AI Compresses Months of Gene Research Into 20 Minutes

What used to take three months in a genetics lab now happens over lunch break, and scientists say it's opening doors to discoveries they never had time to pursue before.

Researchers at Stanford University and other leading labs are using Claude, an AI system, to compress research timelines that once stretched for months into hours or even minutes. The technology handles the tedious work that eats up most of a scientist's time, letting them focus on the creative parts of discovery.

Take genome-wide association studies, which hunt for genetic links to traits like perfect pitch or diseases. The actual genome scanning is straightforward, but analyzing the messy data, cleaning it up, and making sense of what it all means can take months of painstaking work. Using Stanford's Biomni system, powered by Claude, that entire process now takes 20 minutes.

The speed gains are remarkable, but accuracy matters even more in science. In blind tests, the AI designed molecular experiments that matched the work of postdoctoral researchers with over five years of experience. In another trial, it analyzed data from 450 wearable health devices across 30 people in just 35 minutes. Human experts estimated that task would take three weeks.

The technology isn't perfect, and scientists have built in safety checks to catch when the AI goes off track. When working with rare disease diagnosis, researchers found the AI's approach differed from what clinicians would do, so they taught it expert methods step by step. With that guidance, it performed well.

At the Cheeseman Lab, scientists use CRISPR gene editing to remove genes and see what breaks. They can create massive amounts of data, but interpreting it all was the bottleneck. Now the AI handles that interpretation work, letting them run experiments at a scale that wasn't practical before.

Other labs are analyzing 336,000 individual cells from human embryonic tissue in hours instead of weeks. The system confirmed relationships scientists already knew about, but also identified new proteins controlling gene activity that researchers hadn't previously connected to human development.

The Ripple Effect: Scientists say this technology doesn't just make their current work faster. It's changing what kinds of questions they can ask in the first place. Research projects that seemed too time-consuming or expensive to attempt are now becoming feasible. Labs are exploring entirely new approaches they couldn't take before.

Stanford provides free access to researchers working on high-impact scientific projects around the world through their AI for Science program. The goal is to eliminate bottlenecks that have slowed down progress on everything from rare diseases to understanding how life develops.

The scientists we spoke to describe Claude not as a replacement but as a collaborator that works alongside them through all stages of research, from designing experiments to finding patterns humans might miss in massive datasets.

Progress in science often comes down to having enough time and resources to ask the right questions, and now researchers have more of both.