Single-Decoupling Alchemical Binding Free Energy Estimation
The Single-Decoupling Binding Free Energy Method (SDM) is an absolute binding free energy estimation and analysis methodology based on a statistical mechanics theory of molecular association and efficient computational strategies built upon parallel Hamiltonian replica exchange, implicit solvation, and multi-state statistical inference. The method is implemented in OpenMM using the SDM plugin and the SDM workflow packages.The workflow It takes as input .mae files of receptor and ligand plus a definition of the binding site region. Analysis of the results produces, among other things, the estimated values of the binding free energy. See the documentation and example provided.
This work is supported by a generous grant from the National Science Foundation.
AGBNP (see AGBNP and AGBNP2 publications) is an analytic implicit hydration model primarily aimed at alchemical binding free energy calculations. The legacy CPU version of AGBNP is freely available as a portable library written in C. More recently, we have been focusing on a GPU implementation of AGBNP as part of the OpenMM library which is many times faster than the best CPU implementation. The AGBNP OpenMM Plugin is freely available on Github under the GPL license. The plugin currently implements on GPUs AGBNP version 1 (as well as the GaussVol model, see below). The GPU implementation of AGBNP2 is underway.
GaussVol OpenMM Plugin
The GaussVol plugin implements in OpenMM the Gaussian model of Grants & Pickup for the fast calculation of the volume and surface area of macromolecules. The model is the basis of the AGBNP model above. The plugin enables molecular dynamics simulations on GPUs with an effective potential energy function proportional to the solute volume and/or the solute surface area as in, for instance, models of hydrophobic solvation. The GPU implementation is described in a publication in the Journal of Computational Chemistry. The original GaussVol plugin code is available on Github. Development and support of GaussVol continue as part of the AGBNP OpenMM Plugin described above.
Asynchronous Replica Exchange Framework
Asynchronous replica exchange (or ASyncRE for short) is an implementation of the popular parallel replica-exchange conformational sampling algorithm. Unlike traditional synchronous implementations, ASyncRE can scale to many hundreds of replicas over heterogeneous grids of unreliable compute nodes for long running times. Our lab maintains a version of ASyncRE for SDM calculations with OpenMM.
Unbinned-WHAM R Package
UWHAM is an R package for multi-state free energy estimation and thermodynamic reweighting I developed jointly with Zhiqiang Tan at the Department of Statistics at Rutgers University. It provides to R users functionality similar to the MBAR python package. MBAR and UWHAM return identical point estimates of free energies and expectations using different numerical routes. See the UWHAM publication.To install UWHAM enter:
within an R session. Documentation and examples are available within R by entering
> library("UWHAM") > help(uwham)
or in pdf format from
Variance estimation is based on the Fisher or Sandwich formulas, or, for correlated data, by block bootstrap. Processes corresponding to multiple independent simulations, serial tempering or serial Hamiltonian hopping simulations, and parallel replica exchange simulations of both the synchronous and asynchronous varieties are supported.
The UWHAM package is particularly robust in terms of scalability; we have been able to process within minutes large 2D replica-exchange datasets with ~300,000 snapshots from ~200 thermodynamic states. Feedback, comments, and bug reports