Re: Re: Problems to parametrize molecules

From: Evandro Semighini (epsemighini_at_gmail.com)
Date: Fri Apr 24 2015 - 13:54:33 CDT

Hello Brian,

A HUGE thank you for this instructions and for the help, I'll work on it
now.

Have a nice weekend !

Evandro

2015-04-24 15:46 GMT-03:00 Bennion, Brian <bennion1_at_llnl.gov>:

> Hello Evandro,
>
> These are the steps that I followed.
> 1. I took your mol2 file and submitted it to cgenff.paramchem.org.
> Amazingly the server returned an str file with extremely low penalties for
> nearly all the atoms.
> There are a few that will need some work (more on that later).
> 2. I converted all the UNK and b_list names to EVA in the MOL2 file and
> STR file.
> 3. I converted the mol2 file to a pdb with babel -imol2 b_list.mol2 -opdb
> b_list.pdb
> 4. I loaded this pdb file in to vmd192
> 5. I used the molfacture plugin to add the correct atom type (very tedious
> but critical).
> 6. I saved the psf and pdb and top files from molefacture.
> 7. Now you can use FFTK to validate the very few missing parameters. I
> would recommend doing this by reloading the psf and pdb files in to vmd and
> using molefacture to cut the pieces out to make the calculations smaller.
> That is cut out the biphenyl and piperazine portions of your molecule to
> validate the charges and angle that is noted in the STR file.
> Now unless you have access to gaussian you will be stopped at this point
> and will need to judge whether the penalties are such you can live with
> them (and the reviewers as well) or find access to gaussian by paying for
> it or collaborating with someone.
>
> Have a good weekend.
> Brian
>
>
> ------------------------------
> *From:* Evandro Semighini [epsemighini_at_gmail.com]
> *Sent:* Friday, April 24, 2015 10:49 AM
> *To:* Bennion, Brian
> *Cc:* namd-l_at_ks.uiuc.edu; Mayne, Christopher G
>
> *Subject:* Re: namd-l: Re: Problems to parametrize molecules
>
> Hello Brian,
>
> The molecule is attached to the e-mail, in mol2 format.
> .
>
> Thank you all !
>
> Evandro
>
> 2015-04-24 14:27 GMT-03:00 Bennion, Brian <bennion1_at_llnl.gov>:
>
>> Hello Evandro,
>>
>>
>>
>> Your description below is still missing critical information and blending
>> steps together without any stated rationale.
>>
>> At this point it would be best if you posted your initial molecule
>> structure file so others could check it and reproduce your errors.
>>
>>
>>
>> If this is not acceptable for whatever reason lets talk off-line.
>>
>> Brian
>>
>>
>>
>> *From:* owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] *On
>> Behalf Of *Evandro Semighini
>> *Sent:* Friday, April 24, 2015 8:05 AM
>> *To:* Mayne, Christopher G
>> *Cc:* <namd-l_at_ks.uiuc.edu>
>> *Subject:* Re: namd-l: Re: Problems to parametrize molecules
>>
>>
>>
>> Hello Christopher,
>>
>> I did what you said in the first message, with another molecule, and I
>> got the message of "unknown residue" when I try to create the psf file with
>> psfgen, even with the str file from ParamChem, as this unknown residue is
>> the whole molecule.
>>
>> I looked to it again and the residue name is the same at the pdb and the
>> mol2 file, UNK.
>>
>> I only managed to run psfgen dividing the molecule into parts like a
>> benzene ring, a phenyl methyl ketone system and so on, taking it's
>> parameters from the cgenff rtf file and modifying them, removing the atoms
>> where is another part from the original molecule, but when I try to
>> parametrize more than one part toghether in the same molecule, it only
>> parametrizes the first part, as it recozniges it as the only residue there,
>> and the bonds are completelly messed.
>>
>> So, I changed the rtf file to this (benzene):
>>
>> RESI BENZ 0.00 ! C6H6 benzene, adm jr.
>> GROUP
>> ATOM CG CG2R61 -0.115
>> ATOM CZ CG2R61 0.000 ! HD1 HE1
>> ATOM CD1 CG2R61 -0.115 ! | |
>> ATOM HD1 HGR61 0.115 ! CD1--CE1
>> ATOM CD2 CG2R61 -0.115 ! / \
>> ATOM HD2 HGR61 0.115 ! *-CG CZ-*
>> ATOM CE1 CG2R61 -0.115 ! \ /
>> ATOM HE1 HGR61 0.115 ! CD2--CE2
>> ATOM CE2 CG2R61 -0.115 ! | |
>> ATOM HE2 HGR61 0.115 ! HD2 HE2
>>
>> BOND CD1 CG CD2 CG CE1 CD1
>> BOND CE2 CD2 CZ CE1 CZ CE2
>> BOND CD1 HD1 CD2 HD2 CE1 HE1
>> BOND CE2 HE2
>>
>> IC CG CD1 CE1 CZ 0.0000 0.0000 0.0000 0.0000 0.0000
>> IC CD1 CE1 CZ CE2 0.0000 0.0000 0.0000 0.0000 0.0000
>> IC CE1 CZ CE2 CD2 0.0000 0.0000 0.0000 0.0000 0.0000
>> IC CE1 CG *CD1 HD1 0.0000 0.0000 180.0000 0.0000 0.0000
>> IC CE2 CG *CD2 HD2 0.0000 0.0000 180.0000 0.0000 0.0000
>> IC CZ CD1 *CE1 HE1 0.0000 0.0000 180.0000 0.0000 0.0000
>> IC CZ CD2 *CE2 HE2 0.0000 0.0000 180.0000 0.0000 0.0000
>>
>>
>>
>> psfgen runned, resulting in 64 atoms, but, it has a lot of errors like
>> this:
>>
>> psfgen) ERROR: Missing atoms for bond CE2(0) HE2\PAR(0) in residue BENZ:2
>>
>> psfgen) Warning: failed to set coordinate for atom C10 BENZ:2
>> psfgen) no segment
>>
>> All over the molecule, and all the bonds are still wrong.
>>
>> I checked the str, pdb and mol2 files, they are ok.
>>
>>
>>
>> Any clue of where I am missing this process ?
>>
>> I will have to do this customized parameters for every molecule I'm
>> working with ?
>>
>>
>>
>> Thank you,
>>
>> Evandro
>>
>>
>>
>>
>>
>> 2015-04-19 20:59 GMT-03:00 Mayne, Christopher G <cmayne2_at_illinois.edu>:
>>
>> Evandro,
>>
>>
>>
>> Small molecules are typically treated as single residues; certainly
>> anything returned by ParamChem is treated as a single residue. It is
>> critical that the information in your PDB, i.e., resname, atom names, etc.,
>> matches that of your MOL2 file, since this is what ParamChem uses to build
>> the topology. You should specifically check the resname, as ParamChem may
>> assign UNK if the resname in the MOL2 isn't assigned or valid.
>>
>>
>>
>> Regards,
>>
>> Christopher Mayne
>>
>>
>>
>
>

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:21:50 CST