3DNA Forum

Questions and answers => MD simulations => Topic started by: cridemichel on June 18, 2012, 08:57:06 am

Netiquette · Download · News · Gallery · G-quadruplexes · DSSR-Jmol · DSSR-PyMOL · Video Overview · DSSR v2.5.1 (DSSR Manual) · Homepage

Title: fiber and gromacs
Post by: cridemichel on June 18, 2012, 08:57:06 am
Hi All,
I am a new x3dna user and I would like to use the pdb file generated by fiber
as an input file for gromacs.  I tried to do what follows:

1) I used fiber to generate a file bdna.pdb (see attached 'bdna.pdb' file)

2) pdb2gmx  -f bdna.pdb -o output.gro (selecting interactively: amber99 force field and tip3p water)

but I get the following fatal error:

Fatal error:
Atom P in residue A 1 was not found in rtp entry RA5 with 31 atoms
while sorting atoms.

selecting different force fields or water models do not help,
what's wrong? Any idea?

thank you very much
for your help,
Cristiano
Title: Re: fiber and gromacs
Post by: xiangjun on June 18, 2012, 10:21:09 am
Hi Cristiano,

Welcome to join the 3DNA user community! Posting your question on the 3DNA forum is the right step to solve any 3DNA-related problems.

Regarding your issue of 3DNA fiber-generated PDB file, it is likely to be due to a 'special' (customized) PDB format adopted by Gromacs, based on the following error message:

Quote
Atom P in residue A 1 was not found in rtp entry RA5 with 31 atoms while sorting atoms.

It seems Gromacs gets stuck in the first residue -- it is expecting RA5 (presumably for adenine of RNA, on the 5' end?) while 3DNA provides simply "  A" for DNA adenine.

Please try as instructed below, report back what you get, and we will move on from there:

HTH

Xiang-Jun
Title: Re: fiber and gromacs
Post by: cridemichel on June 19, 2012, 05:13:35 am
Hi Xiang-Jun,
first of all thank you for the quick reply.
I tried to generate the sequence GCGCGC with the option -pdbv3 getting
the attached gzipped bdna.pdb file.
If I try to convert it to .gro format using pdb2gmx I get the following
error if I use any AMBER forcce field:

Fatal error:
Atom P in residue DG 1 was not found in rtp entry DG5 with 31 atoms
while sorting atoms.

but at least it works if I use CHARMM27 force field.

About the dickerson dodecam I always get the following fatal error
from pdb2gmx:

Fatal error:
Residue 'SPM' not found in residue topology database

regards
Cristiano




Quote from: xiangjun on June 18, 2012, 10:21:09 am
Hi Cristiano,

Welcome to join the 3DNA user community! Posting your question on the 3DNA forum is the right step to solve any 3DNA-related problems.

Regarding your issue of 3DNA fiber-generated PDB file, it is likely to be due to a 'special' (customized) PDB format adopted by Gromacs, based on the following error message:

Quote
Atom P in residue A 1 was not found in rtp entry RA5 with 31 atoms while sorting atoms.

It seems Gromacs gets stuck in the first residue -- it is expecting RA5 (presumably for adenine of RNA, on the 5' end?) while 3DNA provides simply "  A" for DNA adenine.

Please try as instructed below, report back what you get, and we will move on from there:
  • Download 355d (http://www.rcsb.org/pdb/explore/explore.do?structureId=355d), the classic Dickerson B-DNA dodecamer, and repeat your procedure.
  • Regenerate your fiber model with option -pdbv3 (to have residue names like " DA"),  and repeat your procedure.
  • Check for the documentation of the specifics of the Gromacs PDB format.

HTH

Xiang-Jun
Title: Re: fiber and gromacs
Post by: xiangjun on June 19, 2012, 09:01:46 am
Thanks for posting back your findings. I am glad to see that you are making progress!

Now I can reasonably guess what's happening:


HTH,

Xiang-Jun
Title: Re: fiber and gromacs
Post by: cridemichel on June 20, 2012, 04:49:21 am
Hi Xiang-Jun,
using the -pdbv3 option and removing the 5'-phosphate groups it works with AMBER99 forcefield also.
Nevertheless if I extract the two dna strands and I try to convert them, as you suggested, i.e. if I do:

> get_part 355d.pdb 355d-only-dna.pdb

> pdb2gmx -f 355d-only-dna.pdb -o prova.gro

I still get a fatal error in converting like the following:

Fatal error:
Atom C5M in residue DT 7 was not found in rtp entry DT with 32 atoms
while sorting atoms.

thx for your help,
Cristiano

Quote from: xiangjun on June 19, 2012, 09:01:46 am
Thanks for posting back your findings. I am glad to see that you are making progress!

