AW: problems running amber parm7

From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Tue Jan 20 2015 - 00:19:40 CST

 

Von: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag von Francesco Pietra
Gesendet: Montag, 19. Januar 2015 20:04
An: NAMD
Betreff: namd-l: problems running amber parm7

 

Hello:

 

Hey

 

I am experiencing problems in running amber parm7/rst (parm10) in periodic box tip3p solvation. I used the same settings that proved efficient in past work with parm9 for a protein with a metallic center:

 

 

# Approximations for nonbonded interactions

cutoff 9 # for vdW and elctrostatic interactions

switching on

switchDist 6 # < cutoff

pairListDist 11 # between pair of atoms for vdW and electrost

outputpairlists 1000

margin 0

 

These settings are unusual, defaults are:

 

cutoff 10

switchdist 9 #cutoff -1

pairlistdist 12 #cutoff+2

 

# Multistep integrator parameters

timestep 1.5 # 1.5 fs/step

nonbondedFreq 2 # nonbonded forces every two steps

stepspercycle 20 # redo pairlist every 20 steps

fullelectfrequency 2 # number of timesteps between electr evaluation

 

A timestep of 2fs would be possible when using rigidbonds all.

 

# PME settings

PME on

PMETolerance 1.0e-6

PMEInterpOrder 4 # i.e. cubic spline

PMEGridSizeX 147 # size of fftw grid on x dimension

PMEGridSizeY 115

PMEGridSizeZ 102

PMEGridSpacing 1.0

 

Settings PMEGridSpacing and PMEGridSize* is redundant.

 

# periodic settings

# Don't set the periodic cell basis if you have also specified an .xsc

# cellBasisVector1 138.96 0. 0.

# cellBasisVector2 0. 105.38 0.

# cellBasisVector3 0. 0. 91.97

cellOrigin 71.08314514160156 54.25843811035156 47.91973114013672

 

# output

outputName ./min-05

outputEnergies 500 ;# multiple of fullElectFrequency or viceversa

restartfreq 100

binaryrestart yes

binaryoutput no

# wrapNearest no

# wrapAll on

 

Your restart freq is way too low. You are writing a restart every approx. minute.

 

seed 3971

 

Why you set the seed manually?

 

# Minimize protocol (steps multiple of stepspercycle)

# minimization on # default off

minTinyStep 1.0e-6 # default 1.0e-6

minBabyStep 1.0e-2 # default 1.0e-2

minLineGoal 1.0e-4 # default 1.0e-4

 

Why pumping up the script with so much values that are set to default anyway?

 

velocityQuenching on # default off

maximumMove 1.5 # default 0.75 x cutoff/stepsPerCycle = 0.5

 

 

Minimization proved difficult so that I went to velocityQuenching. That went on well from ts/VelQuen 0.01/0.01 to 1.5/1.5, the last for 1000 steps. Continuing under the latter conditions caused notable denaturation of the protein (loss of alpha helicitiy) while the water box, from cubic became nearly spherical.

 

Now to you actual problem. Away from the comments above, especially cutoff and switchdist, you have changed your parameters to parm10, so it’s not unlikely that your molecule behave slightly different (have you checked what changes have been made by the amber folks). But given the fact that you were not able to simply minimize your system, I’d suspect that something is wrong with your initial input (close contacts etc.). What exactly prevents you from doing a standard minimization (error message)? If definitely everything is ok with your system, than you’d like to put some hbond restraints (extrabonds) to your minimization (and maybe equilibration) to prevent the protein from unfolding.

 

The enzyme comprises several chains and a transition-metal active center, bound to the protein backbone through covalent carbon-carbon bonds. The conformation of the active center is maintained well until min-05 as the protein above, then it suffers somewhat from the protein degradration.

 

I wonder whether more appropriate settings could be used. It is important to mention that no denaturation of the protein was observed under implicit conditions (GBIS/SASA) even on heating to 300K.

 

Thanks for advice

francesco pietra

 

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:21:33 CST