[gmx-users] -inf potential with TPIC

João M. Damas jmdamas at itqb.unl.pt
Sat Jun 14 06:19:12 CEST 2014


Interesting... Does it also happen on the mu or only on the <mu>? I ask
this because the <mu>, unlike the mu, has the volume in it (as per the
Widom insertion equations for the isothermic-isobaric ensemble). If the
"-inf" is only on the <mu>, I'd suggest checking the volume along the
trajectory (using g_energy, for example, or check the .xvg file from the
-tpi flag, which already has the volume, I think). Check the post-processed
trajectory that you feed mdrun -rerun, as your trajectory-processing script
for TPIC may be doing something strange to the box size.

I recognize that the single/double-precision is intriguing, if indeed the
problem is the volume.

João

On Fri, Jun 13, 2014 at 11:10 PM, Rafael I. Silverman y de la Vega <
rsilverm at ucsc.edu> wrote:

> Hi João,
> I dont think its an excessive number of unfavorable insertions. The mu
> getting output with double precision are
> Reading frame      40 time 1770.000   mu -6.570e+01 <mu> -7.654e+01
> Reading frame      50 time 1775.000   mu -4.985e+01 <mu> -7.609e+01
> Reading frame      60 time 1780.000   mu -7.010e+01 <mu> -7.571e+01
> Reading frame      70 time 1785.000   mu -5.669e+01 <mu> -7.540e+01
> Reading frame      80 time 1790.000   mu -4.749e+01 <mu> -7.511e+01
> Reading frame      90 time 1795.000   mu -7.355e+01 <mu> -7.590e+01
>
> to give a fairly representative sample. So, no, they arent near 100, and
> are negative... And the issue seems to happen with later insertions, not
> the first few... One run I just did, only the very last average mu was
> -inf, (not even the mu for that frame, just the average mu after that
> frame) which really seems to imply some sort of overflow problem
>
> Reading frame     180 time 1790.000   mu -1.284e+02 <mu> -1.153e+02
> Reading frame     190 time 1795.000   mu -7.920e+01 <mu>       -inf
> (not the same run as above, this was run in single precision)
>
>
> On Thu, Jun 12, 2014 at 7:06 AM, João M. Damas <jmdamas at itqb.unl.pt>
> wrote:
>
> > Hello Rafael,
> >
> > High energies (energy differences) are not discarded. When summing them
> > through the exponential of their negative, they contribute very little to
> > the final sum (lower energies dominate the sum). Hence, they are
> > "discarded".
> >
> > The "-inf" value on the mu standard error output is normal because,
> unlike
> > the -tpi output, there is no ternary operator checking if the sum of
> > exponentials is equal to zero (when it is, the log of zero gives you the
> > inf value).
> >
> > The change to the double precision not giving you the "-inf" also makes
> > sense, because the sum of exponentials, which on single-precision was
> zero,
> > can now be non-zero due to the higher number of decimal places on
> > double-precision. I am guessing with double-precision you are getting
> very
> > high positive numbers (energies) for the mu, around 1000 maybe?
> >
> > My interpretation is that it is hard to find a "favorable place" to
> insert
> > the water molecule in that cavity, at least with a few frames. I am
> > guessing that with more and more frames, that may change, which is
> natural.
> >
> > I hope this has helped.
> >
> > Cheers,
> > João
> >
> >
> >
> > On Thu, Jun 12, 2014 at 10:45 AM, Justin Lemkul <jalemkul at vt.edu> wrote:
> >
> > >
> > >
> > > On 6/11/14, 8:55 PM, Rafael I. Silverman y de la Vega wrote:
> > >
> > >> I mistyped amber, when I should have typed charmm, amber is totally
> > >> uninvolved in  my calculations. I used the script from
> > >> http://www.gromacs.org/Downloads/User_contributions/Other_software
> > titled
> > >> charmm2gromacs. The output looked fine after running it.
> > >>
> > >> And, I doubt its the cofactor,  just ran the insertion with double
