Special Seminar in Applied Mathematics
Recent years have seen the rapid growth of large-scale biological data, but the effective mining of big data for new biological discoveries remains a significant challenge. To systematically understand cancer progression and drug resistance, I have developed computational frameworks to integrate a large number and variety of public datasets. In the first part, I will present algorithm RABIT, which searches for gene expression regulators in tumor progression. RABIT utilizes Frisch-Waugh-Lovell theorem for efficient feature selection across vast amount of candidate cancer drivers. Our method significantly outperformed previous feature selection algorithms and reveals many previously unidentified aspects of gene regulation in cancer. In the second part, I will present our work on understanding drug resistance mechanisms in targeted cancer therapy. We found high dimensional interaction feature screening from public chemical datasets is particularly predictive of drug resistance characteristics. Our analysis reveals several possibilities for combinatorial treatment to prevent clinical resistance.