[gmx-users] Weird structure after minimization (membrane protein simulation

Chris Neale chris.neale at utoronto.ca
Mon Jan 28 20:59:59 CET 2008

 >Dear All,
 >I am doing the simulation of POPE lipid + Protein, I did my system 
setup using mdrun_hole program. It looks fine to me  
(Figure-A). When I was doing energy minimization (using steepest decent 
and conjugant gradient algorithm), water molecules diffuse a lot, 
structure looks very weird (Figure-B). But only after 1ps mdrun (NVT 
ensemble) it comes back to its normal (Figure-C). But during this 1ps I 
got lots of LINCE warning, all for water molecules. If I continue my 
simulation (till now ~5ns production run) I do not get any problem/warning.
 >So I just want to know should I proceed further, or I have to come 
back to my initial state and resolve this problem?
 >Previously I tried different options by changing value of emtol but I 
could not resolve this problem. So I proceeded. By this mail, I am 
requesting expert comments from you people. Is it normal to Membrane 
simulation or there is some problem in my system? Till now I have not 
encountered any problems/warning.
 >Eagerly waiting for your reply,
 >Best regards,
 >Alok Jain
 >Thanks a lot for your reply/comments and time. I am using TIP4P water 
model, and I really could not understand why it happens, Some of the 
bonds of the water molecules are broken down, and after 1ps MD  they 
make bonds again. Is it not very strange? I have tried to visualize in 
different visualization tool but still problem was persisting. I was not 
able to implement your suggestion regarding tolerance limit of the 
visualization software, I used rasmol, chimera, insightII but could not 
found any such option. I am still trying for that, if I could found it, 
I will inform you the result after that.  I am really worried about 
temporary LINCE warning
 >which I was getting. Is there any way to resolve this issue?

Use VMD and set your representation to "dynamic bonds", then there is a 
sliding bar that determines how to detect bonds.
Easier yet, load your initial (presumable ok) structure from figure A 
into VMD, then load in the figure B structure as a new frame
in the original structure. This will draw the representation of fig B 
using the topology as determined from fig A.

 >I am pasting the em.mdp and my top file below.
 >@chris: Thanks for your time spent on investigating on my problem. 
Thanks for creating the public album. I am sorry to say I could not get 
your statement "In the worst case scenario that I can imagine, temporary 
lincs warning could represent a chiral inversion that will never be 
resolved and never give you any more warning messages, but would 
definitely give you the wrong answer." could you please explain it a 
little more (in layman term) because as I think there is no Chiral 
center in water so what it
 >means by chiral inversion.

Yes, water has no chiral center. But your protein does. In the midst of 
all those LINCS warnings,
you might have had one about your protein, and it is even possible that 
only the first N LINCS errors
are reported (you could ask a developer about that) so possibly you have 
protein angles rotating
too much during minimization steps without knowing it.

Regarding what I mean by a chiral inversion... If forces get too high in 
EM, it is entirely possible that
totally unphysical things can happen. To give an MD example, a long time 
ago (using CHARMM) I was doing
simulated annealing and taking my structure up to 5000K. At that 
temperature, I had some strange rearrangements
where the Ca of a Trp had a "chiral inversion" (perhaps not the correct 
term?) in which my L-Trp became D-Trp.
Then upon cooling, the molecule no longer had enough energy to overcome 
this L to D barrier of the improper and
note that impropers do not enforce L amino acids, they just hinder L->D 
or D->L conversions.

 >I have also plotted the two plots to validate my final structure of 
mdrun_hole program and uploaded these plots at 
http://i269.photobucket.com/albums/jj58/gromacs/hole-depth-atom1.jpg  As 
I pasted below my em.mdp file. I was using FLEXIBLE TIP4P water molecules.
 >define              =  -DFLEXIBLE
 >constraints         =  none
 >integrator          =  steep
 >nsteps              =  10000
 >;       Energy minimizing stuff
 >emtol                    =  100
 >emstep                 =  0.001
 >nstcgsteep           =  1000
 >comm_mode      =  Linear
 >nstcomm             =  1
 >comm_grps        =  Protein_POP SOL
 >ns_type               =  grid
 >rlist                       =  0.9
 >coulombtype       =  PME
 >rcoulomb             =  0.9
 >vdw-type              =  Cut-off
 >rvdw                     =  1.2
 >fourierspacing     =  0.12
 >pme_order          =  4
 >ewald_rtol            =  1e-5
 >optimize_fft         =  yes
 >Tcoupl                 =  no
 >Pcoupl                 =  no
 >gen_vel               =  no

Are you sure about "comm_grps        =  Protein_POP SOL" ??
Try with "comm_grps = System"


 >top file:
 >; Include forcefield parameters
 >#include "/home/lysine/ffoplsaa.itp"
 >; Include chain topologies
 >#include "Protein_A.itp"
 >#include "Protein_B.itp"
 >#include "Protein_C.itp"
 >#include "Protein_D.itp"
 >#include "pope_opls.itp"
 >; Include water topology
 >#include "tip4p.itp"
 >; Position restraint for each water oxygen
 >[ position_restraints ]
 >;  i funct       fcx        fcy        fcz
 >   1    1       10000      10000       10000
 >; Include generic topology for ions
 >#include "ions.itp"
 >[ system ]
 >; Name
 >protein + POPE +  TIP4P water molecules
 >[ molecules ]
 >; Compound        #mols
 >Protein_A           1
 >Protein_B           1
 >Protein_C           1
 >Protein_D           1
 >POPE               269
 >SOL              13800

More information about the gromacs.org_gmx-users mailing list