Now I can reasonably guess what's happening:

  • The program pdb2gmx seems to follow PDB format v3. So it takes residue name such as "  A" as RNA instead of DNA (" DA"). That's why the -pdbv3 option helps.
  • Furthermore, for the HARMM27 force field, pdb2gmx does not like the 5'-phosphate group (atoms P, OP1, and OP2):
    ATOM      1  P    DG A   1      -0.356   9.218   1.848  1.00  1.00           P  
    ATOM      2  OP1  DG A   1      -0.311  10.489   2.605  1.00  1.00           O  
    ATOM      3  OP2  DG A   1      -1.334   9.156   0.740  1.00  1.00           O
    ------------------------------------------------------------------------------
    ATOM    124  P    DG B   7       0.356   9.218 -18.723  1.00  1.00           P  
    ATOM    125  OP1  DG B   7       0.311  10.489 -19.480  1.00  1.00           O  
    ATOM    126  OP2  DG B   7       1.334   9.156 -17.615  1.00  1.00           O 
    If you manually delete the two 5' phosphate fragments, I sense pdb2gmx should work for the AMBER forcce field.

    To verify the above line of thinking, extract only the two DNA chains in 355d as below: 
        get_part 355d.pdb 355d-only-dna.pdb
    Now pdb2gmx should be happy with file '355d-only-dna.pdb' for both AMBER and CHARMM force fields.
  • I performed a quick google search on pdb2gmx (http://manual.gromacs.org/current/online/pdb2gmx.html). It appears to me that users need to provide force-field specific residue types for "uncommon" cases. Dig deeper into Gromacs/pdb2gmx documentation, and post your question into the Gromacs mailing list -- gmx-users (http://www.gromacs.org/Support/Mailing_Lists) would help -- it is more of a Gromacs-related problem than 3DNA fiber-generated PDB files (with the -pdbv3 option).

HTH,

Xiang-Jun
Title: Re: fiber and gromacs
Post by: xiangjun on June 20, 2012, 07:22:58 am
Quote
> get_part 355d.pdb 355d-only-dna.pdb

> pdb2gmx -f 355d-only-dna.pdb -o prova.gro

I still get a fatal error in converting like the following:

Fatal error:
Atom C5M in residue DT 7 was not found in rtp entry DT with 32 atoms
while sorting atoms.
Oops, here I forgot to add the -pdbv3 option: instead of as " C7 ", the 5-methyl group of DT was labelled as " C5M" which pdb2gmx obviously does not like. Unless I am still missing something else, I believe the following should work:

Code: [Select]
get_part -pdbv3 355d.pdb 355d-only-dna.pdb
The last point in my previous reply still holds, i.e., more familiarity with the nuances of proper usage of Gromacs/pdb2gmx would certainly help. If you could post back your effort and progress in that aspect, it'd be great.

Xiang-Jun
Title: Re: fiber and gromacs
Post by: cridemichel on June 20, 2012, 10:29:49 am
Hi Xiang-Jun,
-pdbv3 option does the job and pdb2gmx works now
using the AMBER force field in converting the dickerson dodecamer.
About digging into gromacs docs I will keep you updated.
For now I believe that I will use -pdbv3 option removing
the phosphate group from the duplex,

thank you very much
for your valuable help,
Cristiano


Quote from: xiangjun on June 20, 2012, 07:22:58 am
Quote
> get_part 355d.pdb 355d-only-dna.pdb

> pdb2gmx -f 355d-only-dna.pdb -o prova.gro

I still get a fatal error in converting like the following:

Fatal error:
Atom C5M in residue DT 7 was not found in rtp entry DT with 32 atoms
while sorting atoms.
Oops, here I forgot to add the -pdbv3 option: instead of as " C7 ", the 5-methyl group of DT was labelled as " C5M" which pdb2gmx obviously does not like. Unless I am still missing something else, I believe the following should work:

Code: [Select]
get_part -pdbv3 355d.pdb 355d-only-dna.pdb
The last point in my previous reply still holds, i.e., more familiarity with the nuances of proper usage of Gromacs/pdb2gmx would certainly help. If you could post back your effort and progress in that aspect, it'd be great.

Xiang-Jun

Funded by the NIH R24GM153869 grant on X3DNA-DSSR, an NIGMS National Resource for Structural Bioinformatics of Nucleic Acids

Created and maintained by Dr. Xiang-Jun Lu, Department of Biological Sciences, Columbia University