Application of statistical machine learning in identifying candidate biomarkers of resistant to anti-cancer drugs in ovarian cancer

Drug resistance is one of the major challenges in the treatment of ovarian cancer. To facilitate identification of candidate biomarkers of resistant to platinum-based chemotherapy in ovarian cancer, we employed statistical machine learning techniques and integrative genomic data analysis. We used gene expression, somatic mutation and copy number aberration data of platinum sensitive and resistant tumors from the cancer genome atlas. Using regression tree and module network analysis, we identified genes that both contain mutations (copy number aberration and/or point mutation) and their expressions influence groups of their co-regulated genes for resistant and sensitive tumors. Finally, we compared these two gene lists and their associated pathways to extract a short list of genes as potential biomarkers of resistant to platinum-based chemotherapy.