[gmx-users] Problem with my pull-code or my topology (or both) - help appreciated

Justin Lemkul jalemkul at vt.edu
Tue Aug 9 00:41:59 CEST 2016 


On 8/8/16 5:00 PM, Billy Williams-Noonan wrote:
> Hi Justin,
>
> Thank you for the confirmation that the pull-code is doing what I thought.
> :)
>
> My confusion is in the fact that the stability of my system seemed to
> depend largely on what cluster I used.
>
> If I used this one:
> http://www.synchrotron.org.au/about-us/our-facilities/scientific-computing-a-it/massive
>
> Every umbrella window of my system crashed within 200 ps, generating over
> 1000 LINCS warnings.
>
> But if I did the same thing (no changes) on this cluster (Snowy):
> https://www.vlsci.org.au/page/computer-software-configuration.
>
> Everything runs stably.
>
> And there are never any LINCS warning on the local machines we have either,
> where I simply pull the ligand into position over 400ps, but what I'm
> finding is that sometimes the ligand never moves to the specified COM pull
> distance even with the high force constant, but when I re-equilibrate my
> ensemble it usually does...
>
> Hope that helps explain my confusion...
>

Any number of things could be wrong here.  Do all the GROMACS tests pass on all 
architectures?  Are the compilers, MPI libraries, etc. all free from bugs? 
You'll often find considerable variability across different clusters, and not 
all sysadmins are created equal :)  There are plenty of buggy things completely 
outside GROMACS that could be to blame, but with only this information, it's 
like shooting in the dark.

