#122 Why AI Drug Discovery Is So Hard — Dr Imran Haque (Recursion)
Sep 4, 2023
auto_awesome
Dr Imran Haque, SVP of AI and Digital Sciences at Recursion Pharmaceuticals, discusses why AI drug discovery is challenging, the need for better understanding of drug interactions, the trade-offs in translational models, advancements in computational modeling and data acquisition in biology, and the value of being a generalist in the field.
Decoding biology and understanding its complexities hinder AI drug discovery progress.
AI drug discovery relies on data collection, computational modeling, and continuous experimentation to decode biology and accelerate drug discovery.
Deep dives
The Challenges of AI Drug Discovery
AI drug discovery has garnered significant attention and investment, but no drug discovered using AI has made it to market yet. The complexities and difficulties of decoding biology are the major factors hindering progress in this field. The lack of deep understanding about how biology works and the limitations of linear hypotheses about disease contribute to the challenge. The approach taken by companies like Recursion involves collecting extensive data about biology and chemistry, leveraging deep computation to build models, and continuously testing predictions through experiments. The goal is to decode biology and accelerate the drug discovery process for new diseases and specific patient populations.
The Complexity of Biology
Understanding biology and chemistry is a complex task. The interactions between compounds and proteins are not fully understood, and the effects of these interactions in different contexts and tissues are still unclear. Biology is highly complex and does not adhere to simplistic linear models. The challenge lies in comprehending the multi-dimensional nature of biological systems and moving away from reductionistic approaches. The unpredictability and evolutionary history of biology make it difficult to model and simulate accurately, requiring a more comprehensive and interconnected understanding of biological systems.
The Importance of Data and Models
AI drug discovery relies heavily on data collection, computational modeling, and continuous experimentation. The ability to gather high-dimensional, high-quality data about biology and chemistry is crucial in order to make accurate predictions. By generating data through perturbing different agents and observing their effects on human cells, Recursion builds biologically meaningful models using deep learning and image processing techniques. These models allow for a better understanding of the relationships between different perturbations and guide the screening of chemical compounds. The iterative process of gathering more data, improving models, and testing predictions is key to decoding biology and accelerating drug discovery.
The Future of AI Drug Discovery
The convergence of comprehensive computational modeling, massive data sets, and experimental advancements holds great promise for the future of AI drug discovery. Increasing the flexibility and comprehensiveness of computational modeling in biology, coupled with the scale and scope of data acquisition, will enable deeper insights into biological systems. The ability to accurately predict experimental outcomes and connect different data points will revolutionize the drug discovery process. As this field continues to evolve, AI methods will play an increasingly integral role in designing and developing drugs with greater efficiency and effectiveness.
Despite $billions of investment into the sector — there are still no AI discovered drugs on the market. Why?
Recursion Pharmaceuticals (NASDAQ: RXRX) is one of the hottest AI drug discovery companies on the market. Especially after NVIDIA's $50M investment into them.
Dr Imran Haque (SVP of AI and Digital Sciences) explains why AI drug discovery is so tricky — and when it's arriving.
0:00 Intro
0:42 Decoding biology (the big idea)
2:18 Why is AI drug discovery harder than it looks?
4:48 Simulating the Sims (game) vs AI drug discovery
6:58 Building data for AI drug discovery
14:47 Recursion's successes and failures
24:22 The barriers to AI drug discovery
26:45 How optimistic should we be about AI drug discovery
35:26 Being aggressively generalist for success
People
Dr Imran Haque: https://www.recursion.com/team-members/imran-haque
Dr Imran Mahmud: https://www.imranmahmud.com/
Dr Mustafa Sultan: https://www.musty.io/
Get the Snipd podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share & Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode