AMBER Archive (2007)

Subject: Re: AMBER: antechamber fails with large molecules

From: Francesco Pietra (chiendarret_at_yahoo.com)
Date: Sat Nov 03 2007 - 08:01:41 CST


Hi Ilyas:
That molecule of 500 atoms was a bold starting at a time that I knew even less
than I know now. I mentioned that in the context of the new respgen.c only to
remind my interest. Set aside. There is no theory to deal computationally with
that.

What you are saying (and other subscribers said in parallel interventions) does
not apply to organic chemistry/natural product chemistry, except perhaps in
rare out-of-the-norm cases, if any. At any event, I am not aware of any single
example of treating a complicated organic molecule as residues to be recomposed
later. The reason is simple: any portion is generally highly flexible - which
defeats any choice of residues - and any portion you can conceive interacts
strongly with neighboring portions. No repetitive units, very scarce similarity
with units in other families of compounds, and a behavior that depends on large
sections of the molecule, if not, in many cases, on the whole. That is
certainly the case of my 500-atoms molecule, and even of the other smaller
members of the family.

This is why computational medicinal chemists do docking for very small - or
very rigid - ligands only. Nonetheless, nature is able to dock specifically
with molecules as alrge and as flexible as above.

Perhaps one could calculate partial charges rapidly with antechamber (not for
500-atoms molecule, say 150-atoms) carry out a simulated annealing and/or a
montecarlo sampling and then refining by calculating the partial charges for a
wide variety of low-energy conformers. Finally, averaging.

regards
francesco

--- Ilyas Yildirim <yildirim_at_pas.rochester.edu> wrote:

