MOCCS profile analysis clarifies the cell type dependency of transcription factor-binding sequences and cis-regulatory SNPs in humans

Abstract

Transcription factors (TFs) show heterogeneous DNA-binding specificities in individual cells and whole organisms in natural conditions): de novo motif discovery usually provides multiple motifs even from a single ChIP-seq sample. Despite the accumulation of ChIP-seq data and ChIP-seq-derived motifs, the diversity of DNA-binding specificities across different TFs and cell types remains largely unexplored. Here, we propose MOCCS profiles, the new representation of DNA-binding specificity of TFs, which describes a ChIP-seq sample as a profile of TF-binding specificity scores (MOCCS2scores) for every k-mer sequence. Using our k-mer-based motif discovery method MOCCS2, we systematically computed MOCCS profiles for >10,000 human TF ChIP-seq samples across diverse TFs and cell types. Comparison of MOCCS profiles revealed the global distributions of DNA-binding specificities, and found that one-third of the analyzed TFs showed differences in DNA-binding specificities across cell types. Moreover, we showed that the differences in MOCCS2scores (ΔMOCCS2scores) predicted the effect of variants on TF binding, validated by in vitro and in vivo assay datasets. We also demonstrate ΔMOCCS2scores can be used to interpret non-coding GWAS-SNPs as TF-affecting SNPs and provide their candidate responsible TFs and cell types. Our study provides the basis for investigating gene expression regulation and non-coding disease-associated variants in humans.