Dear Dian,
> Thanks for your kind reply. I pretty much went through the RED tutorial but
> still have a few puzzles I hope you could help me with.
>
> Say, I have a peptide-like molecule which consists of four fragments,
> A-B-C-D. All the bonds connecting these four parts are peptide bonds except
> the bond between A and B. Here is what I will do. Please let me know if
> there is anything wrong.
This means you need to create 4 elementary building blocks, which are
in fact 4 small/whole molecules: this means the 4 fragments A, B C & D
have to be transformed into 4 elementary building blocks/whole
molecules such as: A-x1, x2-B-x3, x4-C-x5 and x6-D; x1-x6 are
connecting groups, which will be involved in charge constraints and
removed after charge fitting.
a fragment does not exist in QM - if you provide fragment A to
Gausssian/Firefly/GAMESS for geometry optimization (or MEP
computation) the program will crash. On the contrary, the QM program
will recognize the whole moelcule A-x1; x1 can be removed and the
empirical fragment A can be designed for an empirical force field by
using a charge constraint during the empirical charge fitting step.
> The first thing is to break it up to four pieces A B C and D
you need to create 4 whole molecules A-x1, x2-B-x3, x4-C-x5 & x6-D
> 1. A is like a alkane chain. Doesn't exist in AMBER library. So I have to
> treat it as a brandnew molecule. I put a CH3 group at the "C-term" (right
> side terminal). Save A-CH3 into A.pdb and follow the RED steps to get the
> parameters.
Yes however - let's say "fragment A" contains 10 carbon atoms:
CH3(CH2)9 this means whole "molecule A-x1" will contain 11 carbons
CH3(CH2)9CH3 and an intra-mcc = set 0 (with the R flag) can be used
during the fitting step to create fragment A. Then, x1 is also
designed based on the fragment B, which will be connected to A.
> 2. B and D are Asp and Ser respectively. As you said, these two can be
> found in Amber, so I don't have to parameterize them.
Yes
> However, how are the
> atoms of these fragments recognized as known amino acids?
You load the Amber force field topology database (FFTopDB; set of
force field libraries) in LEaP; to be recognized the atom & residue
names in the PDB file (experimental data of the entire molecule) have
to match these in the FFTopDB.
> Do I have to
> change the atom names from simple single letter to standard protein atom
> names, e.g. CA, OE1, etc and add the column of residue names in the
> seperate pdb file?
- if the atom & residue names in the PDB file are different and if you
want a match with the FF libs; Yes
- if the atom & residue names in the PDB file are different and if you
do not want a match with the FF libs; No - but in this case you have
to create a new FF lib.
> 3. C is an unknown molecule. I put the ACE and NME caps on each sides and
> save it as C.pdb and do the RED to get the mol2 file.
ok
> Now here are some questions:
>
> (1) After the steps above, I only end up with two mol2 files (A and C).
> What do I do with B and D?
B & D are contained in the Amber FFTopDB
> If somehow I can derive the mol2 files for B and
> D as well from AMBER existing force field database, how do I combine them
> to generate a single prmtop for the entire molecule?
- you load the Amber FFTopDB
- you load your new FF libs
- you load the PDB file of your entire molecule
the match will append based on the rules defined above.
> What do I do with the
> connections between fragments?
if you correctly defined the connecting atom between units (i.e. the
head & the tail), usually you do nothing ;-) the connection between
two units with head/tail defined is automatically performed by LEaP
See
http://q4md-forcefieldtools.org/Tutorial/leap-mol3.php
> (2) Why does RED need multiple conformations to get the charges?
because RESP & ESP charge values depends on conformation
they also depends (to a minor extend) on molecular orientation
> Can I just
> start with one structure for every individual fragment?
fragment? see above: elementary building block i.e. a small/whole molecule.
in general a building block/whole molecule with a well defined
conformation is always involved in charge derivation. which
conformation(s) to be involved is a key problem: in general the lowest
minimum/a are selected; experimental like conformation(s) can also be
selected.
by using 2 conformations in charge fitting a single set of charge
values is derived 'reflecting' the 2 conformations selected: this is
the main idea; here the charge values should be more general and
suitable for MD simulation when using an empirical force field with a
non-polarizable/fixed charge model...
