Biology subjects

Dey, K. K.

Publications and source records attributed to Dey, K. K..

26 records · Page 2Linked to original sources

bioRxiv2020

Integrative approaches to improve the informativeness of deep learning models for human complex diseases

Deep learning models have achieved great success in predicting genome-wide regulatory effects from DNA sequence, but recent work has reported that SNP annotations derived from these predictions contribute limited unique information for human complex disease. Here, we explore three integrative approaches to improve the disease informativeness of allelic-effect annotations (predicted difference between reference and variant alleles) constructed using several previously trained deep learning models: DeepSEA, Basenji and DeepBind (and a related machine learning model, deltaSVM). First, we employ gradient boosting to learn optimal combinations of deep learning annotations, using fine-mapped SNPs and matched control SNPs (on held-out chromosomes) for training. Second, we improve the specificity of these annotations by restricting them to SNPs implicated by (proximal and distal) SNP-to-gene (S2G) linking strategies, e.g. prioritizing SNPs involved in gene regulation. Third, we predict gene expression (and derive allelic-effect annotations) from deep learning annotations at SNPs implicated by S2G linking strategies -- generalizing the previously proposed ExPecto approach, which incorporates deep learning annotations based on distance to TSS. We evaluated these approaches using stratified LD score regression, using functional data in blood and focusing on 11 autoimmune diseases and blood-related traits (average N =306K). We determined that the three approaches produced SNP annotations that were uniquely informative for these diseases/traits, despite the fact that linear combinations of the underlying DeepSEA, Basenji, DeepBind and deltaSVM blood annotations were not uniquely informative for these diseases/traits. Our results highlight the benefits of integrating SNP annotations produced by deep learning models with other types of data, including data linking SNPs to genes.

genetics↗

bioRxiv2020

Unique contribution of enhancer-driven and master-regulator genes to autoimmune disease revealed using functionally informed SNP-to-gene linking strategies

Gene regulation is known to play a fundamental role in human disease, but mechanisms of regulation vary greatly across genes. Here, we explore the contributions to disease of two types of genes: genes whose regulation is driven by enhancer regions as opposed to promoter regions (enhancer-related) and genes that regulate other genes in trans (candidate master-regulator). We link these genes to SNPs using a comprehensive set of SNP-to-gene (S2G) strategies and apply stratified LD score regression to the resulting SNP annotations to draw three main conclusions about 11 autoimmune diseases and blood cell traits (average Ncase=13K across 6 autoimmune diseases, average N =443K across 5 blood cell traits). First, several characterizations of enhancer-related genes defined in blood using functional genomics data (e.g. ATAC-seq, RNA-seq, PC-HiC) are conditionally informative for autoimmune disease heritability, after conditioning on a broad set of regulatory annotations from the baseline-LD model. Second, candidate master-regulator genes defined using trans-eQTL in blood are also conditionally informative for autoimmune disease heritability. Third, integrating enhancer-related and candidate master-regulator gene sets with protein-protein interaction (PPI) network information magnified their disease signal. The resulting PPI-enhancer gene score produced >2x stronger conditional signal (maximum standardized SNP annotation effect size ({tau}*) = 2.0 (s.e. 0.3) vs. 0.91 (s.e. 0.21)), and >2x stronger gene-level enrichment for approved autoimmune disease drug targets (5.3x vs. 2.1x), as compared to the recently proposed Enhancer Domain Score (EDS). In each case, using functionally informed S2G strategies to link genes to SNPs that may regulate them produced much stronger disease signals (4.1x-13x larger{tau} * values) than conventional window-based S2G strategies. We conclude that our characterizations of enhancer-related and candidate master-regulator genes identify gene sets that are important for autoimmune disease, and that combining those gene sets with functionally informed S2G strategies enables us to identify SNP annotations in which disease heritability is concentrated.

genetics↗

bioRxiv2020

A convex optimization framework for gene-level tissue network estimation with missing data and its application in understanding disease architecture

Genes with correlated expression across individuals in multiple tissues are potentially informative for systemic genetic activity spanning these tissues. In this context, the tissue-level gene expression data across multiple subjects from the Genotype Tissue Expression (GTEx) Project is a valuable analytical resource. Unfortunately, the GTEx data is fraught with missing entries owing to subjects often contributing only a subset of tissues. In such a scenario, standard techniques of correlation matrix estimation with or without data imputation do not perform well. Here we propose Robocov, a novel convex optimization-based framework for robustly learning sparse covariance or inverse covariance matrices for missing data problems. Robocov produces more interpretable and less cluttered visual representation of correlation and causal structure in both simulation settings and GTEx data analysis. Simulation experiments also show that Robocov estimators have a lower false positive rate than competing approaches for missing data problems. Genes prioritized based on the average value of Robocov correlations or partial correlations across tissues are enriched for pathways related to systemic activities such as signaling pathways, heat stress factor, immune function and circadian clock. Furthermore, SNPs linked to these prioritized genes provide unique signal for blood-related traits; in comparison, no disease signal is observed for SNPs linked to genes prioritized by the standard correlation estimator. Robocov is an important stand-alone statistical tool for sparse correlation and causal network estimation for data with missing entries; and when applied to GTEx data, it provides insights into both genetic and autoimmune disease architectures.

genetics↗

bioRxiv2020

Improving the informativeness of Mendelian disease pathogenicity scores for common disease

