Researchers from the Medical College of Wisconsin, China's University of Science and Technology, and Zhejiang University have developed cancer sequence analysis software that they claim improves on existing methods for identifying driver mutations — genetic changes that promote tumor survival and growth.
The software, called Driver Genes and Pathways or DrGap, is described in detail in a paper that was published earlier this month in the American Journal of Human Genetics. According to its developers, DrGap provides "a powerful and flexible statistical framework for identifying [cancer] driver genes and pathways" that offers "several significant features not addressed or naïvely obtained by previous methods."
The AJHGpaper explains that DrGap identifies both driver mutations and pathways using a combination of statistical methods such as Poisson modeling and maximum likelihood estimation, and a set of bioinformatics applications for doing things like estimating protein coding regions' depth of coverage and tabulating mutations.
Unlike existing methods, DrGap's statistical framework allows it to incorporate biological data such as protein coding region length, variations in transcript isoforms and mutation types, and differences in background mutation rates, and use this information to "assess the functional significance of mutated genes," the researchers wrote. It also includes methods to analyze pathways of mutated driver genes, which makes it possible to detect "infrequently mutated cancer-associated genes … with sufficient power," the paper states.
Other DrGap features include its ability to consider multiple types of mutations to both reduce the risk of bias in estimating mutation rates and increase the "statistical power compared with a single-type mutation test," as well as statistical improvements that make it possible for researchers to analyze data from tumors with a low prevalence of somatic mutations. It also adjusts protein coding region sizes depending on the sequence coverage to increase sensitivity, and uses a Bayesian approach to estimate the distribution of background mutation rates, the paper states.
These improvements, the developers claim, make DrGap a more accurate and sensitive tool for cancer genome sequencing studies — one that provides a more comprehensive picture of the tumor mutation landscape that could help researchers identify more effective therapeutic targets for treatments, Pengyuan Liu, an associate professor of physiology at MCW and a co-author on the paper, told BioInform.
Liu and colleagues reported that in tests conducted with the software, it not only recaptured a large majority of driver genes previously reported in other studies, but also identified a much longer list of additional candidate genes whose mutations may be linked to cancer.
For example, in one test, described in the AJHG paper that used 623 genes from a Cancer Genome Atlas dataset to compare DrGap's performance with pipelines from two other studies, the researchers report that DrGap was able to identify 59 driver genes compared to 22 and 23 driver genes that were identified by the methods used in the other studies.
In another test using whole-exome sequencing data from non-small cell lung cancer samples and squamous cell cancer samples, DrGap identified 110 and 260 driver gene mutations out of more than 7,700 and 11,000 mutated genes that were identified in the lung cancer and squamous cell cancer samples, respectively, the researchers wrote. In a third test which analyzed colorectal cancer data from TCGA, DrGap reportedly identified 44 driver mutations — 15 of which had been previously identified by TCGA — and 29 additional ones not included in the initial TCGA report.