Who We Are

OpenADMET is an organization with the mission to build open predictive models of safety and toxicity for small molecules to improve humanity’s ability to more reliably, cheaply, and effectively treat disease.

Small molecule therapies have always been and continue to be the dominant way we treat diseases worldwide. Small molecules have unique advantages such as cost, scalability, convenience, the ability to get into every organ/cell and modulate a wide variety of targets/mechanisms – especially as compared to other modalities like antibodies, RNA, and genetic medicines. However, those advantages also give rise to the idiosyncratic and unpredictable properties that pose challenges in drug development, such as where in the body these drugs get to, in what amounts, for how long, and how they interact with unintended targets to cause safety issues. The ability to predict these properties, which are collectively called ADMET (Absorption, Distribution, Metabolism, Excretion and Toxicity) properties, from small molecule structure would broadly enable small molecule drug discovery and development.

OpenADMET is an open effort to build predictive models of ADMET properties and understand the mechanisms by which they arise. This means systematically characterizing the proteins and mechanisms that give rise to these properties (protein structures and scaled functional and genetic assays), understanding how small molecules interact with these mechanisms (through high-throughput nanoscale chemistry to explore/exploit chemical space), and integrative computational models (AI/ML ligand/structural, mechanistic, and physiological modeling).

We see our role as a guide to the community by developing open datasets and computational models. One way to ignite innovation in ADMET modeling is through community blind challenges. Blind challenges can provide accurate benchmarks of current performance and help us understand how much we have left to achieve. The paragon example is the CASP challenge that set up conditions for the “AlphaFold” breakthrough in protein structure prediction. For ADMET challenges, we plan on using both our generated data on anti-targets of broad interest and ADMET data donated from the community, as we did in our first challenge with the ASAP AViDD center.

OpenADMET is a nascent coalition of aligned efforts funded by different organizations. It will be governed by a board and be administered by the Open Molecular Software Foundation (OMSF). Our projects currently involve personnel at OMSF, UCSF, Octant, and MSKCC. Our initial funding is through an ARPA-H grant: “AVOID-OME”. Since then, we’ve additionally been funded by the Gates Foundation & Schrodinger, to expand into toxicity and fundamental molecular properties, and by the Astera Institute, to expand our metabolism dataset coverage.

Join Us!

Reach out at openadmet@omsf.io, join the Discord, or compete in our challenges!