Constraint warning with AMBER PDB file

[Lekpa Duukori]

Hi all,

I have been getting this warning  when I try to run a calculation using PDB and topology files generated by AMBER.
The idea was to apply a constraint to the HBONDS

 *** 12:36:25 WARNING in topology_constraint_util:print_warning_molname     ***
 *** err=-300   MOLNAME (MOL) was defined for constraints, but this         ***
 *** molecule name is not defined. Please check carefully your PDB, PSF     ***
 *** (has priority over PDB) or input driven CP2K coordinates. In case you  ***
 *** may not find the reason for this warning it may be a good idea to      ***
 *** print all molecule information (including kind name) activating the    ***
 *** print_key MOLECULES specific of the SUBSYS%PRINT section.              ***
 *** /opt/cp2k/makefiles/../src/topology_constraint_util.F line   1374 

Using the print key it talks about prints gives molecule names like MOL0 etc, all other parameters appear to be read correctly though.

The cp2k executable I am using is about a day old.

Inputs has been uploaded to the site

Any ideas is appreciated, thanks.

Lekpa

PS: this is s test that reproduces an error I am getting with different input files

[Laino Teodoro]

Dear Lepka,

unfortunately AMBER topology files do not have any information about
molecule names. What is stored there is the residue name, which we
don't really use within CP2K.

As you realized CP2K is assigning new molname: MOL01..
You can either specify the whole list within the appropriate molname
keyword or alternatively just use the other keyword MOLECULE in this
way:

MOLECULE 1..15

I have no clue if you really have 15 molecules in your input or just 2.
This should fix your problem.
Teo

p.s.: just a comment: please upload 1 unique tar (gzipped) file: this
makes life easier.

> --
> You received this message because you are subscribed to the Google
> Groups "cp2k" group.
> To post to this group, send email to cp...@googlegroups.com.
> To unsubscribe from this group, send email to cp2k
> +unsub...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/
> group/cp2k?hl=en.

[yimanu]

Thanks!

[marco.stenta]

Dear Teodoro,
I'm running QM/MM cp2k on a pretty big system.
I'm using amber topology so I need a way to SHAKE the water molecules
I understand I do it in the &MOTION section with something like that, using molecule as, without a psf I cannot specify the molname

  &CONSTRAINT
    &G3X3
      MOLECULE                     31
      DISTANCES                    1.7958  1.7958  2.85444
      ATOMS                        1 2 3
      EXCLUDE_QM
    &END G3X3
  &END CONSTRAINT

Here 31 is the molecule kind number  of the fist water molecule as guessed from the output of a tentative run with verbose output: I get the molecule kind name is MOL30 and the molecule kind number is 31: is it right?

    31. Molecule kind: MOL30                     Number of atoms:              3
                     Atom         Atomic kind name
                        1                       OW
                        2                       HW
                        3                       HW

        The name was automatically generated: T
        Number of molecules:      1
        Molecule list:               31
        Number of bonds:            3



The problem is that I have 47883 water molecules: do I have to repeat this CONSTRAINT  section (or the &G3X3 subsection ) 50 thousand time? the input would be pretty big.
Is there a clever way to SHAKE water molecules in my case?

you write in your previous answer
MOLECULE 1..15

but from the manual I understand MOLECULE takes only one integer: what does your line mean? can I specify for a range instead then for a single molecule?

Is it possible to have the code recognizing a MOLNAME also when one uses an amber topology?
I am forced to use amber topology as I did not find the ambertocharmm converter (leap2fist ?) anymore and I have to shake water, of course.

More in general: what do you mean by "molecule"?
each residue of a protein is assigned to a "molecule": what is the  meaning of that?
TOPOLOGY > PARA_RES is set to TRUE
what if I set it to FALSE?
I get the term molecule has nothing to do with its chemical meaning, but my protein part is composed by more than one sub-units: does that conflicts with PARA_RES FALSE?
Then I get some strange molecules as well like
     1. Molecule kind: _QM_MOL0                  Number of atoms:          10657
this is not the QM part, nor the QM + link and I am lost

thanks for the help
m

[Teodoro Laino]

On Jun 22, 2011, at 8:44 AM, marco.stenta wrote:

The problem is that I have 47883 water molecules: do I have to repeat this CONSTRAINT  section (or the &G3X3 subsection ) 50 thousand time? the input would be pretty big.
Is there a clever way to SHAKE water molecules in my case?

yes (see below)

you write in your previous answer
MOLECULE 1..15

but from the manual I understand MOLECULE takes only one integer: what does your line mean? can I specify for a range instead then for a single molecule?

that's right..  I'll see how tough it is to convert this keyword to a list..

Is it possible to have the code recognizing a MOLNAME also when one uses an amber topology?

not easily, because AMBER topology has to be heavily digested unfortunately and molnames have to be changed.. Clearly one can always hack this procedure, but it is not my intention do so at the moment..

I am forced to use amber topology as I did not find the ambertocharmm converter (leap2fist ?) anymore and I have to shake water, of course.

I decided to embed the leap2fist in cp2k: less maintenance.. you can still dump the PSF and PDB after digestion and use them to re-run cp2k with the proper styles..

At this stage you can decide to give the name you like more to the residues in the PSF, and then use just 1 constraint for WATER (assuming all water have the molname WATER).

More in general: what do you mean by "molecule"?
each residue of a protein is assigned to a "molecule": what is the  meaning of that?
TOPOLOGY > PARA_RES is set to TRUE
what if I set it to FALSE?

PARA_RES distributes the residues instead of molecules.

I get the term molecule has nothing to do with its chemical meaning, but my protein part is composed by more than one sub-units: does that conflicts with PARA_RES FALSE?
Then I get some strange molecules as well like
     1. Molecule kind: _QM_MOL0                  Number of atoms:          10657
this is not the QM part, nor the QM + link and I am lost

no input.. no idea..

cheers,

Teo

thanks for the help
m

--
You received this message because you are subscribed to the Google Groups "cp2k" group.

To view this discussion on the web visit https://groups.google.com/d/msg/cp2k/-/YYO4fTrQUIwJ.

To post to this group, send email to cp...@googlegroups.com.

To unsubscribe from this group, send email to cp2k+uns...@googlegroups.com.

[marco.stenta]

Dear Teodoro,
it worked:
1) first fake run to dump psf and pdb
2) editing the psf as to change the segment name of water molecule
3) good run reading psf/pdb instead of top/crd (TOPOLOGY section)
Thanks again
m

[Chuong]

Hello,

Sorry to bump in an old thread but how do you do a "fake run" to dump a psf and pdb file from the AMBER top/crd files? I also have need to convert amber topology to charmm topology for some reasons.

Yours,

Chuong