IMPACT: Genomic Annotation of Cell-State-Specific Regulatory Elements Inferred from the Epigenome of Bound Transcription Factors

Abstract

Despite significant progress in annotating the genome with experimental methods, much of the regulatory noncoding genome remains poorly defined. Here we assert that regulatory elements may be characterized by leveraging local epigenomic signatures where specific transcription factors (TFs) are bound. To link these two features, we introduce IMPACT, a genome annotation strategy that identifies regulatory elements defined by cell-state-specific TF binding profiles, learned from 515 chromatin and sequence annotations. We validate IMPACT using multiple compelling applications. First, IMPACT distinguishes between bound and unbound TF motif sites with high accuracy (average AUPRC 0.81, SE 0.07; across 8 tested TFs) and outperforms state-of-the-art TF binding prediction methods, MocapG, MocapS, and Virtual ChIP-seq. Second, in eight tested cell types, RNA polymerase II IMPACT annotations capture more cis -eQTL variation than sequence-based annotations, such as promoters and TSS windows (25% average increase in enrichment). Third, integration with rheumatoid arthritis (RA) summary statistics from European (N = 38,242) and East Asian (N = 22,515) populations revealed that the top 5% of CD4 + Treg IMPACT regulatory elements capture 85.7% of RA h2, the most comprehensive explanation for RA h2 to date. In comparison, the average RA h2 captured by compared CD4 + T histone marks is 42.3% and by CD4 + T specifically expressed gene sets is 36.4%. Lastly, we find that IMPACT may be used in many different cell types to identify complex trait associated regulatory elements.

Immune disease variants modulate gene expression in regulatory CD4+ T cells and inform drug targets

“…Collectively, we tested 1,290 unique GWAS loci associated with 14 immune colocalizing signals) included IBD, UC, CD, ALL and RA. The high number of observed colocalizations is consistent with previous work that implicated the role of Tregs in the pathobiology of all of these diseases 5,6,[32][33][34] , as well as the higher number of significant GWAS loci for these traits. Overall, immune-mediated diseases showed more colocalizations with Treg QTLs than non-immune mediated diseases, such as type-2 diabetes or depression ( Figure 3A ).…”

“…GWAS variants associated with common immune-mediated diseases such as inflammatory bowel disease (IBD), type 1 diabetes (T1D) and rheumatoid arthritis (RA) are enriched in active chromatin marks that tag enhancers and promoters in the CD4+ T cell compartment, especially in regulatory T cells (Tregs) [4][5][6] . Tregs are an infrequent yet functionally significant subset of CD4+ T cells, they comprise 2-10% of CD4+ T cells and play an essential homeostatic role in the immune system by suppressing the proliferation and effector functions of conventional T cells.…”

Deep neural networks identify context-specific determinants of transcription factor binding affinity

“…AgentBind automatically captures features determining binding status based on DNA sequence alone. We compared our results with the recently developed IMPACT method 30 , which tackles a similar classification task but instead using a broad range of manually extracted epigenomic features including histone modifications, TF binding, ATAC-seq, and DNAse-Seq profiles. We benchmarked each method on 4 TFs active in CD4+ T cells and applied the same training scheme as was used in the IMPACT study.…”

echolocatoR: an automated end-to-end statistical and functional genomic fine-mapping pipeline

“…We also built in API access to in silico validation datasets, including massively parallel reporter assays (MPRA) (van Arensbergen et al, 2019;Tewhey et al, 2018), S-LDSC heritability enrichment, and predictions from multiple machine learning models trained on tissue-and cell-type-specific epigenomic annotations: Basenji (Kelley et al, 2018), and DeepSEA (Zhou and Troyanskaya, 2015) (provided by (Dey et al)) as well as IMPACT (Amariuta et al, 2019). Lastly, we integrated motifbreakR which uses a comprehensive set of algorithms and position weight matrices (n = 9,933) to assess whether fine-mapped variants fall within sequence motifs and to what extent they disrupt binding to specific transcription factors (Coetzee et al, 2015).…”