From Data Factory to Better Drugs: Part 3 - Creating Maps of Biology

“The data is used to train our models. The models point to the targets. That’s lab-in-the-loop.” Generating millions of data points is only the first step. You need immense computational power to make sense of it (and maybe a good pair of headphones, because BioHive-2 is loud). In Part 3 of our series on our end-to-end platform, Chief Technology Officer Ben Mabey explains how Recursion turns raw data into a novel view of biology. 🖥️Using BioHive-2, one of the fastest supercomputers in the world, we train state-of-the-art foundation models to process every image and sequence we generate. These models create a unique embedding, or fingerprint, for every cellular perturbation we have ever performed. All of this high-dimensional data is compiled into a single massive dataset to form a comprehensive map of biology. By comparing known disease fingerprints against this map, we can find other genetic manipulations that create similar effects. This allows us to identify targets that drive or rescue a disease state, generating new hypotheses and pointing us towards potential novel discoveries. 🚀Follow along over the coming weeks as we continue this series, covering how Recursion maps biology, optimizes precision drug design, and closes the gap between patient and discovery.