Detection of Transcription Factor Co-Binding Patterns in Human Cells via Point Process Models
Transcription factors usually work synergistically to regulate target gene expression. Characterizing their combinatorial binding patterns will provide key step towards elucidating the underlying gene regulatory mechanism. Accordingly, the goal of this thesis is to apply a newly designed test statistic based on inhomogeneous Poisson process and Ripley's K-function to investigate pairwise transcription factor binding patterns using chromatin immunoprecipitation sequencing (ChIP-seq) data. We applied the method to 21 selected transcription factors using ChIP-seq data from two different human cell types. Significant clustering patterns have been detected between most transcription factor pairs, and their optimal binding distances are reported. More interestingly, by comparing the two cell types, we identify tissue-specific co-binding patterns, which implicate tissue-specific transcriptional regulation. In summary, the presented work has demonstrated the development and utility of the designed test statistics on evaluating transcription factor binding patterns, which can help to infer transcription factor interactions on gene expression regulation.