作者: Gregory P Way , Francisco Sanchez-Vega , Konnor La , Joshua Armenia , Walid K Chatila
DOI: 10.1016/J.CELREP.2018.03.046
关键词: Genome 、 Machine learning 、 Selumetinib 、 Transcriptome 、 Regulation of gene expression 、 KRAS 、 HRAS 、 Biology 、 Gene 、 Neuroblastoma RAS viral oncogene homolog 、 Artificial intelligence
摘要: Precision oncology uses genomic evidence to match patients with treatment but often fails identify all who may respond. The transcriptome of these "hidden responders" reveal responsive molecular states. We describe and evaluate a machine-learning approach classify aberrant pathway activity in tumors, which aid hidden responder identification. algorithm integrates RNA-seq, copy number, mutations from 33 different cancer types across Cancer Genome Atlas (TCGA) PanCanAtlas project predict states tumors. Applied the Ras pathway, method detects activation identifies phenocopying variants. model, trained on human can response MEK inhibitors wild-type cell lines. also present data that suggest multiple hits confer increased activity. The is underused precision and, combined machine learning, identification responders.