The chromVAR package takes as inputs 1) aligned sequencing reads, 2) chromatin accessibility peaks (derived from either data aggregated across cells or external resources), and 3) a set of chromatin features representing either motif position weight matrices (PWMs) or genomic annotations (Fig. 1a, Supplementary Fig. 1). For use as input (3) into chromVAR, we have curated a set of human and mouse PWMs from the cisBP database10 that represent a diverse and comprehensive collection of known TF motifs. Alternately, user provided TF motifs or other types of genomic annotations, such as enhancer modules, ChIP-seq peaks, or GWAS disease annotation may be used. chromVAR may also be applied to a collection of kmers—DNA sequences of a specific length k—in order to perform an unbiased analysis of DNA sequence features that correlate with chromatin accessibility variation across the cells or samples.
