Chromatin Accessibility Beyond the Peaks

Abstract

Chromatin state resides at the intersection of trans-acting factors that operate globally over the genome but respond to changing conditions, and DNA sequence which is invariant across conditions but varies locally around genes. Assays that measure features of chromatin state can help us understand this interplay between sequence and trans factors that ultimately governs transcriptional regulation. My dissertation work has been focused on the analysis of a long read chromatin accessibility assay called Fiber-seq. In this body of work I show that this single assay that, on its face, measures chromatin accessibility, is incredibly information rich and may help decode the regulatory logic governing gene expression. I present fiber-views, a software package for analysis of Fiber-seq data at aligned genomic positions. In the application of Fiber- seq to Zea mays I show Fiber-seq detects twice as many ACRs as ATAC-seq in paired samples. Fiber-seq is particularly good at identifying ACRs with short accessible elements and ACRs in repetitive regions, including transposable elements. Finally I present a novel analysis approach converting the single molecule data from Fiber-seq to a set of feature tracks. I show that these feature tracks are able to recapitulate chromatin states typically derived from multiple ChIP-seq assays. Fiber-seq derived features can predict gene expression, capturing nearly 60% of expression variation in maize. Patterns of Fiber-seq features can also be used to categorize ACRs reflecting their function and underlying sequence.