Despite considerable progress on pathogenicity scores prioritizing both coding and noncoding variants for Mendelian disease, little is known about the utility of these pathogenicity scores for common disease. Here, we sought to assess the informativeness of Mendelian diseasederived pathogenicity scores for common disease, and to improve upon existing scores. We first applied stratified LD score regression to assess the informativeness of annotations defined by top variants from published Mendelian disease-derived pathogenicity scores across 41 independent common diseases and complex traits (average N = 320K). Several of the resulting annotations were informative for common disease, even after conditioning on a broad set of coding, conserved, regulatory and LD-related annotations from the baseline-LD model. We then improved upon the published pathogenicity scores by developing AnnotBoost, a gradient boosting-based framework to impute and denoise pathogenicity scores using functional annotations from the baseline-LD model. AnnotBoost substantially increased the informativeness for common disease of both previously uninformative and previously informative pathogenicity scores, implying pervasive variant-level overlap between Mendelian disease and common disease. The boosted scores also produced significant improvements in heritability model fit and in classifying disease-associated, fine-mapped SNPs. Our boosted scores have high potential to improve candidate gene discovery and fine-mapping for common disease.

genetics↗

bioRxiv2019

Evaluating the informativeness of deep learning annotations for human complex diseases

Deep learning models have shown great promise in predicting genome-wide regulatory effects from DNA sequence, but their informativeness for human complex diseases and traits is not fully understood. Here, we evaluate the disease informativeness of allelic-effect annotations (absolute value of the predicted difference between reference and variant alleles) constructed using two previously trained deep learning models, DeepSEA and Basenji. We apply stratified LD score regression (S-LDSC) to 41 independent diseases and complex traits (average N=320K) to evaluate each annotations informativeness for disease heritability conditional on a broad set of coding, conserved, regulatory and LD-related annotations from the baseline-LD model and other sources; as a secondary metric, we also evaluate the accuracy of models that incorporate deep learning annotations in predicting disease-associated or fine-mapped SNPs. We aggregated annotations across all tissues (resp. blood cell types or brain tissues) in meta-analyses across all 41 traits (resp. 11 blood-related traits or 8 brain-related traits). These allelic-effect annotations were highly enriched for disease heritability, but produced only limited conditionally significant results - only Basenji-H3K4me3 in meta-analyses across all 41 traits and brain-specific Basenji-H3K4me3 in meta-analyses across 8 brain-related traits. We conclude that deep learning models are yet to achieve their full potential to provide considerable amount of unique information for complex disease, and that the informativeness of deep learning models for disease beyond established functional annotations cannot be inferred from metrics based on their accuracy in predicting regulatory annotations.

genetics↗

bioRxiv2019

Characterizing and inferring quantitative cell cycle phase in single-cell RNA-seq data analysis

Cellular heterogeneity in gene expression is driven by cellular processes such as cell cycle and cell-type identity, and cellular environment such as spatial location. The cell cycle, in particular, is thought to be a key driver of cell-to-cell heterogeneity in gene expression, even in otherwise homogeneous cell populations. Recent advances in single-cell RNA-sequencing (scRNA-seq) facilitate detailed characterization of gene expression heterogeneity, and can thus shed new light on the processes driving heterogeneity. Here, we combined fluorescence imaging with scRNA-seq to measure cell cycle phase and gene expression levels in human induced pluripotent stem cells (iPSCs). Using these data, we developed a novel approach to characterize cell cycle progression. While standard methods assign cells to discrete cell cycle stages, our method goes beyond this, and quantifies cell cycle progression on a continuum. We found that, on average, scRNA-seq data from only five genes predicted a cells position on the cell cycle continuum to within 14% of the entire cycle, and that using more genes did not improve this accuracy. Our data and predictor of cell cycle phase can directly help future studies to account for cell-cycle-related heterogeneity in iPSCs. Our results and methods also provide a foundation for future work to characterize the effects of the cell cycle on expression heterogeneity in other cell types.

genomics↗

bioRxiv2017

A new sequence logo plot to highlight enrichment and depletion

BackgroundSequence logo plots have become a standard graphical tool for visualizing sequence motifs in DNA, RNA or protein sequences. However standard logo plots primarily highlight enrichment of symbols, and may fail to highlight interesting depletions. Current alternatives that try to highlight depletion often produce visually cluttered logos.\n\nResultsWe introduce a new sequence logo plot, the EDLogo plot, that highlights both enrichment and depletion, while minimizing visual clutter. We provide an easy-to-use and highly customizable R package Logolas to produce a range of logo plots, including EDLogo plots. This software also allows elements in the logo plot to be strings of characters, rather than a single character, extending the range of applications beyond the usual DNA, RNA or protein sequences. We illustrate our methods and software on applications to transcription factor binding site motifs, protein sequence alignments and cancer mutation signature profiles.\n\nConclusionOur new EDLogo plots, and flexible software implementation, can help data analysts visualize both enrichment and depletion of characters (DNA sequence bases, amino acids, etc) across a wide range of applications.

genetics↗

bioRxiv2016

Metabolome Identification by Systematic Stable Isotope Labeling Experiments and False Discovery Analysis with a Target-Decoy Strategy

We introduce a formula-based strategy and algorithm (JUMPm) for global metabolite identification and false discovery analysis in untargeted mass spectrometry-based metabolomics. JUMPm determines the chemical formulas of metabolites from unlabeled and stable-isotope labeled metabolome data, and derives the most likely metabolite identity by searching structure databases. JUMPm also estimates the false discovery rate (FDR) with a target-decoy strategy based on the octet rule of chemistry. With systematic stable isotope labeling of yeast, we identified 2,085 chemical formulas (10% FDR), 892 of which were assigned with metabolite structures. We evaluated JUMPm with a library of synthetic standards, and found that 96% of the formulas were correctly identified. We extended the method to mammalian cells with direct isotope labeling and by heavy yeast spike-in. This strategy and algorithm provide a powerful a practical solution for global identification of metabolites with a critical measure of confidence.

biochemistry↗