Accelerating Therapeutics for Opportunities in Medicine (ATOM)
The ATOM Consortium is a collaborative public-private partnership focused on developing an open, computer-aided, drug discovery platform to shorten the time it takes to develop new and more effective treatments for cancer. The ATOM Research Alliance is the 501(c)(3) non-profit organization that operates ATOM.
ATOM currently involves the following organizations:
- Brookhaven National Laboratory
- NCI’s Frederick National Laboratory for Cancer Research
- Lawrence Livermore National Laboratory
- Oak Ridge National Laboratory
- University of California, San Francisco
Aims of the Project
- Expand the ATOM portfolio of computational models, tools, and data to enable the scientific community to identify candidate molecules, refine them, and explore novel chemical entities.
- Advance a computer-aided drug discovery approach that uses machine learning, predictive data-driven models, high performance computing, and diverse biological data to develop and optimize new drug candidates.
Available Resources
ATOM Modeling PipeLine
The ATOM Modeling PipeLine (AMPL) is an open-source, extensible software workflow for building and using predictive machine learning models for drug discovery. Available on GitHub and built on Deep Chem, AMPL helps incorporate molecular descriptors, properties, and structural data into predictive models. To learn how to use AMPL software, refer to tutorials provided on GitHub in Jupyter notebooks.
ATOM Models
The NCI Predictive Oncology Model and Data Clearinghouse shares predictive oncology computational models and data. The resource provides access to several models trained with AMPL and associated data sets that have been developed through the ATOM Consortium.
More information about ATOM is available on the ATOM Consortium website.
If you are interested in working with or joining ATOM, please contact the ATOM team.