A Random Forest Model for Detecting Somatic Mutations from Unpaired Cancer Samples

Authors

• S Tu
• B Zhou
• DW Huang
• R Schmitz
• GW Wright
• LM Staudt
• CA Johnson

Abstract

In whole exome sequencing studies on human cancer samples without matched normal samples, most SNV callers contaminate the pool of somatic mutation calls with germline calls due to several complications in the cancer genome. We have trained a random forest model on MuTect2 scores from 37 matched-normal diffuse large B-cell lymphoma samples from TCGA. The model’s ROC vastly outperforms that of the toxicity callers that comprise its feature space.

Scientific Focus Area: Computational Biology

This page was last updated on Friday, March 26, 2021