customFunction module as reaction coordinate

[satyajitkhatua09]

Hi,

This is Satyajit Khatua. I am here to gather some expert opinions about studying the conformational dynamics of DNA unwinding. To capture both the rotation and the translocation of helices, I have to use the combination of some colvars that capture the change in distance and rotation around a helical axis.

Do any of you have any idea about the linear combination of any two such colvars (like distance/diatanceZ and spinAngle) to implement inside the customFunction colvar that can capture the conformational change in DNA in terms of simultaneous rotation and translocation?

Best,
Satyajit Khatua

[giacomofiorin]

Hi Satyajit!

To use a linear combination you could use componentCoeff:
https://colvars.github.io/master/colvars-refman-namd.html#sec:cvc_superp
The example below defines a colvar that is the average of two functions:

colvar {
    name xi
    function1 {
        componentCoeff 0.5
        ....
    }
    function2 {
        componentCoeff 0.5
        ....
    }
}

customFunction is much more flexible, because you can use many more mathematical functions, and I do recommend using it. The only caveat is that GROMACS and VMD currently would not support it.

If you have a specific idea in mind, please describe it so that we can confirm whether it's supported or not.

Cheers!

[satyajitkhatua09]

Hi @giacomofiorin,

I will give it a try. One quick question, inside function brackets I have to put the colvar module as is defined in the manual, right?

Best,
Satyajit

[giacomofiorin]

Based on what you mentioned, function1 would be distanceZ and function2 would be spinAngle. Each block that follows them would have their usual parameters, plus the coefficient for linear superposition.

If you stick with linear superposition, I would definitely recommend also using the Dashboard in VMD.

If you use customFunction, the Dashboard would be less useful because, sadly, the VMD developers would not allow us to include the Lepton library, which is needed for that feature.

[satyajitkhatua09]

Hi @giacomofiorin,

Here is the snippet of my configuration file where I have turned the collar module on.

# Incorporating the colvar file
colvars on; # turing collective variable (CV) module on
colvarsConfig colvars.in; # file containing the CV information

Inside the colvars.in file, I have defined the two CVs as follows:

colvarsTrajFrequency 100
colvarsRestartFrequency 500

colvar {
name xi

using the coefficient for linear superposition

function1 {
componentCoeff 0.5
distanceZ {
main {
atomsFile atoms.pdb
atomsCol B
atomsColValue 1.0
}
ref {
atomNameResidueRange CA 3778-3956
}
ref2 {
atomNameResidueRange CA 408-623
}
}
}
function2 {
componentCoeff 0.5
spinAngle {
atoms {
atomsFile atoms.pdb
atomsCol B
atomsColValue 1.0
}
refPositionsFile ref_pos_inpcrd.pdb
axis (-0.5, 0.3, 0.8)
}
}
}

However, when I ran the simulation, it threw the error message attached below. I am unable to understand the real problem here that reflects such behavior.

colvars: ----------------------------------------------------------------------
colvars: Initializing the collective variables module, version "2020-07-07".
colvars: Please cite Fiorin et al, Mol Phys 2013:
colvars: https://dx.doi.org/10.1080/00268976.2013.813594
colvars: in any publication based on this calculation.
colvars: SMP parallelism is enabled; if needed, use "smp off" to override this.
colvars: This version was built with the C++11 standard or higher.
colvars: Using NAMD interface, version "2020-05-04".
colvars: ----------------------------------------------------------------------
colvars: Reading new configuration from file "colvars.in":
colvars: # units = "" [default]
colvars: # smp = on [default]
colvars: # colvarsTrajFrequency = 100
colvars: # colvarsRestartFrequency = 500
colvars: # scriptedColvarForces = off [default]
colvars: # scriptingAfterBiases = off [default]
colvars: ----------------------------------------------------------------------
colvars: Initializing a new collective variable.
colvars: # name = "xi"
colvars: Error: no valid components were provided for this collective variable.
FATAL ERROR: Error in the collective variables module (see above for details)
[Partition 0][Node 0] End of program

[giacomofiorin]

"function1" is only an example, which in your case would be distanceZ

[satyajitkhatua09]

Hi @giacomofiorin,

Thank you, now it is running perfectly.

So, if I want to use the customFunction option, do I have to first manually deduce the functional form that co-relates the two CV modules?

[jhenin]

Hi @satyajitkhatua09 ,
to add to Giacomo's advice: if you already have example structures for the end points (or simulation trajectories), you can load them in VMD and experiment freely with variables using the Colvars Dashboard, until you find one that suits your needs.
Jérôme

[satyajitkhatua09]

Hi @jhenin,

Thank you for letting me know the feature. It is very useful to define the colvar modules.

Best,
Satyajit