Success Story: Using artificial intelligence to make molecule discovery more efficient

NRC collaboration focuses on using AI for health and climate benefits

At the root of new health discoveries like vaccines or drugs, lie molecules with the properties needed for a desired result, such as immunity against a new pathogen, like COVID-19, new antibiotics or a new way to treat an old disease.

The process of identifying the molecules needed for a specific cause often takes more than 10?years, costs over $1?billion (US) and is a setback in many fields beyond healthcare. Artificial intelligence (AI) offers a faster, less expensive way of sorting through the large amount of data and huge candidate space to determine which molecules are the best fit.

National Research Council of Canada (NRC) scientists working in the Artificial Intelligence for Design (AI4D) Challenge program are developing ways to harness the power of AI for advancing scientific discovery for health and?climate benefits, as part of a $2.5?million, 3-year project.?Funding for this project is being provided by the NRC's Collaborative Science, Technology and Innovation Program .

Using AI, the team is currently accelerating the process of discovering molecules that can be used in healthcare and clean energy applications, while reducing costs. The project also focuses on molecule identification for developing green materials for clean energy, and on environmental protection, such as innovative battery technology to power electric vehicles or better methods for carbon capture and storage.

Collaboration leading to new pathways for better health and a cleaner environment

"With billions of potential candidates, searching for a particular molecule can be like looking for a needle in a haystack. AI has the potential to cut years off this search," says Harry Guo , Senior Research Officer at the NRC's Digital Technologies Research Centre , and project lead. "Through exploration and active knowledge-seeking learning algorithms, AI can learn from both the data and experience when, where and how to explore the search space in order to better guide the molecule search."

Harry Guo is partnering on the project with one of the world's leading AI researchers, Yoshua Bengio , recipient of the 2018 Turing Award (the computer science equivalent of the Nobel Prize) and founder of Mila , the Quebec Artificial Intelligence Institute. Mila is the world's largest academic deep learning research centre and a key partner in the Pan-Canadian AI Strategy . The ultimate goal of the collaboration is to build an efficient, high-performance machine learning framework for finding molecules with designated properties.

"I believe AI-driven scientific discovery is going to be a major advancement for scientific research in general and could help us tackle critical challenges in healthcare and the environment," says Yoshua Bengio. "I believe it is very likely this work will be extremely beneficial and transformative for both science and society. In the future, large-scale AI systems may help us come up with greater scientific theories and better evaluate the merits of competing theories to advance knowledge and drive innovation."

Cross-team collaborations

In addition to lead partners AI4D and Mila, project participants include the NRC's Security and Disruptive Technologies Research Centre , as well as external partners at 加拿大麦吉尔大学 and The Hospital for Sick Children . Their participation not only advances the creation of the machine learning framework, but also its application, contributing to the NRC's long-term goal of developing AI tools to accelerate Canada's capacity for scientific and engineering innovations.

"Combining domain and machine learning expertise together into collaborative projects is absolutely necessary to advance artificial intelligence for scientific discovery," says Joel Martin, Chief Digital Research Officer at the NRC. "Multiple perspectives and diverse expertise promote novelty and I know in this project will accelerate innovation."

Faster more affordable breakthroughs for Canadians

Canada's healthcare system is under enormous pressure. Additional pressures are posed by the need to reduce our greenhouse gas emissions to lessen the effects of climate change on the environment and our health. AI has the potential to ease those pressures and support the innovations we need. This project is developing the tools to make an AI-based solution a reality.

Read more stories about NRC research , and how our work contributes to the success of our clients and partners.