BIOINFORMATICS ANALYSIS AND COMPUTATIONAL BIOLOGY MODELING OF GENETIC AND EPIGENETIC ELEMENTS DICTATING CHROMATIN ORGANIZATION, TRANSCRIPTION REGULATION, AND DISEASE STATE IN CANCER GENOMES

Abstract

Cancer is the second leading cause of death worldwide with over 17 million new cases and over 9.6 million deaths annually. Hence, it is one of the biggest research topics of our time. In all our projects, we implemented powerful bioinformatics and computational biology concepts, ideas, and tools to address cancer biology problems. 1) We developed a novel immune signaling-based subtyping approach to stratify head and neck squamous cell carcinoma patients into clinically relevant subtypes with different immune signatures. 2) For the first time in eQTL analysis, we sub-grouped genomically complex lung squamous cell carcinoma patients based on SOX2 activity prior to conducting eQTL analysis. This allowed us to truly identify gene targets that are affected by the overexpression of the stem cell transcription factor. 3) We developed a machine learning model that modeled chromatin organization using a vast array of genetic and epigenetic features. We identified underlying factors that contribute to chromatin structure changes in breast cancer. We also analyzed downstream effects of chromatin structure alterations in breast cancer cells and discovered plausible activation of signaling pathways and oncogenes. 4) Finally, we developed two “sequence-based” methods to study enhancer-promoter interactions. Both methods extracted meaningful information and patterns from DNA sequences and accurately predicted enhancer-promoter interactions. Our projects enhanced our understanding on our genomes and allowed us to predict functional importance of mutations, SNPs, or indels in patient data. Our developed models and pipelines tackled cancer biology problems and can easily be applied to other problems and diseases.