CS Tea: Allie Warren '17
Fighting cancer with machine learning: computational methods in pre-clinical cancer research
The changes that cause cancer cells to grow uncontrolled also confer specific vulnerabilities that normal cells lack. Unfortunately, for most cancers we have not identified these genetic vulnerabilities, or the biological features that are predictive of these vulnerabilities. To tackle this problem, scientists at the Broad Institute, and other partner institutions, are working to create a “cancer dependency map'' that profiles thousands of cancer cell line models for genomic information and sensitivity to genetic and small molecule perturbations. I will discuss some of the challenges of using this multi-faceted data to identify and predict vulnerabilities. The clinical applicability of results derived from cancer cell lines (cancer cells that are grown continuously in a lab) remains an important question, however, due largely to uncertainty as to how well they represent the biological characteristics and drug responses of patient tumors. Direct comparisons of cancer cell lines and patient tumors are complicated by several factors, most notably the variable presence of normal cells in tumor samples that are not present in cancer cell lines. I will discuss an unsupervised alignment method (Celligner) that we developed that allows us to directly compare and visualize these data. I will also talk about my path to working in research at the Broad Institute and how I found that experience, as well as about my current experience in public health, and the realities of data science in a pandemic.
Biography: Allie Warren is an epidemiologist in the Data Science Support Unit at the Washington State Department of Health, Office of Public Health Outbreak Coordination, Informatics, and Surveillance. She develops tools to process, link, and analyze COVID-19 sequencing, testing, and patient data. Prior to that, Allie worked as a computational associate on the Cancer Data Science Team at the Broad Institute, where she worked on efforts to systematically identify the chemical and genetic vulnerabilities of different cancer cells, as well as their predictive biomarkers. Allie graduated from Carleton College in 2017 as a Computer Science major.