Collective Variable Discovery by Deep Encoder-Decoder Model:
The paper is:
"A deep encoder–decoder framework for identifying distinct ligand binding pathways"
By Satyabrata Bandyopadhyay and Jagannath Mondal
J. Chem. Phys. 158, 194103 (2023)
https://doi.org/10.1063/5.0145197
The paper link is: https://pubs.aip.org/aip/jcp/article/158/19/194103/2890463/A-deep-encoder-decoder-framework-for-identifying
Here, in this repository, time-evolution of three different trajectories have been shown with visualization of the snap-shots in vmd along with time-evolution on free energy surface.
(1) a_traj1_4H_6H_vmd_fes.mp4 is for Trajectory-1(a) ---> Ligand Binding Through Helix-4 and Helix-6.
(2) b_traj2_7H_9H_vmd_fes.mp4 is for Trajectory-2(b) ---> Ligand Binding Through Helix-7 and Helix-9.
(3) c_traj3_5H_6H_7H_8H_vmd_fes.mp4 for Trajectory-3(c) ---> Ligand Binding Through Helices-5-6-7-8.
The lig-binding-ae-model.py python file will take the ligand-com-to-C-alpha distances as input vector and after model training and prediction, it will return us the hidden-projected "encoded data". This "encoded data" will be needed for all our future work.
We need to have pre-installed keras and tensorflow library to run our code.