> > >> precision mdrun, it gave reasonable looking results. There seems to be
> > >> some
> > >> sort of issue with unreasonable interaction energies not being
> discarded
> > >> properly.
> > >>
> > >>
> > > It's entirely possible that there's something wrong with the tpic code;
> > > you can certainly file a bug report on Redmine with input files to
> > > reproduce it.
> > >
> > >
> > > -Justin
> > >
> > >
> > >> On Tue, Jun 10, 2014 at 3:41 PM, Justin Lemkul <jalemkul at vt.edu>
> wrote:
> > >>
> > >>
> > >>>
> > >>> On 6/10/14, 6:14 PM, Rafael I. Silverman y de la Vega wrote:
> > >>>
> > >>>  I meant that I only changed to cofactor to gromacs chamm27 format,
> not
> > >>>> amber format. I used charmm27 for the apoprotein as well as the
> > >>>> cofactor.
> > >>>>
> > >>>>
> > >>>>  What script did you use?  I don't understand why Amber99SB is
> > involved
> > >>> here, so I don't know exactly what you were doing to create the
> > topology.
> > >>>   It may not end up being useful information, but it's important to
> > >>> establish the validity of the underlying physical model.
> > >>>
> > >>>
> > >>> -Justin
> > >>>
> > >>>
> > >>>  On Tue, Jun 10, 2014 at 3:09 PM, Justin Lemkul <jalemkul at vt.edu>
> > wrote:
> > >>>>
> > >>>>
> > >>>>
> > >>>>> On 6/10/14, 6:04 PM, Rafael I. Silverman y de la Vega wrote:
> > >>>>>
> > >>>>>   you think it was the cofactor model?
> > >>>>>
> > >>>>>>
> > >>>>>>
> > >>>>>>   If you're using a CHARMM-parametrized cofactor with an Amber99SB
> > >>>>>>
> > >>>>> protein,
> > >>>>> I'd say the simulations are wholly unreliable.  It's certainly the
> #1
> > >>>>> likely cause for nonsensical results.  If you want to get to the
> root
> > >>>>> of
> > >>>>> the issue, you need to start with a solid foundation.
> > >>>>>
> > >>>>> Unless that's not what your last message meant?  When you say "only
> > >>>>> changed the cofactor," that implied to me that you didn't
> > consistently
> > >>>>> convert everything in the system.
> > >>>>>
> > >>>>>
> > >>>>> -Justin
> > >>>>>
> > >>>>>
> > >>>>>   On Tue, Jun 10, 2014 at 2:23 PM, Justin Lemkul <jalemkul at vt.edu>
> > >>>>> wrote:
> > >>>>>
> > >>>>>>
> > >>>>>>
> > >>>>>>
> > >>>>>>  On 6/10/14, 5:01 PM, Rafael I. Silverman y de la Vega wrote:
> > >>>>>>>
> > >>>>>>>    wait, I only changed  the cofactor to gromacs charmm27.ff
> format
> > >>>>>>>
> > >>>>>>>
> > >>>>>>>>
> > >>>>>>>>    Pretty amazing that the simulation even ran with that mash-up
> > of
> > >>>>>>>>
> > >>>>>>>>  parameters :)
> > >>>>>>>
> > >>>>>>> I'd say that's the most likely cause of your problems;
> nonsensical
> > >>>>>>> physical model leads to nonsensical output.
> > >>>>>>>
> > >>>>>>> -Justin
> > >>>>>>>
> > >>>>>>>
> > >>>>>>>
> > >>>>>>>    On Tue, Jun 10, 2014 at 1:58 PM, Rafael I. Silverman y de la
> > Vega
> > >>>>>>> <
> > >>>>>>>
> > >>>>>>>  rsilverm at ucsc.edu> wrote:
> > >>>>>>>>
> > >>>>>>>>     I suppose something might be wrong with the simulations. I
> am
> > >>>>>>>> using
> > >>>>>>>>
> > >>>>>>>>   gromacs 4.6.5, I am simulating a protein with a cofactor,
> > someone
> > >>>>>>>>
> > >>>>>>>>> else
> > >>>>>>>>> parametrized it, I changed from charmm to gromacs amber99sb.ff
> > >>>>>>>>> format
> > >>>>>>>>> with
> > >>>>>>>>> a script on the gromacs website.  I did a typical em/nvt/npt
> > >>>>>>>>> equilibration
> > >>>>>>>>> followed by a 5 ns production run. I picked 250 frames out of
> the
> > >>>>>>>>> trajectory, and altered them to the TPIC input format with a
> > >>>>>>>>> script I
> > >>>>>>>>> wrote, to insert the particle half way between two atoms that I
> > >>>>>>>>> chose.
> > >>>>>>>>> I
> > >>>>>>>>> am
> > >>>>>>>>> inserting a water molecule. When I run the actual insertion
> some
> > of
> > >>>>>>>>> the
> > >>>>>>>>> frames have a reasonable looking mu, and some have -inf mu. I
