Statistical Detection and Biochemical Classification of Cancer Driver Mutation Patterns in Biological Networks

点击翻译按钮获取中文摘要

 DESCRIPTION (provided by applicant): Cancer arises from somatically acquired genetic and epigenetic alterations. While large consortia like The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC) have profiled genomic somatic mutations of thousands of tumor samples from various cancer types based on whole-genome/exome sequencing, meaningful mechanistic interpretation of these gene variation results are still very limited. One basic yet challenging task is to distinguish driver mutations, which are causally implicated in cancer development, from passenger mutations, which occur randomly with neutral effect. Another critical task is to map, trace, and interpret the functional impact of drivr mutations within biological networks. In a network context, driver mutations associated with genes within a pathway often show a mutually exclusive pattern, meaning that each patient carries exactly one mutation in the pathway, which is sufficient to perturb the function of that pathway. Another prominent pattern is that driver mutations of genes from several different pathways may co-occur, since perturbation of multiple pathways is required for tumor formation. Screening for mutual exclusivity and co-occurrence patterns can greatly facilitate the identification of novel sets of related driver gene mutations, their associated driver pathways, and functional relationships between these driver pathways. Although several de novo driver mutation gene set discovery methods have been proposed in the past few years, they have major limitations due to computational feasibility, an inability to deal with mutational heterogeneity across patients, and lack of biochemical interpretation. The overall goal of this proposal is to develop and combine advance sequence variation analyses with complementary biological network analyses into a highly novel systems biology approach that will: i) detect sets of related mutations in driver regulatory/signaling pathways, ii) classify these pathways as stimulated, inhibited, or mixed with respect to their role in the tumor development process, and iii) predict direct metabolic outcomes of these perturbed pathways. Our specific aims are: 1) to develop a statistical method for de novo discovery of mutually exclusive and co-occurrent sets of driver mutations; and 2) to develop a pathway mapping and classification method for related sets of driver mutations. The identification and biochemical interpretation of aggregated tumor mutations from driver mutation gene sets to inhibited/stimulated pathways to perturbed biological network will provide new mechanistic insights in tumor progression at a systems level. Also with this information, potential drug targets in the detected driver pathways can be classified as requiring agonist or antagonist drug development, making drug target evaluation and prioritization much more effective. Furthermore, identification of co-occurrence between specific genes and pathways may aid in the development of multi- therapeutic cancer treatments that are optimized to groups of patients showing the same mutational patterns of co-occurrence.