AW: FEP on an amino acid ligand

From: Norman Geist (
Date: Tue Jan 22 2013 - 00:28:38 CST

> -----Ursprüngliche Nachricht-----
> Von: [] Im
> Auftrag von Thomas Albers
> Gesendet: Dienstag, 22. Januar 2013 02:44
> An: Namd Mailing List
> Betreff: namd-l: FEP on an amino acid ligand
> Hello!

Hi Thomas,

> We would like to run FEP on some amino acids in the binding pocket of
> our amino acid transporter to get an idea of what accounts for the
> observed binding affinities but when I'm running the FEP calculation
> the simulation blows up within a few hundred timesteps. At constant
> pressure, NAMD complains about the periodic cell having become to
> small, and at constant volume one sees atoms moving too fast.

This usually indicates that the periodic box is to small from the start.
Check if you got some atoms out of bounds. Remember that if the box is set
to small and the simulation starts, the atoms will be treated as wrapped and
so maybe superimposed with other atoms, or too close. This would cause the
two situations you mentioned.

> So far, I have built a topology of the amino acid ligand with the help
> of the hybrid topologies (top_all27_hybrid.inp). I also have a model
> of the transporter with its ligand in the binding pocket. With the
> coordinates of the ligand and the internal coordinates in the topology
> file I can build a .psf and .pdb file for the new ligand, and I have
> also built a new model, consisting of the transporter with the FEP
> ligand.
> The ligand is glycine, which is transformed into alanine by turning
> the CH2 group into a CH(CH3) group. These are the alchemical
> parameters:
> alch on
> alchType fep
> alchEquilSteps 5000
> alchFile ALA.fep
> alchCol B
> alchOutFreq 10
> alchOutFile ALA.fepout
> alchDecouple off
> alchVdwLambdaEnd 1.0
> alchElecLambdaStart 0.5
> alchVdWShiftCoeff 6.0
> alchLambda 0
> alchLambda2 .05
> run 10000
> However, when running a normal simulation with the new coordinates
> (the glycine ligand without the exnihilated atoms), this simulation
> runs just fine.
> Something is wrong here, and I cannot see what it is.
> Thomas

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:20:52 CST