Over the years, much progress has been made in cancer research to gather omics and clinical data. Here, we show how genomics and clinical data can be exploited using integrative data analysis and deep learning to identify potential novel gene targets in breast cancer. In brief, we retrieved, processed and built a DL model using genomics, transcriptomics, and proteomics data for 483 Breast Invasive Carcinoma (BRCA) samples available in The Cancer Genome Atlas database. We then conducted functional enrichment and survival analysis to determine potential genes of interest. Our approach identified 83 relevant genes affecting cancer pathways of which 52 genes have no known associated drugs. Furthermore, survival analysis stratified by alterations in each of these 52 genes revealed that patients with alterations in BRF2 had a lower probability of survival than those without alterations. This illustrates that our approach successfully identified BRF2 as a potential drug target. To validate this in-silico, we conducted structure-based virtual screening and we identified eight compounds including Olaparib which is used in the treatment of cancers such as ovarian, breast, pancreatic, and prostate. This shows the power of integrative data analysis and deep learning approaches in identifying candidate drug targets.