Colvars' distancePairs to get minimum distance between molecules

User discussions
,
1

GROMACS version: 2025.3
GROMACS modification: No

I am attempting to run a metadynamics simulation between two molecules with the following config file:


indexFile cpx.ndx
colvarsTrajFrequency 500

colvar {
name r
outputTotalForce yes
lowerBoundary 0.0 # smallest it can physically be
upperBoundary 4.0 # largest that it can physically be
width 0.05 # may need to alter this
extendedLagrangian yes
extendedFluctuation 0.01

# Replaced "distance" with "distancePairs"
distancePairs {
    group1 { indexGroup Ligand }
    group2 { indexGroup BindingSite }
    # "min" tracks the closest approach between any two atoms
    # Use forceVector if you need the direction, but min is standard for PMF
    mode min # suggested by Google Gemini, I can find no reference to this whatsoever on the internet
   }
}

harmonicWalls {
name wall_r
colvars r
outputEnergy yes
forceConstant 10.0
lowerWalls 0.2
upperWalls 2.8
}

metadynamics {
name metad_r
colvars r
outputEnergy yes
writeFreeEnergyFile yes
writeTIPMF yes
writeTISamples yes
hillWeight 1.0 # Suggestion: increase hillWeight for faster separation
gaussianSigmas 0.1
newHillFrequency 50
}

which is a modification of what “worked” for colvars’distanceparameter. I wrote mode min while wishing to select the minimum distance between the two molecules.

However, the output doesn’t make sense:


step         r                     r_r                   ft_r                   E_wall_r               E_metad_r
       0    1.26825742960010e-04  1.26825742960010e-04  0.00000000000000e+00   0.00000000000000e+00   0.00000000000000e+00
     500    1.33747887346636e-04  4.40791848016283e-03 -1.04243672273210e+02   0.00000000000000e+00   9.70221158476317e-01

the distances are unreasonably small. distancePairs makes a matrix from what I understand, and I want the minimum within this matrix. I don’t know what r is, but that’s clearly not the minimum interatomic distance between the molecules.

The reference manual / documentation is very sparse on distancePairs

How can I use metadynamics to drive these two molecules apart from one another, so that I can set the minimum distance between the two molecules, analogous to the parameter that’s output from gmx pairdist?

2

Hi, did you try running that config file? distancePairs is most likely not useful as is (it gives you a vector of values) and mode min does not exist: Google Gemini made it up. The doc is here:

I suggest using distanceInv to have a “minimum distance” kind of variable. (Note that a true min() function would be discontinuous, and unsuitable as a collective variable).

From that input, remove the “writeTI” and “extended*” keywords: you won’t need them.

Lastly, I strongly suggest adding units gromacs at the top and loading this input in the Colvars Dashboard VMD plugin (included in VMD 1.9.4 or later) to test it.

Giacomo