Some questions about restrained-ensemble molecular dynamics (reMD)

From: TeYu Kao (havebeen_at_gapp.nthu.edu.tw)
Date: Wed Jul 24 2019 - 21:13:03 CDT

Hi all,

        I have two simple questions about restrained-ensemble MD, and I already surveyed some resources and can not get answers to my questions. Thus, I am welcome to have any comments and replies, thank you. Also, I will provide my NAMD config file at the end of this letter.

         The goal of my project is trying to get a new structure by guiding crystallography structure with DEER data. Thus, I think restrained-ensemble MD is a matched approach. Basically, I follow the method which is published by Shen et al. in 2015. However, I have some questions about it when I was performing testing simulations.

         First, the gaussian variance to fit the histogram calculated from trajectories in this method seems to be static during the simulation. I’m quite confused about this since the initial gaussian variance from locally enhanced sampling method is usually different from gaussian variance extracted from DEER experimental data. Does it mean that I need to set the gaussianSigma as the variance extracted from DEER experimental data? Is this understanding correct?

         Second, I did not observe the decreasing of MISC energy. As my knowledge, the MISC energy is from the histogram restraint, and the system would prevent current position to make MISC energy drop. However, I did not see this situation happened in any circumstances. My simulation has a difference from Shen et al.’s. Their simulation reduced R1 side chain as a dummy residue, but I did not. I think that would not be the critical factor to lead this problem happened, but I am not sure about this.

        Last, I want to introduce the procedure of my simulation for people who want to know:
        1) I utilized locally enhanced sampling method to generate R1 side chain rotamer, the simulation was performed for 1 ns. In my system, I replicate 10 R1 side chains for 2 residues, so I have 10 * 10 distance in my system.

        2) I used Colvars module to calculate the histogram of this 10 * 10 distance and apply histogram restraint on it. Theoretically, the histogram in trajectories would get close to reference distance distribution eventually.

        My config file is showing below:

## structure input ##
structure replica.psf
coordinates replica.pdb

set inputname LES_test
binCoordinates $inputname.coor
binVelocities $inputname.vel
extendedSystem $inputname.xsc
## structure input ##

## parameter input ##
paraTypeCharmm on
parameters par_all27_prot_lipid.inp

les on
lesFactor 10

colvars on
colvarsConfig Restraints-DEER.in
## parameter input ##

## boundary ##
wrapWater on
wrapAll on
## boundary ##

## VDW force ##
exclude scaled1-4
1-4scaling 1.0
cutoff 12.
switching on
switchdist 10.
pairlistdist 13.5
## VDW force ##

## force par ##
timestep 2.0
rigidBonds all
nonbondedFreq 1
fullElectFrequency 2
stepspercycle 20
## force par ##

## elec force ##
PME yes
PMEGridSizeX 40
PMEGridSizeY 50
PMEGridSizeZ 40
## elec force ##

## T control ##
langevin on
langevinDamping 1
langevinTemp
## T control ##

## P control ##
useGroupPressure yes
useFlexibleCell yes
useConstantArea no

langevinPiston on
langevinPistonTarget 1.01325
langevinPistonPeriod 200.
langevinPistonDecay 50.
langevinPistonTemp $temperature
## P control ##

## output par ##
outputName reLES_test
restartfreq 100
dcdfreq 100
xstFreq 1000
outputEnergies 50
outputPressure 50
## output par ##

## execution ##
firsttimestep 1001000
run 1000000
## execution ##

        Also, the colvars config file is showing below:

Colvarstrajfrequency 500
Colvarsrestartfrequency 500

colvar {

    name DEER_22_176
    outputValue no
    distancePairs {
        forceNoPBC yes
        group1 {
            atomsFile LES.pdb
            atomsCol B
            atomsColValue 22
        }
        group2 {
            atomsFile LES.pdb
            atomsCol B
            atomsColValue 176
        }
     }
}

histogramRestraint {

    name hr_22_176
    colvars DEER_22_176
    refHistogramFile new_structure_22176.dat
    forceConstant 50000
    lowerBoundary 9.4775
    upperBoundary 54.1687
    width 0.6983
    gaussianSigma 1.2
    outputEnergy yes
    writeHistogram on
}

---------------------------------------
Te-Yu Kao (高得瑜)
Ph.D. student
National Tsing Hua University
B25, Chemistry Building, No. 101, Sec. 2, GaungFu Rd., East Dist., Hsinchu City 30014, Taiwan
Email: havebeen_at_gapp.nthu.edu.tw
Tel: +886-3-5715131 #33346

This archive was generated by hypermail 2.1.6 : Mon Nov 18 2019 - 23:20:48 CST