Building AbilityGPT – A GPT model to create novel biological therapeutics

Antibody therapeutics are a fast-growing class of treatments for many significant unmet medical needs. Ability biologics is a Montréal-based startup company that has significant unique proprietary data, sequences and activities for over 40 targets, for training deep-learning models, combined with laboratory facilities to create and test antibodies proposed by computer models. Ability Biologics will use the latest computer science and deep-learning technologies with our large data set to bring proposed therapeutics to reality.
The objective of this project is to explore the use of GPT pretrained-models trained on our unique data set of human antibody sequences. Then characterize the models by comparing to measured data and similar efforts from scientific literature, and eventually by synthesizing and characterizing the antibodies in our laboratories. The ability of the models to generate novel therapeutics with properties such as binding, solubility and developability will be explored as future projects.

Faculty Supervisor:

Mathieu Blanchette

Student:

Partner:

Ability Biotherapeutics

Discipline:

Life Sciences

Sector:

Health and Related Sciences & Technology; Artificial Intelligence

University:

McGill University

Program: