Understanding drug response to improve precision cancer medicine

One unexploited opportunity in current biomedical research is the integration of multi-omics data into interpretable and actionable computational models that surpass single gene biomarkers. Approaches routinely used in molecular biology fail to cope with the multidimensionality of high-throughput data, whereas routine statistical methods and machine learning algorithms cannot account for the complex interactions found in biological systems. These limitations are particularly tangible in translational cancer research because oncogenes are tangled in a network of signaling proteins, the number of genomic features greatly surpasses the number of samples, and most tumors have complex genetic profiles. I aim at developing innovative computational methods that exploit the richness of recent profiling efforts and available patient data to advance our understanding of cancer biology and guide treatment personalization.