Improving protein-ligand binding models using multidimensional, mixed-confidence training data

There currently are no approved treatments for COVID-19, as global case counts increase daily. High infectivity and long time-to-recovery for COVID-19 cases is straining health-care systems globally. There is an urgent need to discover therapeutics that treat patients and improve clinical outcomes. Research and Development timelines associated with vaccines and new chemical entities will not reach clinics in time to mitigate the current patient surge, leaving drug repurposing as the most practical short-term solution. Most repurposing efforts to date focus on drugs that have previously shown antagonistic activity against other viruses. A broad patient population with diverse pre-existing conditions calls for as many diverse treatment strategies as possible. This proposed project will apply AI technologies to mine the biological networks formed between virus and human proteins and predict existing FDA-approved drugs that can interrupt these interactions.

Faculty Supervisor:

Bo Wang

Student:

Hamed Janani

Partner:

Cyclica

Discipline:

Medicine

Sector:

Professional, scientific and technical services

University:

University of Toronto