, fetal brain; F, female; M, male; and NPC, neural progenitor cell 33.…”

mentioning

confidence: 99%

Noncoding de novo mutations contribute to autism spectrum disorder via chromatin interactions

Kim

et al. 2019

Preprint

Three-dimensional chromatin structures regulate gene expression across genome. The significance of de novo mutations (DNMs) affecting chromatin interactions in autism spectrum disorder (ASD) remains poorly understood. We generated 931 whole-genome sequences for Korean simplex families to detect DNMs and identified target genes dysregulated by noncoding DNMs via long-range chromatin interactions between regulatory elements. Notably, noncoding DNMs that affect chromatin interactions exhibited transcriptional dysregulation implicated in ASD risks. Correspondingly, target genes were significantly involved in histone modification, prenatal brain development, and pregnancy. Both noncoding and coding DNMs collectively contributed to low IQ in ASD. Indeed, noncoding DNMs resulted in alterations, via chromatin interactions, in target gene expression in primitive neural stem cells derived from human induced pluripotent stem cells from an ASD subject. The emerging neurodevelopmental genes, not previously implicated in ASD, include CTNNA2, GRB10, IKZF1, PDE3B, and BACE1.Our results were reproducible in 517 probands from MSSNG cohort. This work demonstrates that noncoding DNMs contribute to ASD via chromatin interactions. probands=29%, MSSNG probands=28%, and Korean siblings=21%) ( Supplementary Fig. 2c).To validate those chromatin interactions within fetal human brains, we utilized published Hi-C data of neuronal progenitor cells 11 as a baseline with which to compare Hi-C interactions for the NCD genes (see Methods). Reflecting the intensities of the Hi-C interactions, distancenormalized interaction frequencies for the proband NCD genes were significantly higher than those at baseline (Korean probands P=2.3x10 -4 , MSSNG probands P=5.0x10 -8 , Student's twosided t-test) ( Fig. 1b and Supplementary Fig. 2d). These results indicated that the chromatin interactions between NCD genes and noncoding DNMs were validated using the independent Hi-C data.Examining distances of the chromatin interactions, we observed, in the probands, 269 of the 363 NCD genes (74.1%) were discordant to genes nearest to the 230 DNMs (Fig. 1c). Moreover, 232 of the 363 NCD genes (63.9%) in the probands showed chromatin interactions over a distance of 100kb-500kb, whereas only 23 of the 129 genes nearest to the noncoding DNMs (17.8%) showed interactions over the same distance (P=2.4x10 -19 , Proportion test) ( Fig. 1c and Supplementary Fig. 2e). These results suggested that majority of the noncoding DNMs might affect their target NCD genes via long-range chromatin interactions.We then evaluated the functional properties of the 363 NCD genes in the probands in comparison to 129 of the nearest genes in the probands and to 113 NCD genes in the siblings.After 10,000 permutations, using genes differentially expressed in brains of ASD patients 12,13 , we noted significant overlap only for the proband NCD genes (Proband NCD genes P=0.01, proband nearest genes P=0.71, and sibling NCD genes P=0.12, Random permutation test) ( Fig. 1d). Besides, the expression le...