This week on The Genetics Podcast, Patrick is joined by Dr. James Zou, Associate Professor of Biomedical Data Science at Stanford University. They discuss the concept of virtual biotechs and how AI agents link single cell data to clinical trial success, Paper to Agent and turning papers into living virtual authors, clinical AI from EchoNet to SleepFM, and the role of academia alongside big AI labs.
Show Notes
0:00 Intro to The Genetics Podcast
00:59 Welcome to James
01:55 An overview of James’s virtual biotech and its AI scientist agents
04:11 How single cell evidence behind drug targets predicts clinical trial success
06:33 Behind the scenes of how autonomous the virtual biotech agents are
08:06 Missing proprietary trial data as the next gap for the virtual biotech
09:44 Scaling the number of collaborating AI agents as a new frontier for drug discovery
12:29 How teamwork and structure evolved among the agents
15:53 Behind the idea of Paper to Agent and virtual authors
19:01 What converting over 100 papers into agents revealed about reproducibility
23:14 The story behind EchoNet, an FDA-cleared AI tool for cardiac ultrasound
26:28 Predicting risk for over 100 future diseases from one night of sleep
27:40 Transferring SleepFM from sleep labs to consumer wearables
28:43 Combining multimodal health data to predict disease
30:10 Longitudinal blood proteomics as the biggest data gap for health AI
31:18 The role of academia as an open, neutral counterweight to big AI labs
32:51 How James’s path from Cambridge math and Harvard machine learning led to biology
34:41 Closing remarks
Find out more
Virtual biotech paper (https://www.science.org/doi/10.1126/science.aeg6779)
Paper2Agent paper (https://www.nature.com/articles/s41586-026-11044-y)