Re: [AMBER] TI Softcore potentials and SHAKE

From: Kepa K. Burusco <>
Date: Mon, 1 Oct 2012 00:12:10 +0100 (BST)


Hi Brian (and all),

Thanks for the info.

So, just to be sure that I understood you correctly: if I run a TI simulation asking for jfastw=0 (by default), ntc=1 and ntf=1, do you mean that "sander" automatically recognises TIP3P water molecules as RIGID ones by default and does the calculation SHAKEing the solvent but not the solute? Because this is the case I am interested in.

Kepa K.


 De: Brian Radak <>
Para: Kepa K. Burusco <>; AMBER Mailing List <>
Enviado: Viernes 28 de septiembre de 2012 19:04
Asunto: Re: [AMBER] TI Softcore potentials and SHAKE

Hi Kepa,

Three point solvent molecules are identified in the prmtop (and by Leap too maybe?). I believe that SETTLE and not SHAKE is used to constrain those when ntc > 1 and jfastw = 0 (default). If you set jfastw = 4 then normal SHAKE is used. I think you can only get away with ntf > 1 when all bonds/angles/dihedrals are constrained (i.e. when not using noshakemask).


On Fri, Sep 28, 2012 at 1:27 PM, Kepa K. Burusco <> wrote:

>I am trying to run some TI calculation tests, in explicit water solvent (TIP3P model) with AMBER 11 for a glucose molecule (GLYCAM06 force field) for the alpha to beta anomer transformation (GLA and GLB residue names). The NVT equilibration includes a heating slope of 100ps. followed by 200ps. at 298 K, (all in the same simulation step). I am using lambdas from 0 to 1 in growing intervals of 0.1 and a refinement at the edges (0.01; 0.05; 0.95 and 0.99).
>Shortly after I sumbitted the job, the system failed (all of the lambda values) according to SHAKE problems, giving the error message:
>"Softcore potentials require ntf=1 because SHAKE constraints on some bonds might be removed"
>(See output files for lambda 0.01 attached)
>I am aware that SC potentials don't get on very well with SHAKE algorithm and I used the "noshakemask" keyword to UNSHAKE the residues that contain the atoms involved in the transformation in both V0 and V1 states. So, the only residues under SHAKE in my system were the water molecules (the solvent).
>The AMBER11 manual says that "noshakemask" must be used to unshake residues (which I did) and AMBER12 manual says also that "just setting noshakemask might not be enough, since this flag does not affect the settle
>routine that handles rigid waters". The tutorials A9 and A6 don't say so much about this and I did not find any answer in the amber list.
>Is it possible to run TI calculations applying shake only to solvent molecules? Can anyone give a clue to solve this problem?
>Thank you very much.
>Kepa K.
>AMBER mailing list

================================ Current Address =======================
 Brian Radak                                             :     BioMaPS Institute for Quantitative Biology
 PhD candidate - York Research Group       :     Rutgers, The State University of New Jersey
 University of Minnesota - Twin Cities         :     Center for Integrative Proteomics Room 308
 Graduate Program in Chemical Physics     :     174 Frelinghuysen Road,
 Department of Chemistry                          :     Piscataway, NJ 08854-8066                                 :
Sorry for the multiple e-mail addresses, just use the institute appropriate address.
AMBER mailing list
Received on Sun Sep 30 2012 - 16:30:03 PDT
Custom Search