Building integrative machine learning framework for precision oncology

Traditional cancer treatments have followed a “one size fits all” approach, which limits efficacy and often results in significant side effects.
This research project aims to develop an approach to predict the impact of cancer missense mutations on the drug-protein interactions of cancer treatments. The approach will use the patient’s own genomic profile and will help to tailor cancer treatments for the patient. This will reduce side effects and costs to the patient by selecting optimal treatment options for individual patients. Using binding affinity as a measure of drug efficacy this research will follow techniques similar to prior works, with the use of graph representation learning for the drugs and targets. This project will also explore several ideas to allow for better results by considering the uniqueness of this problem, such as including information from both wild-type and the mutated protein.
The significance of this work to the partner organization (Princess Margaret Cancer Center) is improved patient care and potential improvements in efficiency/costs with the use of such an automated system.

Faculty Supervisor:

Arvind Gupta

Student:

Partner:

University Health Network

Discipline:

Computer science

Sector:

Health and Related Sciences & Technology

University:

University of Toronto

Program: