A Novel Approach for Identifying Cancer Drivers in Coding andNon-coding Genomic Elements

Cataloging cancer drivers is crucial in cancer research. Current methods for identifying cancer drivers,especially in non-coding regions of the genome, mainly rely on detecting signals of positive selec tion using two approaches: mutational burden tests, which compare observed and background mutationrates, and functional impact tests, which incorporate functional relevance of mutations in a genomicregion. However, previous studies have had limited success in identifying potential drivers in non coding regions. This paper introduces a new statistical test combining functional impact and muta tional burden tests to better prioritize driver candidates in both coding and non-coding elements.Methods:This study used somatic PCAWG consensus callsets of SNV/Indels of ۲۲۵۳ non-hyper mutated samples across۳۳ cancer types from the PCAWG project. The somatic variants were annotated using CADD scores. Thebackground mutation rates were estimated using the DriverPower gradient boosting model based on a fea ture matrix of ۱۳۷۳ genomic and epigenomic features influencing mutation rates. This calculation took intoaccount the genomic coordinates of both coding and non-coding elements and randomly generated bins, excludingblacklisted regions of the genome. The study then employed a novel statistical test based on probabilis tic graphical models that incorporate both mutation recurrence and CADD score to prioritize cancer driverelements. Finally, the p-values were corrected for multiple testing with the Benjamini-Hochberg procedure.Results and Discussion:The performance of this approach was compared to other PCAWG driver discovery methodsnamely ActiveDriverWGS, DriverPower, NBR, oncodriveFML, MutSig, ncdDetect, and ExInAtor interms of AUPR and AUC measures. The study resulted in AUC (۰.۶۷۹۷۵۱, ۰.۵۴۷۰۷۷۶) and AUPR(۰.۱۹۵۰۳۵, ۰.۰۸۰۸۲۹) and outperformed other computational methods applied to the PCAWG data.

کلیدواژه ها:

Cancer driver ، non-coding genomic elements ، Somatic mutations ، Mutational burden tests ، Functional impacttests