Amber (https://ambermd.org) is a molecular simulation software package that has maintained an active development community and user base for over 40 years. Amber is distributed in two parts: Amber which is licensed out of UCSF (and is free for academics/non-profits) and contains the GPU-accelerated MD/FE engine, and AmberTools which is free and open source, and consists of several independent packages, including its own flexible MD code. AmberTools is standalone, whereas Amber requires AmberTools. Amber and AmberTools are released in alternate years in a 2-year cycle.
The York Lab is extensively involved in the development of both Amber and AmberTools. In Amber, the York Lab develops the GPU-accelerated free energy simulation and related enhanced sampling methods as part of pmemd/pmemd.cuda. In AmberTools, the York Lab develops the Sander, HFDF and FE-ToolKit. Amber and AmberTools are both available for download from the Amber web site: https://ambermd.org.
Outside of Amber, the York Lab develops the Amber Drug Discovery Boost package (which includes FE-Workflow, as well as patch updates to pmemd/pmemd.cuda, designated “FE-engine”, and FE-ToolKit; FE-ToolKit is also released with AmberTools). The Amber-DD Boost package is available for download from the Laboratory for Biomolecular Simulations Research (LBSR) GitLab repository: https://gitlab.com/RutgersLBSR.
As part of the York Lab’s effort to create cyberinfrastructure for sustained scientific innovation, a number of interfaces/APIs have been developed in order to integrate different quantum and machine learning codes for advanced applications. Examples of quantum chemical codes include: QUICK, HFDF, ORCA and DFTB+. Examples of machine learning potentials such as DeePMD-kit and TorchANI. These interfaces/APIs enable tight integration to enable the creation of a rich set of multi-layered hybrid “QM/MM+∆MLPs” potentials that can be used in molecular dynamics/free energy simulations. The simulations can be further enhanced by interfaces/APIs to PLUMED, i-PI (universal force engine) and WESTPA to enable use of generalized coordinates and collective path variables (PLUMED), path integral molecular dynamics for treatment of nuclear quantum effects (i-PI), and weighted ensemble simulations (WESTPA). These are under various stages of development, and available through the Amber web site.
The York Group has developed deep-learning potentials for biochemical reactions catalyzed by protein and nucleic acid enzymes, and for prediction of drug binding to biological targets. For this purpose we have been involved in the development of DeePMD-kit, a powerful open-source software package that facilitates molecular dynamics simulations using machine learning potentials using a number of advanced features, and DP-GEN (Deep Potential GENerator), a set of tools designed to generate deep-learning potentials for atomistic simulations.
These programs have been interfaced with Sander for training new deep-learning potentials and using them in molecular dynamics and free energy simulations. The deep-learning potentials developed can be downloaded from the LBSR GitLab repository.
As part of the process of training deep-learning potentials, testing and validating new force fields and enhanced sampling methods, benchmark datasets must be created. These datasets are of stand-alone importance as part of the infrastructure needed to develop and test new methods and models. We have generated a number of such datasets that we make available to the broader community to build from. As these datasets are collected and curated, they will become available for downloaded from the LBSR GitLab repository.
The York Lab is active in a number of education and outreach efforts. These include:
York Lab Amber Tutorials : tutorials, activities and training materials for conducting alchemical free energy and free energy surface simulations, analyzing results and making high-level corrections using advanced features and recommended best practices. As these are developed, they will accessible through our web page (coming soon!) and available for download on the LBSR GitLab repository.
Teaching Materials and Instructional Videos : teaching materials in quantum chemistry, statistical mechanics and computational chemistry, including instructional videos are currently under development and will be released as they become mature and full available.
OrbitalExplorer : online interactive tools, games and active learning material that are designed to engage learners to learn atomic orbitals.