Computational Modeling of Gene Regulation

人类基因组包含超过22,000个基因，这与单细胞芽殖酵母或果蝇相比并不多。那么是什么让人类变得如此复杂呢？答案在于人类基因被调节或开启和关闭的极其复杂和微妙的方式。这些规则中的小故障往往会导致蛋白质产品在错误的数量、错误的时间或错误的地点被制造出来，这反过来又会导致癌症等疾病。为了更好地了解人类细胞的工作和功能，了解22,000个基因中的每一个是如何被调节的，以及如何打开和关闭一组基因来完成所需的细胞任务是至关重要的。

The human genome contains over 22,000 genes, which remarkably is not much more than the single cell budding yeast or a fruit fly. So what makes humans so much more complex? The answer lies in the immensely complex and delicate ways by which the human genes are regulated or turned on and off. Glitches in these regulations often result in protein products being made at a wrong amount or at wrong time or wrong place, which in turn can cause diseases such as cancer. In order to better understand how human cells work and function, it is crucial to understand how each of the 22,000 genes is regulated and how groups of genes are turned on and off to fulfill the required cellular task. Supported by a previous Discovery Grant, my students and I have made significant progress in developing computing methods in studying gene regulation and genetic variations. For example, we developed a suite of software tools that can accurately predict microRNA target genes and identify miRNA regulations that are important in human diseases. We also developed algorithms that can integrate transcription factors (TFs), microRNAs and other regulation elements into a comprehensive regulatory network, and computational tools for analyzing such network. In the next funded period, we will further improve and extend these algorithms so that they can run efficiently and quickly on a genomic scale. Our software tool will be able to quickly scan genome sequences and identify mutations that disrupt regulatory elements. These tools will be able to integrate different type of regulatory elements, such as microRNAs, transcription factors (TF), RNA binding proteins (RBP), RNA structure, and DNA or RNA methylation, into one comprehensive network model. Interestingly many of these regulatory mechanisms can cross-talk to each other. For example, microRNAs and RBPs can compete for the same target position in the 3'UTR, and methylation in these target sites can mask the binding by microRNAs or RBPs. Our integrated regulatory network that will help researchers to quickly and accurately identify and prioritize mutations or mutated regulations. With the rapid advancement of next generation sequencing technology and reduction of sequencing cost, we envision that tens of thousands or even millions of human genome sequence will become available in the next years. Fast, efficient, and accurate computational tools will be needed to analyze these genomes and identify actionable mutations. The methodology that we are developing will help to fill these needs.