> (3) If it is all about getting mol2 files, can I just break the molecule up
> in the same way (also add the caps and save as 4 seperate pdbs) and create
> the mol2 files with semi-em method (Antechamber), although it might not be
> as accurate as RESP charges. And then the problem again becomes how to
> stitch them up to create a complete prmtop.
You need to start from X whole molecules to create X fragments and
constraints are not correctly applied when using Antechamber (nmol
=1); from what I know Antechamber does not deal with the building
block approach; charge equivalencing is not correctly performed so
far; I just learned they are also pseudo-rounding off errors, etc...
Is it my role to report these problems? I do not think so...
> (4) I know there are many ways to add the ACE/NME caps and save it as a 3-D
> pdb structure. Do you have any recommendations of what software to use?
- Use Java applet in R.E.D. Server
See
http://q4md-forcefieldtools.org/REDS/faq.php#20
- Use xLEaP and the "sequence" command...
See
http://q4md-forcefieldtools.org/Tutorial/Tutorial-1.php#21
regards, Francois
> On Fri, Feb 3, 2012 at 1:04 AM, FyD <fyd.q4md-forcefieldtools.org> wrote:
>
>> Dear Dian,
>>
>>
>> I was following the tutorial to run RED to get the charges for my
>>> molecule.
>>> It takes for ever for the gaussian calculation to finish since it is a
>>> huge
>>> molecule (>250 atoms).
>>>
>>
>> I did NOT tell you to compute your entire molecule by using Gaussian
>> instead of divcon/mopac/sqm ;-) similar problems should append in your case
>> even if I prefer to use QM than semi-emp.
>>
>>
>> Is there any short cut to do this? I mean in
>>> principle Antechamber should do it, right?
>>>
>>
>> I do NOT support the Antechamber approach.
>>
>>
>> Although it didn't generate a
>>> mol2 file, but the sqm.out file did have the charge information for each
>>> atom (Mulliken charges). Maybe I am wrong, but can I manual create a mol2
>>> based on the charges and other bonded terms from ANTECHAMBER_BOND_TYPE? Or
>>> can I just run the charge calculation with fragment of the molecule, which
>>> is the center of the peptide for my case, get the parameters of the rest
>>> and then combine these two parts?
>>>
>>
>> I think I told you all what you have to do in my former email. I am sorry
>> of this is not clear. Let's take an example:
>> Imagine you have the following 5 residue long peptide with one modified
>> amino-acid, AA*:
>>
>> 1 2 3 4 5
>> Gly-Asp-AA*-Lys-Gly
>>
>> All peptide bonds should be 'trans' except if you have a good reason
>> (proline, ..., ???)
>>
>> -1) Concerning 1 & 5:
>> . you need terminal fragments for 1 & 5: N-term. and C-term.
>> . do 1 & 5 bear a charged group? 'Yes' at pH = 7 (the MD conditions)
>> . do you need to parameterize 1 & 5? 'No' they are already in the Amber
>> force field topology database (FFTopDB)
>>
>> -2) Concerning 2 & 4:
>> . you need central fragments for 2 & 4
>> . do 2 & 4 bear a charged group? 'Yes' at pH = 7
>> . do you need to parameterize 2 & 4? 'No' they are already the Amber
>> FFTopDB
>>
>> -3) Concerning 3:
>> . you need a central fragment for 3
>> . does 3 bear a charged group at pH = 7? well, you have to find out ;-)
>> . do you need to parameterize 3?
>> -> check in the literature
>> -> if unknown: 'Yes' see tutorials at q4md-forcefieldtools.org
>>
>> http://q4md-forcefieldtools.**org/Tutorial/<http://q4md-forcefieldtools.org/Tutorial/>
>> For instance:
>>
>> http://q4md-forcefieldtools.**org/Tutorial/Tutorial-3.php#24<http://q4md-forcefieldtools.org/Tutorial/Tutorial-3.php#24>
>> vs
>>
>> http://q4md-forcefieldtools.**org/Tutorial/Tutorial-3.php#25<http://q4md-forcefieldtools.org/Tutorial/Tutorial-3.php#25>
>>
>> I hope this helps...