-Justin

> Best regards,
>
> Billy
>
> On 8 August 2016 at 20:05, Justin Lemkul <jalemkul at vt.edu> wrote:
>
>>
>>
>> On 8/7/16 7:33 PM, Billy Williams-Noonan wrote:
>>
>>> Hi All,
>>>
>>> Just an update.  I regenerated the topology for the linear peptide with
>>> the
>>> ATB and found myself with similar problems.
>>>
>>> Will this pull until the Z-component of the COM distance is 3.2 nm and
>>> hold
>>> it in place?
>>>
>>> pull                    = yes
>>>
>>> pull_ngroups            = 2
>>> pull_ncoords            = 1
>>> pull_coord1-groups      = 1 2
>>> pull_group1_name        = reference_pull_group
>>> pull_group2_name        = Pull_Group
>>> pull_coord1-type        = umbrella
>>> pull_coord1-geometry    = direction
>>> pull_coord1-dim         = N N Y
>>> ;pull_coord1-rate       = 0.00005
>>> pull_coord1-k           = 1000
>>> pull_coord1-vec         = 0 0 1
>>> pull_coord1-init        =  *3.20*
>>> pull_coord1-start       = no
>>>
>>> pull_print-components   = yes
>>>
>>> pull_nstxout = 10
>>> pull_nstfout = 1
>>>
>>> Is this doing what I think it is doing?  Because it has worked for every
>>> other example I have been using.
>>>
>>>
>> I just posted in a similar thread about this - in effect, yes, but if your
>> starting distance deviates a lot from 3.2, the initial forces are going to
>> be massive and potentially unstable.
>>
>> -Justin
>>
>> Best regards,
>>>
>>> Billy
>>>
>>> On 4 August 2016 at 17:28, Billy Williams-Noonan <
>>> billy.williams-noonan at monash.edu> wrote:
>>>
>>> Hi Gromacs Users,
>>>>
>>>> I have been pulling peptides from two separate proteins (let's call them
>>>> A
>>>> and B) by first taking the crystal structure pose and pulling the ligand
>>>> to
>>>> progressively more distant points from the binding site, and then
>>>> sampling
>>>> at that point.
>>>>
>>>> I first pull the ligand out of the binding site over 400ps with a high
>>>> force constant (10000 kJ/mol/nm^2), and then sample at that point with a
>>>> lower force constant (1000 kJ/mol/nm^2)
>>>>
>>>> The pull-code I have been using in my .mdp file is this:
>>>>
>>>> *400 ps Pull*
>>>>
>>>> pull                    = yes
>>>>
>>>> pull_ngroups            = 2
>>>> pull_ncoords            = 1
>>>> pull_coord1-groups      = 1 2
>>>> pull_group1_name        = reference_pull_group
>>>> pull_group2_name        = Pull_Group
>>>> pull_coord1-type        = umbrella
>>>> pull_coord1-geometry    = direction
>>>> pull_coord1-dim         = N N Y
>>>> pull_coord1-k           = 20000
>>>> pull_coord1-vec         = 0 0 1
>>>> pull_coord1-init        =* pull-distance  *
>>>> pull_coord1-start       = no
>>>>
>>>> pull_print-components   = yes
>>>>
>>>> pull_nstxout = 10
>>>> pull_nstfout = 1
>>>>
>>>> *20 ns sampling:  *
>>>>
>>>> pull                    = yes
>>>>
>>>> pull_ngroups            = 2
>>>> pull_ncoords            = 1
>>>> pull_coord1-groups      = 1 2
>>>> pull_group1_name        = reference_pull_group
>>>> pull_group2_name        = Pull_Group
>>>> pull_coord1-type        = umbrella
>>>> pull_coord1-geometry    = direction
>>>> pull_coord1-dim         = N N Y
>>>> ;pull_coord1-rate       = 0.00005
>>>> pull_coord1-k           = 1000
>>>> pull_coord1-vec         = 0 0 1
>>>> pull_coord1-init        =  *pull-distance*
>>>> pull_coord1-start       = no
>>>>
>>>> pull_print-components   = yes
>>>>
>>>> pull_nstxout = 10
>>>> pull_nstfout = 1
>>>>
>>>> This has worked for two ligands binding to A and one ligand binding to B
>>>> with varying degrees of agreement with experiment.
>>>>
>>>> However, when I try to move my ligands into position with one
>>>> linear-peptide ligand binding to B, the ligand moves to 2.7 nm between
>>>> the
>>>> two pull-group's COM... and then refuses to move further.  The initial
>>>> COM
>>>> pull distance was 1.8 nm, and the final pull distance should be 3.2 nm
>>>> between the two pull groups' COM.
>>>>
>>>> A cyclic peptide version of this molecule was pulled from the binding
>>>> site
>>>> successfully and its topology was generated with the ATB. The linear
>>>> version of this peptide was generated with gmx pdb2gmx and this is the
>>>> *only* difference in the protocol between the two protocols.
>>>>
>>>> So either the pull-code I am using is not doing what I think it is
>>>> (pulling at specific distances along the Z-direction) *OR* the topology
>>>> being generated by *pdb2gmx* is wrong (but I don't think my input is
>>>> defective).  However, this would make sense given I am getting a lot of
>>>> LINCS warnings at some of my positions...
>>>>
>>>> Specifically:
>>>>
>>>>
>>>> *WARNING (1):*
>>>> *lincs-warnangle can not be larger than 90 degrees, setting it to 90.*
>>>>
>>>> Any help would be appreciated... :)
>>>>
>>>> Best regards,
>>>>
>>>> Billy...
>>>>
>>>> --
>>>> Billy Noonan*    |    *PhD Student    *|*    Bsci ( *Adv* ), IA Hon
>>>>
>>>> *LinkedIn Profile
>>>> <http://www.linkedin.com/profile/preview?locale=en_US&trk=
>>>> prof-0-sb-preview-primary-button>
>>>> **|*   +61420 382 557
>>>>
>>>> Monash Institute for Pharmaceutical Sciences ( *MIPS* )
>>>> Royal Parade, Parkville, 3052
>>>>
>>>>
>>>>
>>>
>>>
>> --
>> ==================================================
>>
>> Justin A. Lemkul, Ph.D.
>> Ruth L. Kirschstein NRSA Postdoctoral Fellow
>>
>> Department of Pharmaceutical Sciences
>> School of Pharmacy
>> Health Sciences Facility II, Room 629
>> 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 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================

More information about the gromacs.org_gmx-users mailing list