Inferring differential protein binding from time-series chromatin accessibility data

Due to internal and external factors, the epigenomic landscape is constantly changing in ways that are linked to changes in gene expression. Chromatin accessibility data, such as MNase-seq, provide valuable insights into this landscape and have been used to compute chromatin occupancy profiles. Multiple datasets generated over time or under different conditions can thus be used to study dynamic changes in chromatin occupancy across the genome.

Our existing model, RoboCOP, computes a genome-wide chromatin occupancy profile for nucleosomes and hundreds of transcription factors. Here, we present a new method called DynaCOP that takes multiple chromatin occupancy profiles and uses them to generate a series of nucleosome-guided difference profiles. These profiles identify differentially binding transcription factors and reveal changes in nucleosome occupancy and positioning. We apply DynaCOP to chromatin occupancy profiles derived from deeply sequenced time-series MNase-seq data to study differential chromatin occupancy in the yeast genome under cadmium stress. We find strong correlations between the observed chromatin changes and changes in transcription.

https://github.com/HarteminkLab/RoboCOP.