[AMBER] NFE, PMD, and MULTI_RMSD questions

From: Russell Davidson <rbdavid.rams.colostate.edu>
Date: Tue, 13 Mar 2018 17:26:53 -0600

Dear all,

I am currently attempting to run a set of 1D umbrella sampling simulations
using the NFE method, specifically the pinned MD namelist, to obtain the
relative free energies of various conformations of a loop region in a
protein. This loop region is highly mobile during unbiased simulations,
having RMSD values that range up to ~17 Ang relative to the crystal
structure. The protocol used for this RMSD analysis uses an alignment
landmark (C_alpha atoms of protein residues that have low mobility over the
course of the trajectory) to align the analysis frame to the reference
frame, subsequently followed by a RMSD calculation of the loop region.

To enhance sampling of the conformations of the loop region, I would like
to use the MULTI_RMSD reaction coordinate but am running into unexpected
results. I start a US window from a frame with a 17 Ang RMSD value with the
harmonic biasing potential centered at 17 Ang. Additionally, I am applying
a restraint (using ntr=1 in the &cntrl namelist) to the alignment landmark,
thereby preventing COM translation and rotation of my protein away from the
reference structure (crystal structure). With this simulation setup, the
collective variable output file is reporting RMSD values of ~2 Ang. I
hypothesize that the MULTI_RMSD colvar aligns the biased atom selection to
the reference cartesian coordinates (defined in the cv_r variable) before
calculating the RMSD. Is this true? Can anyone point me to the code that is
used to calculate the RMSD values? Any further insight into the MULTI_RMSD
colvar would be greatly appreciated.

Russell Davidson
AMBER mailing list
Received on Tue Mar 13 2018 - 16:30:01 PDT
Custom Search