Statistical models for clinical genomics of cancer

Ovarian cancer is the most fatal gynecological cancer in North American women and the fifth most common cause of cancer death. Breast cancer is the most common cancer in women worldwide. Recent approaches to improving clinical outcomes for these two diseases have focused on defining distinct subtypes within ovarian and breast tumours that differ in their clinical outcomes and responses to therapy. Preliminary evidence suggests that subtypes can be detected from biopsies by performing state-of-the-art molecular tests to determine specific molecules (called markers) that distinguish the subtypes. It is also expected that an even smaller subset of markers could be used as indicators for determining prognosis and for directing therapy tailored to the subtype. A number of BC research programs are currently working collaboratively to identify and characterize the subtypes of ovarian and breast cancers, using more than 2,000 breast cancer and 400 ovarian cancer tumours for which clinical outcomes are known. This work generates massive amounts of molecular data (more than 100,000 data points per tumour). Previously supported by MSFHR funding for his PhD training, Dr. Sohrab Shah focuses his post doctoral work on developing bioinformatics (the application of computer science tools and research to biology) and statistical modeling approaches that can help pinpoint potential markers among the reams of data. Shah’s research is key to developing tools that help uncover the molecular characteristics of the subtypes of breast and ovarian cancers, and provide state-of-the-art classifiers for improved outcomes for patients with these devastating diseases.