Project ID: plumID:21.028
Name: From Enhanced Sampling to Reaction Profiles
Archive: https://github.com/EnricoTrizio/TargetedDiscriminantAnalysisCVs/archive/refs/heads/main.zip
Category: methods
Keywords: collective variables, multi-state, machine learning, Deep-TDA
PLUMED version: 2.7
Contributor: Enrico Trizio
Submitted on: 06 Jul 2021
Last revised: 09 Dec 2021
Publication: E. Trizio, M. Parrinello, From Enhanced Sampling to Reaction Profiles. The Journal of Physical Chemistry Letters. 12, 8621–8626 (2021)

PLUMED input files

File Compatible with
OAMe_G2/deepDA_enhanced_sampling/plumed.dat tested on v2.10 tested on master
OAMe_G2/deepTDA_enhanced_sampling/plumed.dat tested on v2.10 tested on master
OAMe_G2/unbiased/bound/plumed.dat tested on v2.10 tested on master
OAMe_G2/unbiased/unbound/plumed.dat tested on v2.10 tested on master
alanine/deepDA_enhanced_sampling/plumed.dat tested on v2.10 tested on master
alanine/deepTDA_enhanced_sampling/plumed.dat tested on v2.10 tested on master
benzoquinone_PT/1_unbiased/plumed.dat tested on v2.10 tested on master
benzoquinone_PT/2_biased/plumed.dat tested on v2.10 tested on master
hBromination/1D_deepTDA/plumed.dat tested on v2.10 tested on master with LOAD
hBromination/2D_deepTDA/plumed.dat tested on v2.10 tested on master with LOAD
hBromination/unbiased/anti_markovnikov/plumed.dat tested on v2.10 tested on master with LOAD
hBromination/unbiased/markovnikov/plumed.dat tested on v2.10 tested on master with LOAD
hBromination/unbiased/reagents/plumed.dat tested on v2.10 tested on master with LOAD

Last tested: 07 Oct 2024, 17:36:14

Project description and instructions
This egg contains the input files to reproduce the simulations with Deep-TDA CV of alanine folding, a calyxarene host-guest system in explicit solvent, the multi-state hydrobromination of propene and a step-wise intramolecular double proton transfer reaction. The Deep-TDA CV is expressed as the output of a feed-forward neural network trained imposing that the configurations in the different minima are distributed in the projected CV space according to a preassigned target distribution. Note that the build fails because PLUMED needs to be manually compiled with Libtorch to import the trained model from Pytorch with the interface from Bonati et al., J. Phys. Chem. Lett., 2020, 11, 2998–3004. We have made available a Google Colab Notebook here with a tutorial for the training of the Deep-TDA CV and indications on the PLUMED configuration required. The input files for the training are also included in this egg.

Submission history
[v1] 06 Jul 2021: original submission
[v2] 09 Dec 2021: updated doi

Badge
Click on the image below and get the code to add the badge to your website!
plumeDnest:21.028