> In these papers, they deal with POPC molecule and calculate the resp
> charges for POPC so that they can form a 128 POPC bilayer structure. They
> choose 12 conformations for POPC and they show that with this choice, they
> can predict some of the experimental results. Whether 12 is enough or not,
> I do not know. I am sure they have their reasons on why they have chosen
> 12 conformations for POPC.
>
> It is, still, following the resp protocol, though I am not sure how they
> followed the resp protocol. POPC looks more complicated than an RNA
> nucleotide, but it is still possible to follow the resp protocol to
> calculate the charges. After the resp charge calculation for POPC, they
> create the molecule which has 128 POPC in it.
>
> So, for a 500 atom system, can you calculate the resp charges? Quantum
> Mechanical optimization can take forever for this system. If you want to
> stick with fixed charge method, you have to divide your structure into sub
> structures. Dr. SantaLucia's group created a database for modified RNA
> nucleotides.
>
> http://ozone3.chem.wayne.edu:8080/Modifieds/jsps/move.jsp#
>
> In their paper (Aduri, R., Psciuk, B. T., Saro, P., Taniga, H., Schlegel,
> H. B., SantaLucia, J., Jr. (2007) "AMBER Force Field Parameters for the
> Naturally Occurring Modified Nucleosides in RNA", J. Chem. Theory and
> Comput., 3, 1464 -1475.), they explain how they calculated the resp
> charges. For some big complicated nucleotides, they made some
> approximations such as diving that particular nucleotide into 2 parts; one
> is the sugar part and the other one is the base part. They optimize and
> calculate the resp charges of these two parts seperately. At the end, they
> combine these two parts to create the library file for that particular
> complicate nucleotide.
>
> So, even though you cannot calculate the resp charges for a 500 atom
> system directly, you can make approximation by diving the structure into
> small segments and calculate the resp charges for those segments
> seperately. At the end, you can combine the resp charges to create your
> library file for 500 atom system. How to divide your structure into
> sub segments/structures depend on how your structure looks like. The most
> important point of these sub segments should be that they should not
> interact with each other crucially (at least the interaction should not be
> too big). For instance, in an RNA nucleotide, you can divide the
> nucleotide into sugar + base because the interaction between these two is
> not too big (at least that is what the approximation is in some of the
> resp calculations in the above database).
>
> By the way, when you use antechamber to get the charges, they won't be
> the resp charges.
>
> Best,
>
> On Fri, 2 Nov 2007, Francesco Pietra wrote:
>
> > Hopefully polarizable force fields will put - they alone - the house in
> order.
> > I doubt, simply because it seems too good.
> >
> > People using the fixed charge strategy have long devised to mediate between
> > conformers. A very careful piece of work to this regard came recently from
> > Hungary, concerning the POPC lipidic double layer (J Comput Chem 2007, 28,
> > 2051-8). Is such type of huge work absolutely needed routinely? Other
> groups -
> > in the context of hugh-level publication - still treat the same POPC with
> fixed
> > charges for a single conformer (JACS 2007, 129, 6970-1). Surely, it depends
> on
> > the precision to attain. My conformers, plagued by huge degrees of freedom
> and
> > differing in energy by 1 kcal/mol, can't be treated computationally in
> > condensed phase at any level of theory.
> >
> > francesco
> >
> >
> > --- Ilyas Yildirim <yildirim_at_pas.rochester.edu> wrote:
> >
> > > As I wrote in my previous email, polarizable force fields are still under
> > > development. In amber99 force field, there is a parameter defined
> > > as polarizability, but this is for Mg2+ and F- ions. I am not sure how
> the
> > > polarizability is implemented to AMBER 9, but I am not aware of a fully
> > > constructed polarizable force field. Others might know more about this.
> > >
> > > Best,
> > >
> > > On Fri, 2 Nov 2007, Eddie Men wrote:
> > >
> > > > This is a good question (or one I also have):
> > > >
> > > > "A better solution might be to use a polarizable force field, which
> will
> > > > change the charges of the atoms by time, depending on the neighbouring
> > > > atoms they are interacting. Polarizable force fields are still in
> > > > development stage, and I am not aware of any fully working polarizable
> > > > force field."
> > > >
> > > > I tried a model with resp charges for ATP braking it into ADP and PO3,
> > > > however when computing free energies the ATP breaks spontaneously at
> > > > very low temperature ( because both molecules are negative charged),
> > > > so I changed to a QM/MM description.
> > > >
> > > > Is it possible to model that reaction with polarizable forece fields?.
> > > > What about mixing force fields?.
> > > >
> > > >
> > > > Eddie
> > > >
> > > >
> > > > Best,
> > > >
> > > >
> > > > Ilyas Yildirim wrote:
> > > > > I know this question was discussed before, but I, also, do not
> > > understand
> > > > > the reason to calculate the resp charges for a system like a molecule
> > > > > which has 500 atoms in it. My understanding of charges and its
> importance
> > > > > in a structure is as follows:
> > > > >
> > > > > Some charge distrubution (point charges in this case) has to be
> chosen
> > > > > such that it is going to mimic the Q.M. potential around the
> molecule.
> > > > > Currently, constant charges are used in molecules, meaning the
> charges
> > > > > do not change by simulation time (except for some polarizable force
> field
> > > > > implementations, which is still new). The critical question is this:
> For
> > > > > this molecule, is there only one conformation or more than one? If it
> is
> > > > > a small molecule, probably it will choose one particular
> conformation,
> > > > > which implies that one charge distribution might be reasonable enough
> to
> > > > > mimic the quantum mechanical potential around this molecule.
> > > > >
> > > > > If the structure is too big, then it will probably have more than one
> > > > > conformation to choose. Without knowing the conformations this
> molecule
