PARALLEL FACTOR CHIP SPIKE-IN DECONVOLUTION
Open Access
- Author:
- Wekhande, Siddarth
- Graduate Program:
- Bioinformatics and Genomics
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- November 07, 2019
- Committee Members:
- Shaun Mahony, Thesis Advisor/Co-Advisor
Istvan Albert, Committee Member
Qunhua Li, Committee Member
George H Perry, Program Head/Chair - Keywords:
- ChIP-seq spike in deconvolution
bioinformatics
ChIP-seq analysis - Abstract:
- Recent advances in chromatin immunoprecipitation protocol combined with high-throughput sequencing (ChIP-seq) have enabled the creation of more robust normalization strategies than those previously utilized in the literature. Parallel factor ChIP-seq is a recently developed protocol that is based on the addition of a constant amount of second antibody which serves as an internal control (spike-in) in each sample. Doing so allows us to quantify any relative changes between the global occupancy of the transcription factor of interest to the spike-in signal, while controlling for differences in ChIP-efficiency. However, the computational pipeline developed for the analysis of parallel factor ChIP-seq depends on sharp, non-overlapping signals, which is not suitable when one is interested in studying broad signals, such as those seen for histone modifications. In this thesis, we propose a novel computational workflow that takes advantage of the parallel factor ChIP-seq protocol to perform absolute comparisons between samples. We first deconvolve the spike-in signals in a peak-distribution shaped manner, and use the total deconvolved signal as a normalization factor. This allows us to scale the target broad signals of a protein of interest to the same scale for absolute comparison. We apply this computational workflow on a set of simulated parallel factor ChIP experiments to show successful deconvolution and scaling factor estimation, and on a set of parallel factor ChIP-seq experiments designed to study global changes in H3 Lys 27 trimethylation (H3K27me3) levels in a neuronal cell differentiation time series. In particular, we are interested in quantifying the histone modification domain changes over the Hox gene clusters, a region of interest when studying transcriptional dynamics during early neural differentiation. We show that we are able to successfully deconvolve spike-in signals from the samples, apply a scaling factor for absolute comparison, and observe an accumulation of H3K27me3 repressive mark over the Hox gene clusters during the neural differentiation time course.