Re: [AMBER] Mutation of a protein residue in TI computation with pmemd.MPI

From: <>
Date: Fri, 17 Jul 2015 06:30:40 +0000

Dan has already told you how to principally solve this. But the description of what you are planning to do does not seem to match your input file. I understand that you want to do an elestrostatic transformation only first but what crgmask does is to only set the atom charges to zero. Apparently you switch off the charges of ALA completely. If that is really your plan you would do a linear transformation of ALA(charge) to ALA(uncharged). Although if you mutate PHE->ALA the steps would seem to be: 1) PHE(charged)->PHE(uncharged); 2) PHE(uncharged)->ALA(uncharged); 3) ALA(uncharged)->ALA(charged).

An electrostatic only transformation would (currently) require you to create and intermediate that has all vdW and bonded parameters of PHE but the charges of ALA. As you would do that in a linear transformation you would need to fill up all disappearing atoms of PHE with dummies in ALA. These dummies are then defined as softcore atoms in the second step. Overall, this protocol seems to be more attractive to me but requires more work to set up (I am writing a tool to do exactly this but it is still only dealing with non-covalently bound molecules yet). Alternatively, pmemd would implement lambda paths which would make this much simpler. Dan, is this still on your TODO list?

From: M Olivia Kim []
Sent: 16 July 2015 23:13
Subject: [AMBER] Mutation of a protein residue in TI computation with pmemd.MPI


I am trying to run a FEP simulation for mutation of a protein residue using the pmemd.MPI module. I'm mutating a Phe to Ala - following the Amber14 manual, I prepared the prmtop and inpcrd files using where the redundant bonding terms for the overlapping residues are deleted. So the final pdb for my system looks like that protein consists of residues 1-177 and the residues 178 is Ala (separated by TER), which is the mutant for the Phe in the WT.

I was first trying to run the non-soft core simulation, where the atomic charges of the WT protein residue are changed to the charges of the mutated residue along with lambda. However, since the atom numbers of the WT residue and mutant are different, I couldn't even run the minimization. The part corresponding to TI in my minimization input file is:

  icfe = 1, clambda = 0.0,
  ifsc = 0,
  timask1 = '.1019,1020,1021,1022,1023,1024,1025,1026,1027,1028,1029,1030,1031,1032,1033,1034,1035,1036,1037,1038', timask2 = '.2780,2781,2782,2783,2784,2785,2786,2787,2788,2789',
  crgmask = '.2780,2781,2782,2783,2784,2785,2786,2787,2788,2789',

And the error message I got is:

TI Mask 1 .1019,1020,1021,1022,1023,1024,1025,1026,1027,1028,1029,1030,1031,1032,1033,1034,1035,1036,1037,1038; matches 20 atoms
TI Mask 2 .2780,2781,2782,2783,2784,2785,2786,2787,2788,2789; matches 10 atoms
ERROR: timask1/2 must match the same number of atoms for non-softcore run

Is there any way I can correct my input or prmtop file so that I can run this simulation? Or is this type of simulation (mutation of a protein residue to another that has different number of atoms) not supported by pmemd version of TI in Amber yet? I'll look forward to any advice or comments. Thanks in advance!

AMBER mailing list

AMBER mailing list
Received on Fri Jul 17 2015 - 00:00:03 PDT
Custom Search