# Re: constant velocity SMD

From: JC Gumbart (gumbart_at_ks.uiuc.edu)
Date: Fri Jan 26 2007 - 11:59:46 CST

1) What direction do you want to pull it in? Presumably away from
the protein along some path; you will have to decide this. You may
find however that you will need to restrain the protein in some way,
otherwise you will likely pull it right along with your ligand!

2) The center of mass does not become the smd atom as the center of
mass is still just a fictitious point. When you define all the atoms
as smd atoms, effectively the center of mass is pulled, but the force
is applied to all the atoms weighted by their masses.

3) Actually, a timestep of 1 fs is best for any MD, but I don't think
there is any unique benefit in the case of SMD (someone correct me if
I missed something). Don't forget though, the SMD velocity is
defined in terms of Angstroms/timestep so if you use 2 fs, then your
real velocity will be half what is in the configuration file.

On Jan 26, 2007, at 11:11 AM, Narender Singh Maan wrote:

> Dear Users,
>
> I have some questions about setting up my system (a protein-ligand
> system with constant velocity SMD)…In this i am trying to pull the
> ligand out of the protein. So,
>
> 1) as it says in namd/smd tutorial it keep one atom fixed
> (beta 1) and smd-atom moving (occupancy 1)…but in my situation i
> just want to pull the ligand out of the protein's binding site and
> do not want to keep anything fixed (in protein). So in this
> situation I have all my beta column set to zero. So the question is
> how do I find out the vector direction? (since it requires to find
> the vector direction that links fixed and smd atom, and I don't
> have the fixed atom!)
>
> 2) I can define a fixed Center of mass of my ligand and make
> it as smd-atom and pull it out from that point but how can I define
> all the atoms (as smd-atoms) of my ligand and pull them out
> together?..
>
> 3) Many of the related papers that I see for SMD shows the
> timestep of 1fs!!..why is it so??..normally in MD we keep it
> 2fs…..is it that SMD is more stable with 1fs timestep??..
>
>
> This here is my coordinate xx.pdb file (segname PRO1 is my protein
> part and segname INH1 is my ligand)
>
> ATOM 3670 HG23 THR 234 -2.295 20.477
> -15.764 0.00 0.00 PRO1
>
> ATOM 3671 C THR 234 -5.395 20.707
> -12.464 0.00 0.00 PRO1
>
> ATOM 3672 OT1 THR 234 -6.336 20.719
> -13.291 0.00 0.00 PRO1
>
> ATOM 3673 OT2 THR 234 -5.441 21.306
> -11.370 0.00 0.00 PRO1
>
> ATOM 3674 C29 RPR 1 -1.666 -5.800
> 10.832 0.00 0.00 INH1
>
> ATOM 3675 N30 RPR 1 -2.030 -6.192
> 9.650 0.00 0.00 INH1
>
> ATOM 3676 H30 RPR 1 -2.660 -6.969
> 9.662 0.00 0.00 INH1
>
> This here is my smdfile xx.ref file
>
> ATOM 3670 HG23 THR 234 -2.295 20.477
> -15.764 1.00 0.00 PRO1
>
> ATOM 3671 C THR 234 -5.395 20.707
> -12.464 1.00 0.00 PRO1
>
> ATOM 3672 OT1 THR 234 -6.336 20.719
> -13.291 1.00 0.00 PRO1
>
> ATOM 3673 OT2 THR 234 -5.441 21.306
> -11.370 1.00 0.00 PRO1
>
> ATOM 3674 C29 RPR 1 -1.666 -5.800
> 10.832 1.00 0.00 INH1
>
> ATOM 3675 N30 RPR 1 -2.030 -6.192
> 9.650 1.00 0.00 INH1
>
> ATOM 3676 H30 RPR 1 -2.660 -6.969
> 9.662 1.00 0.00 INH1
>
>
> And this here is a part of my xx.conf file
>
> ___________________________________________________________________
>
> fixedAtoms on
>
> fixedAtomsFile xx.ref
>
> fixedAtomsCol B (???)
>
> SMD on
>
> SMDfile xx.ref
>
> SMDk 5;# 347.395 pNA
>
> SMDvel 0.00001 ; # 10A/ns since ts is 1fs
>
> SMDDir -0.3534 -0.4135 0.8391 ; # (??just a test)
>
> SMDOutputFreq 100
>
> __________________________________________________________________
>
> Any suggestions would be of great help
>
> Thank you
>
> singh
> P.S. I searched the list for these questions but please pardon me