SRA

Enhancers and promoters commonly occur in accessible chromatin characterized by depleted nucleosome contact; however, it is unclear how chromatin accessibility is governed. We show that a logic of cis-acting DNA sequence features can predict the majority of chromatin accessibility at high spatial resolution. We develop a new type of high-dimensional machine learning model, the Cooperative Chromatin Model (CCM), that is capable of predicting a large fraction of genome-widepromoters chromatin accessibility at basepair-resolution in a range of human and mouse cell types from DNA sequence alone. We confirm that a CCM accurately predicts chromatin accessibility, even of a vast array of synthetic DNA sequences, with a novel CrispR-based method of highly efficient site-specific DNA library integration. CCMs are directly interpretable and reveal that a logic based on local, non-specific cooperation, largely among pioneer TFs, is sufficient to predict a large fraction of cellular chromatin accessibility in a wide variety of cell types. Overall design: Dnase-seq on human and mouse cells as well as massively parallel report assay (MPRA) validation using CRISPR editing of native genomic loci.