Michael J. Purcaro

The paper updates Factorbook, a TF-centric database, by expanding its motif catalog and integrating it with ENCODE data to annotate candidate regulatory sites and provide tools for variant and disease analysis.

The paper annotates 18 chromatin states in 66 mouse epigenomes across twelve tissues during development using ChromHMM, identifying bivalent states with evolutionary conservation and gene silencing functions.

The paper benchmarks enhancer-gene prediction methods using a curated dataset, finding TargetFinder performs best but has limited cross-cell-type effectiveness.

The paper develops LR-DNase, a logistic regression model that predicts TF binding sites using DNase-seq and genomic sequence features, outperforming existing methods.