Re: [AMBER] trajectory of minimization using generalized Born

From: Jason Swails <jason.swails.gmail.com>
Date: Fri, 20 Mar 2015 09:23:50 -0400

> On Mar 20, 2015, at 9:13 AM, Edwin Helbert Aponte Angarita <helbert2a.gmail.com> wrote:
>
> Thanks Jason,
>
> Using:
> find /gpfs/apps/amber/amber12.0/ -iname \*patch\*
> I didn't find $AMBERHOME/.patches/Amber_Applied_Patches neither
> patch_amber.py. I will ask what patches have been applied. I also did more
> testing and got interesting results.
>
> I want to simulate a peptide, and I am trying to simulate with and without
> counter ion (has net charge of +1). In summary, the system with counter ion
> minimized, but the one without counter ion worked with 8 processors but not
> with 16.I am using these scripts with tleap -f

Counter ions are inappropriate for systems without periodic boundary conditions. With GB, salt effects are accounted for by a Debye screening parameter.

>
> Without counter ion:
>
> source /gpfs/apps/amber/amber12.0/dat/leap/cmd/leaprc.ff99SB
> AcA14KG3A14K = sequence { ACE ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA
> ALA ALA ALA LYS GLY GLY GLY ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA
> ALA ALA CLYS }
> saveoff AcA14KG3A14K AcA14KG3A14K_linear-ff99SB-noCion.lib
> savepdb AcA14KG3A14K AcA14KG3A14K_linear-ff99SB-noCion.pdb
> saveamberparm AcA14KG3A14K AcA14KG3A14K-ff99SB-noCion.prmtop
> AcA14KG3A14K-ff99SB-noCion.inpcrd
> quit
>
> With counter ion:
>
> source /gpfs/apps/amber/amber12.0/dat/leap/cmd/leaprc.ff99SB
> AcA14KG3A14K = sequence { ACE ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA
> ALA ALA ALA LYS GLY GLY GLY ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA ALA
> ALA ALA CLYS }
> addions AcA14KG3A14K Cl- 0
> saveoff AcA14KG3A14K AcA14KG3A14K_linear-ff99SB-CIon.lib
> savepdb AcA14KG3A14K AcA14KG3A14K_linear-ff99SB-CIon.pdb
> saveamberparm AcA14KG3A14K AcA14KG3A14K-ff99SB-CIon.prmtop
> AcA14KG3A14K-ff99SB-CIon.inpcrd
> quit
>
> And the following mdin file:
>
> &cntrl
> imin=1,
> maxcyc=2000,
> ncyc=1000,
> cut=999.,
> rgbmax=999.,
> igb=1,
> ntb=0,
> ntpr=100,
> ntwx=100,
> ioutfm=1,
> /
>
> I am setting -O -i -o -p -c -r -x properly and these are the results
>
> Without counter ion, 16 processors: failed with ERROR IN SETPAR()
> Without counter ion, 8 processors: succeeded, all minimization steps
> completed
> With counter ion, 16 processors: Succeeded, all minimization steps completed
>
> Can you reproduce this error?

I'll try later when I'm at my computer, but a couple comments. First, you should tell us what program you are actually using. I'm guessing sander, but I can't recall where you actually stated.

Second, the igb=1 model is very old, and vastly superior GB models (like igb=8) are available and better suited for protein and peptide simulations.

Finally, your system is quite small. With only 34 residues, I wouldn't expect your simulation to scale particularly well with more than 8 processors. You can also try using pmemd to see if that can circumvent the problem. But you might be stuck using only 8 CPUs (although I encourage you to do some quick benchmarking to find the optimum number of processors).

HTH,
Jason

--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Mar 20 2015 - 06:30:06 PDT
Custom Search