Practical AI Deep learning technology for drug discovery
Mar 9, 2021
Abraham Heifets, Co-founder and CEO of Atomwise, modifies the landscape of drug discovery using AI. He dives into the use of deep learning for predicting molecule binding, tackling diseases once deemed 'undruggable.' Heifets discusses the synergy of AI with traditional methods, highlighting breakthroughs in protein structure understanding. He also emphasizes the importance of innovative collaborations in AI-driven drug development, showcasing how these advances could revolutionize treatments for diseases like cancer and Alzheimer's.
AI Snips
Chapters
Transcript
Episode notes
Physicists' Early Attempts
- Abraham Heifets discusses the history of computer approaches to chemistry prediction, starting with physicists' attempts.
- Physicists' initial approach involved complex quantum mechanical simulations, which proved computationally expensive.
Chemists' Approach
- Chemists, building upon the physicists' work, incorporated chemical knowledge and features like hydrogen bonds.
- This approach aimed to predict binding, recognizing its significance in how medicines interact with proteins.
Drug Discovery Basics
- Heifets explains drug discovery basics: medicines act as "monkey wrenches," blocking malfunctioning proteins.
- Effective drugs must bind strongly to target proteins while avoiding interactions with healthy ones.