> > think
> > >>>>>>>>> it
> > >>>>>>>>> may
> > >>>>>>>>> have to do with cavity size, some frames have too small of a
> > >>>>>>>>> cavity,
> > >>>>>>>>> so
> > >>>>>>>>> the
> > >>>>>>>>> interaction energy of the inserted water diverges...
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>> On Mon, Jun 9, 2014 at 6:40 PM, Justin Lemkul <jalemkul at vt.edu
> >
> > >>>>>>>>> wrote:
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>>   On 6/9/14, 6:41 PM, Rafael I. Silverman y de la Vega wrote:
> > >>>>>>>>>
> > >>>>>>>>>>
> > >>>>>>>>>>     Hi all, I am having a problem with -inf mu coming out of
> > TPIC
> > >>>>>>>>>>
> > >>>>>>>>>>   calculations.
> > >>>>>>>>>>
> > >>>>>>>>>>> Anyone have some suggestions on how to avoid this? Tabulated
> > >>>>>>>>>>> potentials?
> > >>>>>>>>>>> Or
> > >>>>>>>>>>> can I modify the source code to reject the absurd values that
> > >>>>>>>>>>> cause
> > >>>>>>>>>>> this? I
> > >>>>>>>>>>> read a few old discussions on this, I didnt see much on how
> to
> > >>>>>>>>>>> handle
> > >>>>>>>>>>> this.
> > >>>>>>>>>>>
> > >>>>>>>>>>>
> > >>>>>>>>>>>     I'd be much more concerned that something is going wrong
> in
> > >>>>>>>>>>> the
> > >>>>>>>>>>>
> > >>>>>>>>>>>   simulations.  I have no experience with TPIC, but if you
> post
> > >>>>>>>>>>> more
> > >>>>>>>>>>>
> > >>>>>>>>>> detailed
> > >>>>>>>>>> information about what you're doing and which Gromacs version
> > >>>>>>>>>> you're
> > >>>>>>>>>> using,
> > >>>>>>>>>> you're more likely to get some suggestions.
> > >>>>>>>>>>
> > >>>>>>>>>> -Justin
> > >>>>>>>>>>
> > >>>>>>>>>> --
> > >>>>>>>>>> ==================================================
> > >>>>>>>>>>
> > >>>>>>>>>> Justin A. Lemkul, Ph.D.
> > >>>>>>>>>> Ruth L. Kirschstein NRSA Postdoctoral Fellow
> > >>>>>>>>>>
> > >>>>>>>>>> Department of Pharmaceutical Sciences
> > >>>>>>>>>> School of Pharmacy
> > >>>>>>>>>> Health Sciences Facility II, Room 601
> > >>>>>>>>>> University of Maryland, Baltimore
> > >>>>>>>>>> 20 Penn St.
> > >>>>>>>>>> Baltimore, MD 21201
> > >>>>>>>>>>
> > >>>>>>>>>> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
> > >>>>>>>>>> http://mackerell.umaryland.edu/~jalemkul
> > >>>>>>>>>>
> > >>>>>>>>>> ==================================================
> > >>>>>>>>>> --
> > >>>>>>>>>> Gromacs Users mailing list
> > >>>>>>>>>>
> > >>>>>>>>>> * Please search the archive at http://www.gromacs.org/
> > >>>>>>>>>> Support/Mailing_Lists/GMX-Users_List before posting!
> > >>>>>>>>>>
> > >>>>>>>>>> * Can't post? Read
> http://www.gromacs.org/Support/Mailing_Lists
> > >>>>>>>>>>
> > >>>>>>>>>> * For (un)subscribe requests visit
> > >>>>>>>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_
> > >>>>>>>>>> gmx-users
> > >>>>>>>>>> or
> > >>>>>>>>>> send a mail to gmx-users-request at gromacs.org.
> > >>>>>>>>>>
> > >>>>>>>>>>
> > >>>>>>>>>>
> > >>>>>>>>>>
> > >>>>>>>>>>     --
> > >>>>>>>>>
> > >>>>>>>>>  ==================================================
> > >>>>>>>>
> > >>>>>>>
> > >>>>>>> Justin A. Lemkul, Ph.D.
> > >>>>>>> Ruth L. Kirschstein NRSA Postdoctoral Fellow
> > >>>>>>>
> > >>>>>>> Department of Pharmaceutical Sciences
> > >>>>>>> School of Pharmacy
> > >>>>>>> Health Sciences Facility II, Room 601
> > >>>>>>> University of Maryland, Baltimore
> > >>>>>>> 20 Penn St.
> > >>>>>>> Baltimore, MD 21201
> > >>>>>>>
> > >>>>>>> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
> > >>>>>>> http://mackerell.umaryland.edu/~jalemkul
> > >>>>>>>
> > >>>>>>> ==================================================
> > >>>>>>> --
> > >>>>>>> Gromacs Users mailing list
> > >>>>>>>
> > >>>>>>> * Please search the archive at http://www.gromacs.org/
> > >>>>>>> Support/Mailing_Lists/GMX-Users_List before posting!
> > >>>>>>>
> > >>>>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >>>>>>>