>>
>> regards, Francois
>>
>>
>>
>> On Tue, Jan 31, 2012 at 11:17 PM, FyD <fyd.q4md-forcefieldtools.org>
>>> wrote:
>>>
>>> Dear Dian,
>>>>
>>>>
>>>> Maybe I should not call it peptide in the first place. It is a fairly
>>>> big
>>>>
>>>>> molecule (>250 atoms) with some parts resemble amino acids like Glu,
>>>>> Gln,
>>>>> Ser, Arg etc. That's why I treated it as a big ligand. How would you
>>>>> parameterize a molecule like this in Amber? Could you kindly take a look
>>>>> at
>>>>> my pdb file? Many thanks.
>>>>>
>>>>>
>>>> You need to:
>>>> - load the Amber Force Field Topology Database (FFTopDB) to look at which
>>>> amino-acid residues are already available (i.e. the .off or .lib files).
>>>> - check the peptide bonds in your peptide: do you really want cis peptide
>>>> bonds
>>>> - check the charged groups of your amino-acids in your peptide: for
>>>> instance, do you really want a COOH group instead of COO(-); at pH 7 the
>>>> carboxylate group is present
>>>> - among the amino-acid residues in your peptide isolate these that are
>>>> not
>>>> in the Amber FFTopDB: you will need to generate a new force field library
>>>> for these modified/unknown residues (check in the literature if your
>>>> modified/unknown residue is not already known/parameterized; search in
>>>> the
>>>> Amber and R.E.DD.B. databases as well).
>>>>
>>>> Then to learn how to create the central, N-terminal & C-terminal
>>>> fragments
>>>> for a modified/unknown residue, read:
>>>> http://onlinelibrary.wiley.****com/doi/10.1002/jcc.540161106/**
>>>> **abstract<http://**onlinelibrary.wiley.com/doi/**
>>>> 10.1002/jcc.540161106/abstract<http://onlinelibrary.wiley.com/doi/10.1002/jcc.540161106/abstract>
>>>> **>
>>>>
>>>>
>>>> See also tutorials at q4md-forcefieldtools.org
>>>> http://q4md-forcefieldtools.****org/Tutorial/Tutorial-1.php#**10<
>>>> http://q4md-**forcefieldtools.org/Tutorial/**Tutorial-1.php#10<http://q4md-forcefieldtools.org/Tutorial/Tutorial-1.php#10>
>>>> >
>>>> http://q4md-forcefieldtools.****org/Tutorial/Tutorial-3.php#**24<
>>>> http://q4md-**forcefieldtools.org/Tutorial/**Tutorial-3.php#24<http://q4md-forcefieldtools.org/Tutorial/Tutorial-3.php#24>
>>>> >
>>>> http://q4md-forcefieldtools.****org/Tutorial/Tutorial-3.php#**25<
>>>> http://q4md-**forcefieldtools.org/Tutorial/**Tutorial-3.php#25<http://q4md-forcefieldtools.org/Tutorial/Tutorial-3.php#25>
>>>> >
>>>>
>>>>
>>>> regards, Francois
>>>>
>>>>
>>>>
>>>> On Tue, Jan 31, 2012 at 1:06 AM, FyD <fyd.q4md-forcefieldtools.org>
>>>>
>>>>> wrote:
>>>>>
>>>>> Dear Dian Jiao,
>>>>>
>>>>>>
>>>>>> You should look at the output of the semi-empirical tool used
>>>>>> (divcon/mopac/sqm/...): It looks like the job failed... My
>>>>>> understanding is that a job performed with a semi-empirical method
>>>>>> should be fast (in particular, if the optimization threshold is loose).
>>>>>>
>>>>>> You reported "pep.pdb": does it mean an entire peptide is involved in
>>>>>> the computation? in general, one wants to generate a new force field
>>>>>> library only for a modified residue; the regular residues are handled
>>>>>> by the Amber Force Filed Topology DataBase...
>>>>>>
>>>>>> regards, Francois
>>>>>>
>>>>>>
>>>>>> > I was trying to model this peptide with the following command:
>>>>>> >
>>>>>> > antechamber -i pep.pdb -fi pdb -o pep.mol2 -fo mol2 -c bcc.
>>>>>> >
>>>>>> > After about 24 hours, the job "finished" except that it didn't
>>>>>> generate
>>>>>> > pep.mol2 file. All other output files seem okay, including
>>>>>> > ANTECHAMBER_BOND_TYPE.AC0, ANTECHAMBER_AM1BCC_PRE.AC, sqm.in and
>>>>>> sqm.out.
>>>>>> > So what happened to mol2? Was the file written to somewhere else?
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Feb 10 2012 - 00:30:03 PST
