IDENTIFYING CONCORDANCE BETWEEN OPEN CHROMATIN AND GENE EXPRESSION

Abstract

Characterizing the impact of changes in chromatin accessibility on gene expression is essential to both obtaining an understanding of how gene mis-regulation leads to disease and predicting the effects of non-coding mutations. We continue the work presented in Marstrand & Storey (2014) identifying genes showing cell-specific concordance between expression and chromatin accessibility among 20 cell lines using the angle ratio statistic (ARS), a novel statistic developed to detect concordant outliers between two datasets. This paper extends the analysis to 126 cell lines and reworks the generation of null statistics for the ARS method. Our work proposes several extensions to the ARS method which use redundant cell lines to increase the power to detect associations in these groups. We find that enriched GO terms for significant genes called by ARS match cell line function and demonstrate that correlated ARS lines correspond to biological groupings. Finally, we identify local regions which significantly contribute to expression and chromatin accessibility concordance and report transcription factor binding sites that are over and under represented in these sequences. The ARS statistical framework can be extended to multiple dimensions to incorporate additional genome-wide measurements of chromatin accessibility, applied to alternative splicing instead of expression, or used to evaluate potential enhancers.