Any limitation for system size to run Metadynamics with plumed!

[Anjai]

Dear Users,

I am interested to explore the conformations of a large protein (920 residues, ~ 14000 atoms) in solution. The simulation system size with a dodecahedron box with water in is about 200,000 atoms. I am using gromacs MD simulation with the following options

; neighbour searching
cutoff-scheme   = Verlet; ns_type             = grid; nstlist             = 10; rcoulomb            = 1.50 ; rvdw                = 1.50;  pbc                 = xyz; periodic_molecules  = no
; electrostatic
coulombtype         = PME; pme_order           = 4 ; fourierspacing  = 0.16 
; vdw
vdw-type            = Cut-off;
; constraints
constraints              = all-bonds; constraint-algorithm     = lincs; lincs_iter               = 4
; temperature
Tcoupl              = v-rescale; tc_grps             = Protein Non-Protein; tau_t               = 0.1 0.1; ref_t               = 300.0 300.0
; pression
Pcoupl              =  no;
; initial velocities
gen_vel             = yes; gen_temp            = 300.000; gen_seed            = 1536

For the Metadynamics following options are used

pro: GROUP NDX_FILE=index.ndx NDX_GROUP=Protein
WHOLEMOLECULES ENTITY0=pro
GYRATION TYPE=RADIUS ATOMS=pro LABEL=rg

metad: METAD ARG=rg PACE=500 HEIGHT=0.7 SIGMA=0.12 FILE=HILLS

# monitor the two variables and the metadynamics bias potential
PRINT STRIDE=500 ARG=rg,metad.bias FILE=COLVAR

In the literature I didn't see anyone using Metadyanmics for such large systems. My results after running simulation for about 100ns are not very positive. I have few testcases of the same size protein, some of them blown-up after running for 80 ns and some are running smoothly but even after 100 ns didn't see convergence from CV vs time.

Blowing-up message was
----------------------------------------------------------------------------------------------------------------------------------------------------
Not all bonded interactions have been properly assigned to the domain decomposition cells

A list of missing interactions:
               LJ-14 of  57536 missing      8
          exclusions of 264356 missing      8

Molecule type 'Protein'
the first 10 missing interactions, except for exclusions:
               LJ-14 atoms 1858 4575           global  1858  4575
               LJ-14 atoms 1866 4575           global  1866  4575
               LJ-14 atoms 1884 4575           global  1884  4575
               LJ-14 atoms 1906 4575           global  1906  4575
               LJ-14 atoms 1931 4575           global  1931  4575
               LJ-14 atoms 1933 4575           global  1933  4575
               LJ-14 atoms 1944 4575           global  1944  4575
               LJ-14 atoms 2325 4575           global  2325  4575

-------------------------------------------------------
Program mdrun_mpiPlumed, VERSION 4.6.5
gromacs-4.6.5plumed/src/mdlib/domdec_top.c, line: 393
Fatal error:
16 of the 392794 bonded interactions could not be calculated because some atoms involved moved further apart than the multi-body cut-off distance (0.908498 nm) or the two-body cut-off distance (1.5 nm), see option -rdd, for pairs and tabulated bonds also see option -ddcheck
For more information and tips for troubleshooting, please check the GROMACS
------------------------------------------------------------------------------------------------------------------------------------------------------------

I appreciate anyone share their experience on using metadynamics for such large protein systems.Objective of these simulations is to find the conformation states with free energy minima.

[Giovanni Bussi]

Hi,

I am not sure that the gyration radius can tell you anything meaningful in a 920 residue protein.

Giovanni

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.
To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/3a0851a7-b9b3-4863-be33-948038fdfbd0%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Massimiliano Bonomi]

Hello,

consider also that a protein of 920 residues to which you are applying a bias potential to
explore its conformational space will sample extended conformations.
I suspect that the cell size you are using is too small to accomodate these extended conformations.

Anyway, I would suggest to read the extensive literature on the use of metadynamics to study the
conformational/folding landscape of proteins and the issues regarding the choice of collective variables.
I am not sure which is the maximum size studied so far, but I don’t think it is more than ~100 residues.

Max

[Anjai]

Thank you very much for the reply. I have only SAXS experimental data for this system and I didn't find any CV to incorporate the SAXS data on-fly in metadynamics simulation except using Rg value calculated from that SAXS curve. I agree that Rg is too degenerated but i thought atleast i can screen the conformations from minima Rg posteriori based on the SAXS data. Any other better CV recommended for such large  system.

To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.

[Anjai]

Thank you very much for the reply. I think what you mentioned about the box size to accommodate extended conformations is correct. That may be the reason for blowing-up of simulation. But if I increase box size more my system size will become very expensive as I am using explicit water. In the literature as you mentioned no studies for such large proteins. But I don't quite understand why it is not been used, technically I guess it can apply to any system size.Is it suggestible to use implicit water instead of explicit water to increase the convergence of CV.