TOOme: A novel computational framework to infer cancer tissue-of-origin by integrating both gene mutation and expression.

• DOI: 10.1200/jco.2020.38.15_suppl.3591
• Publication Date: 2020-05-25T15:24:00Z
• Authors: Wei Gao; Jianwei Yang; Changhua Zhuo; Jialiang Yang; Sha Huang; Geng Tian; Jinyuan Lin
• Keywords: 03 medical and health sciences; 0302 clinical medicine; 3. Good health

3591 Background: Metastatic cancers require further diagnosis to determine their primary tumor sites. However, the tissue-of-origin for around 5% tumors could not be identified by routine medical according. With development of machine learning techniques and accumulation big cancer data from TCGA GEO, it is now feasible predict computational tools. Methods: Developed a framework infer tissue-o. Results: Applied TOOme containing 7,008 non-metastatic samples across 20 solid including BLCA, BRCA, CESC, COAD, GBM so on. 74 genes gene expression profile 6 mutation are selected random forest process, which can divided into two categories: (1) type specific genes, highly expressed or mutated only in one (2) those several with different levels rates. Function analysis indicates that significantly enriched gland development, urogenital system hormone metabolic thyroid generation prostate According multiple-label classification method, performs best 10-fold cross-validation prediction accuracy 96%. We also use 19 metastatic 256 downloaded GEO as independent testing data, achieves 89%. The validation better than using (i.e., 95%) (83%) alone. Conclusions: provides quick yet accurate alternative traditional methods inferring tissue-of-origin. In addition, combining somatic expressions outperform