Synthetic data to explore transcriptional regulation of differentially expressed genes in ovarian cancer

Abstract

In major diseases like cancer, many genes are dysregulated. While differential expression (DE) analyses help identify these genes, they do not reveal the transcriptional regulatory landscape between control and disease states. Here, we developed a process-driven model based on generalized stochastic transcriptional regulation (TR) to study DE genes in ovarian cancer. We generated over 39,000 synthetic gene expression profiles based on RNA-seq dataset of 11 samples (5 control FTE and 6 ovarian cancer), and fitted the model parameters to real data achieving between 82 and 99% transcriptome-wide similarity through various data analytics. Through the model we investigated the top 100 DE genes’ transcriptional regulation. Notably, GSTA3 , SNTN , DNAI2 and KIF19 , which have been associated with cancer, possess significantly larger transcriptional quantal and RNA degradation rates, leading to their greater variability in cancer. Overall, this mechanistic understanding provides deep functional insights into unchartered transcriptome-wide gene regulation and could potentially guide gene-targeted interventions for cancer or other diseases in the future.