The last decade has brought about an exciting convergence of technologies that has exponentially increased our ability to answer complex questions, both in and outside of the medical world. We couldn’t help but wonder what efficiencies we could create and what problems we could solve if we applied the same sort of data science to drug discovery that powers facial recognition, helps autonomous vehicles to navigate and can analyze a human genome for 1/10,000th of what it cost just over a decade ago.

To find out, we created our own convergence — a rare, tight-knit collaboration between data scientists, software engineers, biologists, and automation specialists constantly asking biological questions and iterating across the experimental biology life cycle to generate high-dimensional, interconnected, biologically relevant datasets. In doing so, we’re standing on the shoulders of the giants of biomedical research. After all, phenotypic drug discovery has been responsible for the discovery of more first-in-class drugs than any other single approach. We reimagined this practice in the age of artificial intelligence, and established the first digital biology company.

Building a map of human biology

We’re in the process of breaking every known gene and measuring the resulting changes in images of multiple human cell types. Every week we ask tens of thousands of questions about everything from genetics to immuno-oncology to diseases of aging as we aggregate the world’s largest biological image set, fit for the purpose of machine learning.

Massive human impact

Our bold ambition is no less than to create a map of human cellular biology. Along the way, we’re leveraging our work to find novel treatments for disease and then partnering with the world’s most successful development companies to get them to patients as quickly as possible.

We’ve already developed a massive database of biological images, each of which is relatable over time to all the others we produce. As the massive search space of biological perturbations, both genetic and otherwise, is filled with data from our image sets and analysis, we’ve  started to understand and model the complex interactions that compounds have with various conditions.

As we continue building our foundation with an image-based approach, we’ll layer in other high-dimensional datasets such as gene expression, proteomics, and metabolomics to improve the reliability and resolution of the map we’re building.

Our targeted approach is what’s leading us to a high-resolution map of biology faster and more inexpensively than you might expect. This map in turn will allow us to significantly increase the speed and decrease the cost of finding treatments, maybe even to the point where we can identify just the right drug for the right patient at the right time. That’s the ultimate goal of everything Recursion does: improving and even extending lives.