Decode the regulatory genome

Genome sequence encodes the regulatory programs that govern how organisms develop and function. We build AI models that uncovers the sequence rules and molecular underpinnings to how the genome encodes regulatory function — how regulatory elements specify when and where genes express, how chromatin is organized, and how 3D genome architecture emerges. Beyond predictive capabilities, our methods aim to gain insight into the mechanisms underlying this function and enable increasingly precise and flexible design of genomic sequences.

Nature has run billions of years of natural experiments and optimization on genome regulation across species. We aim to uncover the hidden treasures from this enormous scale of experimentation — AI models for genomic sequences can enable us to understand regulatory programs across the tree of life, including organisms whose biology has remained largely inaccessible to experimental study. By reading the genomes of diverse species, we can identify the principles that govern regulatory design, understand why certain solutions are conserved while others are reinvented, and gain insights into human genome function that would be invisible from studying a single species alone.

We are at an inflection point where AI is transforming what computational biology can achieve. We are interested in advances on two fronts: (1) AI methods built for understanding biological sequences and data and (2) agentic AI for scientific discovery. We believe that agentic AI systems will fundamentally expand what biologists can do — offering entirely new ways of exploring hypotheses at scale, integrating knowledge across domains, and driving important discoveries.