normal run fails with calculated dihedrals

  • Hegedus
  • Hegedus's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
2 years 2 months ago #7737 by Hegedus
run fails with calculated dihedrals was created by Hegedus
Hi,

I am trying Martini3b. After generating .itp, I also run the bbsc.sh.
However, if I run step6.0 (mdrun) then I get the following error:
"gromacs Function type Restricted Angles does not support currently free energy calculations"
I do not try to do free energy calculations.
(I use gromacs 2018.1)

If I use the unmodified protein_A.itp then the run starts. (Soon I also have an error, but first I will try to solve it - a lot of LINCS warnings and "tMPI error: Receive buffer size too small for transmission (in valid comm)")

Therefore I think that it is caused by bbsc.sh

I found two bug-like issues:

1. martinize .itp output:
This is defined #define RUBBER_FC 500.000000
but 500.000 is hard-coded in the following lines
1 233 6 0.74960 500.00000
instead of
1 233 6 0.74960 RUBBER_FC

2. addDihedral.tcl
Most likely
puts "measuring SC BB BB SC dihedrals"
should start in a new line AND as a comment.

Thanks, Tamas

Please Log in or Create an account to join the conversation.

More
2 years 2 months ago #7738 by riccardo
Replied by riccardo on topic run fails with calculated dihedrals
Hi Tamas,

The error you get sounds like being related to the use of Restricted Bending potential (angle function number 10 in GMX 2018) when running free energy calculations: do you have ReB potentials in your itp (check the angle section for angles with function type = 10) after the use of bbsc.sh? Do you have "free-energy = yes" in your mdp?

Please Log in or Create an account to join the conversation.

More
2 years 2 months ago #7739 by jbarnoud
Replied by jbarnoud on topic run fails with calculated dihedrals
Hi,

Regarding the 2 bugs:

* A version of gromacs (I do not know which one) was not honouring the RUBBER_FC variable which lead us to use the value in the ITP instead of the variable.
* About addDihedral.tcl, ideally the line should not even be written in the final file. I am surprised it did not cause problems earlier... You can comment line 191 of the tcl script that writes the line. The script should not be necessary any more with the upcoming version of martinize.

Please Log in or Create an account to join the conversation.

  • Hegedus
  • Hegedus's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
2 years 2 months ago #7740 by Hegedus
Replied by Hegedus on topic run fails with calculated dihedrals
Hi,

Thanks for the suggestions and info. It seems to be solved.
I do not have type=10
Yes, I try to do free energy calculations :-) I have it at the bottom of the mdp file. I forgot that I adopted it from charmm-gui output (soft-core-minimization).

I do not know what was the problem. Most likely because of the lot of the manual steps involved at this moment with my martini3 system setup I made something wrong.

By the way, I have a hole for allowing lipid flip-flop. I also wanted to confirm something regarding the lipid tail constraint coordinate file. I would like to check some formatting issue, but I do not find the description of a constraint coordinate file anywhere... Do you have a link, or page number in the gromacs manual? Thanks!

Have a nice day, Tamas

Please Log in or Create an account to join the conversation.

More
2 years 2 months ago #7741 by jbarnoud
Replied by jbarnoud on topic run fails with calculated dihedrals
Good to hear that your problem is solved. The file for the restrains is a gro or a PDB file with the atom at their reference position.

Please Log in or Create an account to join the conversation.

  • Hegedus
  • Hegedus's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
2 years 2 months ago #7742 by Hegedus
Replied by Hegedus on topic run fails with calculated dihedrals
I was wrong again. An input file was wrong thus a different simulation was running.

I still get the Restricted Angles does not support currently free energy calculations. So I turned off the above mentioned free-energy calculation.

Then it runs but it seems I have a gromacs bug.
tMPI error: Receive buffer size too small for transmission (in valid comm)
mailman-1.sys.kth.se/pipermail/gromacs.o...018-June/120624.html

Have you seen this phenomenon?
Should I install and use gromacs 2016?

Thanks, Tamas

Please Log in or Create an account to join the conversation.

More
2 years 1 month ago #7743 by riccardo
Replied by riccardo on topic run fails with calculated dihedrals
Never seen the tMPI error. I would say that yep, trying another GMX version is a good option.

I am still puzzled that you get the Restricted Angles-related error. Maybe it's not just function type 10 but there are also some other function types of "Restricted" type? Can you double-check that?

Please Log in or Create an account to join the conversation.

  • Hegedus
  • Hegedus's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
2 years 1 month ago #7744 by Hegedus
Replied by Hegedus on topic run fails with calculated dihedrals
The newest gmx version (2018.2) doesn't show the tMPI error.

I am sorry: I have tried too many things to get my simulation run; I also tried it without running bbsc.sh - then I did not have function type 10. But after bbsc.sh, yes, I have

[ angles ]
; SC-BB-BB and BB-BB-SC
2 1 3 10 100 5 ; M(1)-M(1)-I(2) SBB
etc...

This is the last section of the bbsc-modified itp file.

At this moment I have running minimization with bbsc (ftype 10) and turned-off soft-core-minimization including the free-energy calc.

One additional thing comes into my mind: I have an unstructured tail in my protein, so I removed a few RUBBER_BANDS. I am not sufficiently expert to decide if this is enough and OK or it may cause problems.

Please Log in or Create an account to join the conversation.

  • Hegedus
  • Hegedus's Avatar Topic Author
  • Offline
  • Fresh Boarder
More
2 years 1 month ago #7749 by Hegedus
Replied by Hegedus on topic run fails with calculated dihedrals
Hi,

Summary, a possible small bug, a question:

The simulation completed without any problem if I do not do soft-core-minimization (free-energy calc) in the first step of the energy minimization (charmm-gui step6.0 for vesicles).

The dynamics seems much better compared to martini2, even simply by eye visualization.

grompp gives warning for an atom of POPS, so I guess that somewhere you call it CN0 and somewhere else CNO (insane.py?):
Warning: atom name 4027 in system.top and system.pdb does not match (CNO - CN0)

PIP2 (DPP2) would be important for me, but not parametrized yet in Martini3. Can I expect to have it in the near future?

Thanks,
Tamas

Please Log in or Create an account to join the conversation.

Time to create page: 0.115 seconds