> > > > > can choose, a constant charge distribution model will never work. The
> > > best
> > > > > thing to do (if someone wants to use constant charge model) is to
> divide
> > > > > this molecule into sub-molecules (like residues) and then to
> calculate
> > > the
> > > > > charges for these sub-molecules (this is still an approximation
> though).
> > > > >
> > > > > A better solution might be to use a polarizable force field, which
> will
> > > > > change the charges of the atoms by time, depending on the
> neighbouring
> > > > > atoms they are interacting. Polarizable force fields are still in
> > > > > development stage, and I am not aware of any fully working
> polarizable
> > > > > force field.
> > > > >
> > > > > Best,
> > > > >
> > > > > On Thu, 1 Nov 2007, Junmei Wang wrote:
> > > > >
> > > > >
> > > > >> Personally I don't think there is a good approach to get good
> charges
> > > (such as HF/6-31G* RESP). AM1BCC may be a good candidate. The problem is
> that
> > > antechamber may not handle such large molecules well. In amber10, I have
> > > improved the code a little bit to handle large molecules much better, in
> a
> > > long run I also plan to develop a even simpler, but more reliable charges
> > > (compared to am1-bcc) for arbitrary organic molecules. This method will
> be
> > > available in amber11.
> > > > >>
> > > > >> Best
> > > > >>
> > > > >> Junmei
> > > > >>
> > > > >> ----- Original Message ----
> > > > >> From: Francesco Pietra <chiendarret_at_yahoo.com>
> > > > >> To: amber_at_scripps.edu
> > > > >> Sent: Thursday, November 1, 2007 3:29:26 AM
> > > > >> Subject: Re: AMBER: antechamber fails with large molecules
> > > > >>
> > > > >> You may remeber that I tried unsuccessfully to get the partial
> charges
> > > with
> > > > >> Antechamber in Amber9 for a 500-atoms non-repetitive molecule. Prof
> > > Case, and
> > > > >> others, suggested, however, that, even if calculated, the partial
> > > charges would
> > > > >> be unreliable for such a large residue.
> > > > >>
> > > > >> Is the new code intended to solve this question too?
> > > > >>
> > > > >> At any event, do you plan to make the new code available? An when? I
> > > have
> > > > >> frequently problems of large ligands.
> > > > >>
> > > > >> Thanks
> > > > >>
> > > > >> francesco pietra
> > > > >>
> > > > >>
> > > > >> --- Junmei Wang <junmwang_at_yahoo.com> wrote:
> > > > >>
> > > > >>
> > > > >>> Could you send me the file? I have modified the respgen.c code a
> > > little bit
> > > > >>> to handle large molecules, but it is not in amber9.
> > > > >>>
> > > > >>> Best
> > > > >>>
> > > > >>> Junmei
> > > > >>>
> > > > >>> Dear amber users:
> > > > >>>
> > > > >>> I am trying fit RESP charges with antechamber to a large
> > > > >>> organic molecule (342 atoms) with the command:
> > > > >>>
> > > > >>> antechamber –i biscalix.log –fi gout –o biscalix_resp.mol2
> > > > >>> -fo mol2 –c resp
> > > > >>>
> > > > >>> This fitting process failed with the following errors:
> > > > >>>
> > > > >>> “The number of the path atoms exceeds
> > > > >>> MAXPATHATOMNUM(1900000) for atom[148],extend the size and
> > > > >>> reallocate the memory automatically reallocate memory for
> > > > >>> pathscore[148] failed
> > > > >>> Info: the atom number exceeds the MAXATOM, reallocate
> > > > >>> memory automatically
> > > > >>> Amber 9 RESP
> > > > >>> Unit 5 Error on OPEN: ANTECHAMBER_RESP1.IN
> > > > >>> Error: cannot run "resp -O -i ANTECHAMBER_RESP1.IN -o
> > > > >>> ANTECHAMBER_RESP1.OUT -e ANTECHAMBER.ESP -t qout" in
> > > > >>> resp() of charge.c properly, exit”.
> > > > >>>
> > > > >>> Could someone give me any idea about how to solve this
> > > > >>> type of problems?
> > > > >>>
> > > > >>> Thanks in advance,
> > > > >>> Vânia Calisto
> > > > >>>
> > > > >>>
> > > > >>>
> > > > >>>
> > > > >>> __________________________________________________
> > > > >>> Do You Yahoo!?
> > > > >>> Tired of spam? Yahoo! Mail has the best spam protection around
> > > > >>> http://mail.yahoo.com
> > > > >>>
> > > > >> __________________________________________________
> > > > >> Do You Yahoo!?
> > > > >> Tired of spam? Yahoo! Mail has the best spam protection around
> > > > >> http://mail.yahoo.com
> > > > >>
> -----------------------------------------------------------------------
> > > > >> The AMBER Mail Reflector
> > > > >> To post, send mail to amber_at_scripps.edu
> > > > >> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> > > > >>
> > > > >>
> > > > >>
> > > > >>
> > > > >>
> > > > >> __________________________________________________
> > > > >> Do You Yahoo!?
> > > > >> Tired of spam? Yahoo! Mail has the best spam protection around
> > > > >> http://mail.yahoo.com
> > > > >>
> > > > >
> > > > >
> > > >
> > > > -----------------------------------------------------------------------
> > > > The AMBER Mail Reflector
> > > > To post, send mail to amber_at_scripps.edu
> > > > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> > > >
> > > >
> > >
> > > --
> > > Ilyas Yildirim
> > > ---------------------------------------------------------------
> > > = Department of Chemistry - =
> > > = University of Rochester - =
> > > = Rochester, NY 14627-0216 - Ph.:(585) 275 67 66 (Office) =
> > > = http://www.pas.rochester.edu/~yildirim/ =
> > > ---------------------------------------------------------------
> > >
> > > -----------------------------------------------------------------------
> > > The AMBER Mail Reflector
> > > To post, send mail to amber_at_scripps.edu
> > > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> > >
> >
> >
> > __________________________________________________
> > Do You Yahoo!?
> > Tired of spam? Yahoo! Mail has the best spam protection around
> > http://mail.yahoo.com
> > -----------------------------------------------------------------------
> > The AMBER Mail Reflector
> > To post, send mail to amber_at_scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> >
> >
>
> --
> Ilyas Yildirim
> ---------------------------------------------------------------
> = Department of Chemistry - =
> = University of Rochester - =
> = Rochester, NY 14627-0216 - Ph.:(585) 275 67 66 (Office) =
> = http://www.pas.rochester.edu/~yildirim/ =
> ---------------------------------------------------------------
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu