derivation of potentials for dihedral

68 views
Skip to first unread message

Ali Khodayari

unread,
Feb 20, 2020, 12:29:28 PM2/20/20
to vo...@googlegroups.com

Dears,

 

I am trying to derive potentials for a CG cellobiose-water system. I am using IBI. The non-bonded interactions are derived using the distributions taken from csg_stat, and I am trying to use csg_boltzmann to get the potentials for bonded interactions.

 

As the simulations crash at the very first step of the iteration, I tried to add the potentials one by one (first bonds, then angles, and at the end dihedrals).

 

An error which I face regularly is:

 

A tabulated bond interaction table number 1 is out of the table range: r

0.501462, between table indices 501 and 502, table length 501

 

I believe this is due to high forces between atoms at some point, shooting the atoms to distances higher than the table’s length.

 

After solving the issues with the bonds and angles, I am now trying to get a nice potential for the dihedrals in my system. Yet, the potentials for the dihedrals always look very weird (see table_d1.xvg). Without including this table, the simulations in gromacs at least run for 1000 steps (1ps), but including this crashes the simulations at the very first step. I am using the following commands for csg_boltzmann:

 

csg_boltzmann --cg "cellobiose.xml;water.xml" --top md_at.tpr --trj md_at.trr < boltzmann_cmds2

 

having Boltzmann_cmds as:

 

tab set T 300

tab set smooth_pot 1

tab set n 101

tab dihedral_yrry.pot *:dihedral-YRRY:*

q

 

The output is already looking very strange (see dihedral_yyry.pot)

Then I use the following commands to get the table:

 

BOND_FILE=$"dihedral_yrry.pot"

SETTINGS="convert_dihedral.xml"

 

# cut bad sampled regions at the boundaries

sed -e '1,33d' -e 's/$/ i/' $BOND_FILE | tac | sed -e '1,0d' | tac > "$BOND_FILE.cut"

 

csg_call table smooth "$BOND_FILE.cut" "$BOND_FILE.smooth"

csg_resample --in "$BOND_FILE.smooth" --out "$BOND_FILE.refined" --grid 0:0.001:3.135

csg_call table extrapolate --function quadratic --region left "$BOND_FILE.refined" "$BOND_FILE.refined1"

csg_call table extrapolate --function quadratic --region right "$BOND_FILE.refined1" "$BOND_FILE.cur"

csg_call --ia-type dihedral --ia-name dihedral --options $SETTINGS convert_potential gromacs "$BOND_FILE.cur" table_d${2}.xvg

 

and the .xml file looks like this:

 

<cg>

  <bonded>

      <bondtype>dihedral</bondtype>

    <name>dihedral</name>

    <min>0</min>

    <max>3.1415</max>

    <step>0.001</step>

    <inverse>

      <gromacs>

        <table>table_d1.xvg</table>

      </gromacs>

    </inverse>

  </bonded>

 

  <inverse>

    <gromacs>

      <pot_max>1e6</pot_max>

      <table_end>3.1415</table_end>

      <min>0.0</min>

      <max>3.1415</max>

      <table_bins>0.004</table_bins>

    </gromacs>

  </inverse>

</cg>

 

I have tried to cut different edges of the potential, but they all seem not logical to me, and moreover, cause the simulation to crash at the very beginning.

 

Would you please kindly let me know if I am doing something wrong and/or how I can resolve this issue?

 

My best,

Ali

table_d1.xvg
dihedral_yrry.pot

Christoph Junghans

unread,
Feb 20, 2020, 6:48:05 PM2/20/20
to vo...@googlegroups.com
All that looks pretty reasonable to me, we recently found a bug in
dihedral potential conversion see:
https://github.com/votca/csg/pull/500
but in general dihedral IBI is pretty still pretty experimental in IBI
and we should most likely add a warning about that to the code.
I, personally, have not worked with dihedral potentials enough to give
any advice on this matter.

Christoph
>
>
>
> My best,
>
> Ali
>
> --
> You received this message because you are subscribed to the Google Groups "votca" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/003b01d5e813%2416be71a0%24443b54e0%24%40gmail.com.



--
Christoph Junghans
Web: http://www.compphys.de

Ali Khodayari

unread,
Feb 21, 2020, 5:43:53 AM2/21/20
to vo...@googlegroups.com
Dear Christoph,

Thank you for your response. Shall I replace the new potential_to_gromacs.sh with the one I already have?

I have another question. Could it be because of the type of dihedral I chose that the potential does not exhibit high peaks on the two ends? The thing that does not show it good to me is that I have very low energies on the two ends of the parabolic shape.

How can I adjust the amplitude of the two sides? For some of the potentials I see very large peaks of 50000 and for some only 100 after extrapolating the two sides. I also tried diffent extrapolators for my potentials.

My regards,
Ali
To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAHG27e4b2UfYXQY9yyvZZGWxxDgWgUotf6wbBRcaQdx4H_e5JA%40mail.gmail.com.

Christoph Junghans

unread,
Feb 21, 2020, 8:18:44 AM2/21/20
to vo...@googlegroups.com
On Fri, Feb 21, 2020 at 3:43 AM Ali Khodayari <a.khod...@gmail.com> wrote:
>
> Dear Christoph,
>
> Thank you for your response. Shall I replace the new potential_to_gromacs.sh with the one I already have?
Yeah, just copy the new version over in your installation.

>
> I have another question. Could it be because of the type of dihedral I chose that the potential does not exhibit high peaks on the two ends? The thing that does not show it good to me is that I have very low energies on the two ends of the parabolic shape.

I honestly don't know.

>
> How can I adjust the amplitude of the two sides? For some of the potentials I see very large peaks of 50000 and for some only 100 after extrapolating the two sides. I also tried diffent extrapolators for my potentials.
The extrapolation is done by taking the average slope of last 3 valid
points on each side and that sometimes over/underestimates the slope,
but usually the extrapolated part of the potential doesn't really
matter much (for bonds and angles at least).

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/001001d5e8a3%24ccee9a00%2466cbce00%24%40gmail.com.
Reply all
Reply to author
Forward
0 new messages