Combined burden and functional impact tests for cancer driver discovery using DriverPower

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Cancer genome sequencing aims to discover driver mutations.
• The ICGC/TCGA Pan-Cancer Analysis of Whole Genomes (PCAWG) Consortium analyzed 2658 cancer genomes.
• Identifying driver mutations in both coding and non-coding regions is crucial.

Purpose of the Study:

• To introduce DriverPower, a novel software package for identifying driver mutations.
• To leverage mutational burden and functional impact evidence for enhanced driver discovery.
• To analyze whole genome sequencing data from the PCAWG project.

Main Methods:

• DriverPower utilizes 1373 genomic features to model background mutation rates.
• The software incorporates functional impact scores to improve driver identification accuracy.
• A background mutation model explains up to 93% of regional mutation rate variance.

Main Results:

• DriverPower identified 217 coding and 95 non-coding driver candidates in 2583 PCAWG cancer genomes.
• The software demonstrated superior performance compared to six other methods.
• DriverPower achieved the highest F1 score for both coding and non-coding driver discovery.

Conclusions:

• DriverPower is an effective computational framework for identifying cancer driver mutations.
• The software enhances driver discovery by integrating mutational burden and functional impact.
• This approach improves the accuracy of identifying both coding and non-coding drivers.