> > >>>>>>> * For (un)subscribe requests visit
> > >>>>>>>
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users
> > >>>>>>> or
> > >>>>>>> send a mail to gmx-users-request at gromacs.org.
> > >>>>>>>
> > >>>>>>>
> > >>>>>>>   --
> > >>>>>>>
> > >>>>>> ==================================================
> > >>>>>
> > >>>>> Justin A. Lemkul, Ph.D.
> > >>>>> Ruth L. Kirschstein NRSA Postdoctoral Fellow
> > >>>>>
> > >>>>> Department of Pharmaceutical Sciences
> > >>>>> School of Pharmacy
> > >>>>> Health Sciences Facility II, Room 601
> > >>>>> University of Maryland, Baltimore
> > >>>>> 20 Penn St.
> > >>>>> Baltimore, MD 21201
> > >>>>>
> > >>>>> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
> > >>>>> http://mackerell.umaryland.edu/~jalemkul
> > >>>>>
> > >>>>> ==================================================
> > >>>>> --
> > >>>>> Gromacs Users mailing list
> > >>>>>
> > >>>>> * Please search the archive at http://www.gromacs.org/
> > >>>>> Support/Mailing_Lists/GMX-Users_List before posting!
> > >>>>>
> > >>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >>>>>
> > >>>>> * For (un)subscribe requests visit
> > >>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users
> > or
> > >>>>> send a mail to gmx-users-request at gromacs.org.
> > >>>>>
> > >>>>>
> > >>>>>  --
> > >>> ==================================================
> > >>>
> > >>> Justin A. Lemkul, Ph.D.
> > >>> Ruth L. Kirschstein NRSA Postdoctoral Fellow
> > >>>
> > >>> Department of Pharmaceutical Sciences
> > >>> School of Pharmacy
> > >>> Health Sciences Facility II, Room 601
> > >>> University of Maryland, Baltimore
> > >>> 20 Penn St.
> > >>> Baltimore, MD 21201
> > >>>
> > >>> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
> > >>> http://mackerell.umaryland.edu/~jalemkul
> > >>>
> > >>> ==================================================
> > >>> --
> > >>> Gromacs Users mailing list
> > >>>
> > >>> * Please search the archive at http://www.gromacs.org/
> > >>> Support/Mailing_Lists/GMX-Users_List before posting!
> > >>>
> > >>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >>>
> > >>> * For (un)subscribe requests visit
> > >>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users
> or
> > >>> send a mail to gmx-users-request at gromacs.org.
> > >>>
> > >>>
> > > --
> > > ==================================================
> > >
> > > Justin A. Lemkul, Ph.D.
> > > Ruth L. Kirschstein NRSA Postdoctoral Fellow
> > >
> > > Department of Pharmaceutical Sciences
> > > School of Pharmacy
> > > Health Sciences Facility II, Room 601
> > > University of Maryland, Baltimore
> > > 20 Penn St.
> > > Baltimore, MD 21201
> > >
> > > jalemkul at outerbanks.umaryland.edu | (410) 706-7441
> > > http://mackerell.umaryland.edu/~jalemkul
> > >
> > > ==================================================
> > > --
> > > Gromacs Users mailing list
> > >
> > > * Please search the archive at http://www.gromacs.org/
> > > Support/Mailing_Lists/GMX-Users_List before posting!
> > >
> > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >
> > > * For (un)subscribe requests visit
> > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > > send a mail to gmx-users-request at gromacs.org.
> > >
> >
> >
> >
> > --
> > João M. Damas
> > PhD Student
> > Protein Modelling Group
> > ITQB-UNL, Oeiras, Portugal
> > Tel:+351-214469613
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > posting!
> >
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> > * For (un)subscribe requests visit
> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > send a mail to gmx-users-request at gromacs.org.
> >
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>



-- 
João M. Damas
PhD Student
Protein Modelling Group
ITQB-UNL, Oeiras, Portugal
Tel:+351-214469613


More information about the gromacs.org_gmx-users mailing list