Publication date: Apr 23, 2019
Biomolecular simulations are computationally expensive. This limits their application in drug or protein design and related fields. Several methods have been developed to address this problem. These methods often use an artificial force or potential acting on selected degrees of freedom known as collective variables. This requires explicit calculation of a collective variable (and its derivatives) from molecular structure. For collective variables that cannot be calculated explicitly or such calculations is slow we developed anncolvar package (https://github.com/spiwokv/anncolvar). This package approximates collective variables using artificial neural networks. It was tested on Isomap low dimensional representation of cyclooctane derivative or solvent-accessible surface area of Trp-cage miniprotein.
No Explore or Discover sections associated with this archive record.
|14.1 MiB||Input files (input structure, topology, Plumed input) for simulations used to demonstrate functionality of anncolvar (https://github.com/spiwokv/anncolvar). Input files for metadynamics simulation of cyclooctane derivative in vacuum with three Isomap CVs, metadynamics (mtd), parallel tempering (ptmd) and parallel tempering metadynamics (ptmtd) of Trp-cage in water with solvent-accessible surface area and alpha-RMSD CVs are provided. Tested on OpenMPI4.0.0, Gromacs 2018.5 and Plumed2.5.0. Scripts provided.|
|4.2 KiB||README.txt file with descriptions.|