The AI-Driven Multiscale Investigation of the RAS/RAF Activation Lifecycle (ADMIRRAL) project was created under the Cancer Moonshot℠ program and is part of the NCI-DOE Collaboration.
The aims of the ADMIRRAL project are to study the dynamic reorganization in the RAS-RAF protein complex during activation and protein domain movement while interacting with the molecular membrane. The approach used by the ADMIRRAL team uses experimental data and approaches for model parameterizations as well as validates hypotheses generated by the machine-learning-driven computational models.
The ADMIRRAL project utilizes a novel combination of molecular dynamics (MD), coarse-grain models, and deep learning to explore the RAS/RAF interaction across increasingly biologically relevant time scales. At the atomic scale, multiple large-scale molecular dynamics simulations are executed creating very large ensembles of MD trajectories, coarse-grain Martini force field models are used to study macros-scale dynamics, and density functional theory is utilized to study the largest scale models allowing the largest timesteps. Deep learning is used to maintain coherence across the different space and time scale models.
The Multiscale Machine-Learned Modeling Infrastructure (MuMMI) software was developed to support the multiscale simulation requirements of the ADMIRRAL project and was awarded best paper at the SC19 supercomputing conference.
Datasets for the early runs for the ADMIRRAL project (previously known as NCI-DOE project Pilot 2) are available through the NCI Predictive Model and Data Clearinghouse (MoDaC) and the Dataset Finder. This dataset comprises the largest collection of publicly available RAS molecular dynamics trajectories.
Publications related to ADMIRRAL are available in the Publications Library
The ADMIRRAL project is co-led by:
- Fred Streitz, Lawrence Livermore National Laboratory
- Dwight Nissley, Frederick National Laboratory for Cancer Research
The RAS Initiative is a national mission at the Frederick National Laboratory for Cancer Research and serves as a hub for the ADMIRRAL project and collaboration.
Contact us at this link or by email at computational-cancer-tech@nih.gov.