[gmx-users] LINCS warning on galactose molecule

[Mark Abraham]

On 23/03/2011 9:45 PM, Anna Marabotti wrote:
> Dear gmx-users,
> I'm experimenting a LINCS warning on my system, an enzyme with ATP, 
> galactose and Mg. We started from the crystallographic structure, made 
> some accommodation on the topology of the sugar in order to check for 
> partial charges (according to Justin Lemkul's suggestions on PRODRG 
> server), then performed a minimization, NVT and NPT PR-MD and the 
> production MD. After about 5 ns of simulation the MD stopped with the 
> LINCS error:
>
> Step 2562553, time 5125.11 (ps) LINCS WARNING
>
> relative constraint deviation after LINCS:
>
> rms 0.000102, max 0.001903 (between atoms 7643 and 7642)
>
> bonds that rotated more than 30 degrees:
>
> atom 1 atom 2 angle previous, current, constraint length
>
> 7641 7640 45.3 0.1000 0.1000 0.1000
>
> Step 2562554, time 5125.11 (ps) LINCS WARNING
>
> relative constraint deviation after LINCS:
>
> rms 0.000104, max 0.001558 (between atoms 7643 and 7642)
>
> bonds that rotated more than 30 degrees:
>
> atom 1 atom 2 angle previous, current, constraint length
>
> 7641 7640 36.5 0.1000 0.1000 0.1000
>
> Step 2562565, time 5125.13 (ps) LINCS WARNING
>
> relative constraint deviation after LINCS:
>
> rms 0.000318, max 0.007712 (between atoms 7641 and 7640)
>
> bonds that rotated more than 30 degrees:
>
> atom 1 atom 2 angle previous, current, constraint length
>
> 7641 7640 31.5 0.1005 0.1008 0.1000
>
> Step 2562566, time 5125.13 (ps) LINCS WARNING
>
> relative constraint deviation after LINCS:
>
> rms 0.000374, max 0.009231 (between atoms 7641 and 7640)
>
> bonds that rotated more than 30 degrees:
>
> atom 1 atom 2 angle previous, current, constraint length
>
> 7641 7640 38.2 0.1008 0.1009 0.1000
>
> Step 2562567, time 5125.13 (ps) LINCS WARNING
>
> relative constraint deviation after LINCS:
>
> rms 0.000310, max 0.007658 (between atoms 7641 and 7640)
>
> bonds that rotated more than 30 degrees:
>
> atom 1 atom 2 angle previous, current, constraint length
>
> 7641 7640 38.6 0.1009 0.1008 0.1000
>
> ............................................................
>
> Step 2563563, time 5127.13 (ps) LINCS WARNING
>
> relative constraint deviation after LINCS:
>
> rms 0.000070, max 0.001640 (between atoms 7641 and 7640)
>
> bonds that rotated more than 30 degrees:
>
> atom 1 atom 2 angle previous, current, constraint length
>
> 7641 7640 57.0 0.0998 0.1002 0.1000
>
> -------------------------------------------------------
>
> Program mdrun, VERSION 4.0.3
>
> Source code file: constr.c, line: 136
>
> Fatal error:
>
> Too many LINCS warnings (1000)
>
> If you know what you are doing you can adjust the lincs warning 
> threshold in your mdp file
>
> or set the environment variable GMX_MAXCONSTRWARN to -1,
>
> but normally it is better to fix the problem
>
> -------------------------------------------------------
>
> We checked the system to find the problem. The atoms 7641 and 7640 
> correspond to atom O1 and H12 of the galactose. With VMD we visualized 
> the trajectory, but we did not see anything anomalous on this bond. 
> Length and angles are quite normal. This OH group is near the 
> phosphate of ATP but the distances seem to be compatible with an 
> (expected) H-bond. I don't think the problem could be an incorrect 
> stabilization because it seems to me that in that case the problem 
> would have appear during the first step of MD, not after 5 ns. In any 
> case, stabilization went correctly, with no errors or warnings. I 
> checked on the gmx-users list and found some messages of Xavier 
> Periole experimenting some LINCS problems with gmx 4.0.4, that seems 
> to resemble my situation. I followed the thread, but I did not see 
> anything that could be of help for me. Some hypotheses were about the 
> gmx version (mine is 4.0.3), communications between CPUs, virtual 
> sites. Looking at the trajectory, I don't think it's a problem of 
> periodic images too close. The thread stops with Xavier saying he 
> would like to test the effects of using -rdd and -rcon on LINCS 
> warning, but I did not find the rest of the history.
> Does anybody (and especially Xavier, if possible!) have some hints 
> about my problem? Or could it be a problem of galactose topology?

Usually the problem has nothing to do with LINCS or -rdd or -con - 
they're just symptoms of machinery that breaks first.

I'd start by assuming my topologies were not well-formed somehow, and 
divide and conquer to locate the problem. Can you simulate galactose in 
vacuo? In a small water box? Same for ATP. Same for enzyme.

Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110323/48ab08e2/attachment.html>