Deep learning sequence-based ab initio prediction of variant effects on expression and disease risk

Abstract

Key challenges for human genetics, precision medicine and evolutionary biology include deciphering the regulatory code of gene expression and understanding the transcriptional effects of genome variation. However, this is extremely difficult because of the enormous scale of the noncoding mutation space. We developed a deep learning-based framework, ExPecto, that can accurately predict, ab initio from a DNA sequence, the tissue-specific transcriptional effects of mutations, including those that are rare or that have not been observed. We prioritized causal variants within disease- or trait-associated loci from all publicly available genome-wide association studies and experimentally validated predictions for four immune-related diseases. By exploiting the scalability of ExPecto, we characterized the regulatory mutation space for human RNA polymerase II-transcribed genes by in silico saturation mutagenesis and profiled > 140 million promoter-proximal mutations. This enables probing of evolutionary constraints on gene expression and ab initio prediction of mutation disease effects, making ExPecto an end-to-end computational framework for the in silico prediction of expression and disease risk.

Genetic Profiles in Pharmacogenes Indicate Personalized Drug Therapy for COVID-19

“…However, as we indicated in Figure 3, most of nonsynonymous mutations are rare variations. Therefore, a reliable way to predict potential pharmacogenetic variation is also important to improve the efficacy or avoid serious toxicity [23] . The third challenge is that a suitable strategy for precision medicine in COVID-19 treatment is still lack.…”

Evaluating the informativeness of deep learning annotations for human complex diseases

“…Disease risk variants identified by genome-wide association studies (GWAS) lie predominantly in non-coding regions of the genome 1,2,3,4,5,6,7 , motivating broad efforts to generate genome-wide maps of regulatory marks across tissues and cell types 8,9,10,11 . Recently, deep learning models trained using these genome-wide maps have shown considerable promise in predicting regulatory marks directly from DNA sequence 12,13,14,15,16,17,18 . In particular, these studies showed that variant-level deep learning annotations (based on the reference allele) attained high accuracy in predicting the underlying chromatin marks 13,14,15,16 , and that models incorporating allelic-effect deep learning annotations (absolute value of the predicted difference between reference and variant alleles) attained high accuracy in predicting disease-associated SNPs 13,14,15,16 ; additional applications of deep learning models, including analyses of signed allelic-effect annotations, are discussed in the Discussion section.…”

“…Recently, deep learning models trained using these genome-wide maps have shown considerable promise in predicting regulatory marks directly from DNA sequence 12,13,14,15,16,17,18 . In particular, these studies showed that variant-level deep learning annotations (based on the reference allele) attained high accuracy in predicting the underlying chromatin marks 13,14,15,16 , and that models incorporating allelic-effect deep learning annotations (absolute value of the predicted difference between reference and variant alleles) attained high accuracy in predicting disease-associated SNPs 13,14,15,16 ; additional applications of deep learning models, including analyses of signed allelic-effect annotations, are discussed in the Discussion section. However, it is unclear whether deep learning annotations at commonly varying SNPs contain unique information for complex disease that is not present in other annotations.…”

“…Here, we evaluate the informativeness for complex disease of both variant-level and allelic-effect annotations at commonly varying SNPs constructed using two deep learning models previously trained on tissue-specific regulatory features (DeepSEA 13,15 and Basenji 16 ); we also train a new deep learning model to prioritize SNPs within non-tissue-specific features (BiClassCNN). We apply stratified LD score regression 5,19 (S-LDSC) to 41 independent diseases and complex traits (average N =320K) to evaluate each annotation's informativeness for disease heritability conditional on a broad set of coding, conserved, regulatory and LD-related annotations from the baseline-LD model 19 and other sources (imputed Roadmap and ChromHMM annotations 11,20,21,22 ); as a secondary metric, we also evaluate the accuracy of models that incorporate deep learning annotations in predicting disease-associated or fine-mapped SNPs 23,24 .…”

Characterising the genetic basis of immune response variation to identify causal mechanisms underlying disease susceptibility

“…86 Massively parallel assays have also been used in the context of both splicing 88,89 and miRNA-mediated regulation, 90,91 showing, for instance, that up to 16% of splice disrupting variants are located in deep intronic regions. 88 With the development of deep learning frameworks for sequence-based predictions, 92,93 data generated by these assays, will now fuel the construction of predictive models. These models will, in turn, allow quantifying the regulatory impact of novel genetic variants on transcriptional activity, 86 alternative splicing, 94 or miRNA-mediated regulation.…”