Tumor-specific Causal Inference Reveals Functional Impacts of Somatic Genome Alterations in Individual Tumors
Chunhui Cai1,2Kevin N.Lu1,2Gregory F.Cooper1,2Shuping Xu3Zhenlong Zhao3Xueer Chen1,2Adrian V.Lee2,3,4,5Nathan Clark2,6Vicky Chen1,2Songjian Lu1,2Lujia Chen1,2Alexandros Labrinidis2,7Harry S Hochheiser1,2Uma Chandran1,2Xia Jiang1,2Q.Jane Wang3Xinghua Lu1,2,5
1. Department of Biomedical Informatics,University of Pittsburgh2. Center for Causal Discovery,University of Pittsburgh3. Department of Pharmacology and Chemical Biology,University of Pittsburgh4. Magee Women's Cancer Research Center,University of Pittsburgh5. University of Pittsburgh Cancer Institute,University of Pittsburgh6. Department of Computational Biology and Systems Biology,University of Pittsburgh7. Department of Computer Science,University of Pittsburgh
摘要：Identifying causative somatic genome alterations（SGAs） driving an individual tumor could both provide insight into disease mechanisms and support personalized modeling for precision oncology.Here,we present a Tumor-specific Causal Inference（TCI） framework that infers causal relationships between SGAs and molecular phenotypes（e.g.,transcriptomic,proteomic,or metabolomic changes） within a specific tumor.We applied the TCI algorithm to 4,468 tumors across 16 cancer types from The Cancer Genome Atlas（TCGA） and identified those SGAs that causally regulate the differentially expressed genes（DEGs）within each tumor.TCI identified 424 SGAs that had a significant functional impact on transcription in tumors,including most（86%） of the previously reported drivers as well as many novel candidate drivers.Our computational evaluation of these SGAs and DEGs support that the causal relationships inferred by TCI are statistically robust and biologically sensible,and preliminary experimental results support the predicted novel functional impact of previously understudied SGAs.