SMuRF: Portable and accurate ensemble-based somatic variant calling

Abstract

ABSTARCTO_ST_ABSSummaryC_ST_ABSSMuRF is an ensemble method for prediction of somatic point mutations (SNVs) and small insertions/deletions (indels) in cancer genomes. The method integrates predictions and auxiliary features from different somatic mutation callers using a Random Forest machine learning approach. SMuRF is trained on community-curated tumor whole genome sequencing data, is robust across cancer types, and achieves improved accuracy for both SNV and indel predictions of genome and exome-level data. The software is user-friendly and portable by design, operating as an add-on to the community-developed bcbio-nextgen somatic variant calling pipeline.\n\nContactskanderupamj@gis.a-star.edu.sg