AMBER Archive (2009)

Subject: RE: [AMBER] Thermodynamic Intergration problem

From: Xiaohui Jiang [Contractor] (xjiang_at_bioanalysis.org)
Date: Wed Mar 04 2009 - 11:00:06 CST

Hi Ross,

Thanks for the reply. It does work with end point MD. As I mentioned both structure are fine in "charge off" and "charge on" parts of TI simulations.

BTW, how to set the box info?

Thanks!

Xiaohui (Jeff) Jiang, Ph.D.
Research Scientist
Phone: (301) 619-1986
Fax: (301) 619-1983
 
DoD Biotechnology HPC Software Applications Institute
BHSAI/MRMC, ATTN: MCMR-TT
Building 363 Miller Drive
Fort Detrick, MD 21702-5012

-----Original Message-----
From: amber-bounces_at_ambermd.org [mailto:amber-bounces_at_ambermd.org] On Behalf Of Ross Walker
Sent: Wednesday, March 04, 2009 11:45 AM
To: 'AMBER Mailing List'
Subject: RE: [AMBER] Thermodynamic Intergration problem

Hi Xiaohui,

Have you checked that you can run either end point on it's own. I.e. not
doing TI just straight MD?

The code has hung at a point where it typically would if you were trying to
do a periodic boundary simulation which your input suggests you are since
you have ntb=1 but your inpcrd file does not contain any box information:

| peek_ewald_inpcrd: Box info not found in inpcrd

Hence the type of run you are doing is not compatible with your input
coordinates. You should check this.

I agree that we should probably work on a better error message here. The
issue is that in parallel it is not always easy to cleanly trap errors when
some of the processors are sitting at barriers etc hence why it just
waits...

Good luck,
Ross

> -----Original Message-----
> From: amber-bounces_at_ambermd.org [mailto:amber-bounces_at_ambermd.org] On
> Behalf Of Xiaohui Jiang [Contractor]
> Sent: Wednesday, March 04, 2009 7:36 AM
> To: amber_at_ambermd.org
> Subject: [AMBER] Thermodynamic Intergration problem
>
> Hi,
>
> I am using TI with soft-core to study relative binding free energy. I
> followed the tutorial and tested several additional cases with T4 lysozyme
> by making small changes in the ligand. Everything went well.
>
> Problem happened when I tried to "mutant" a protein residue (V111T) with
> the same ligand (benzene). The "charge off" and "charge on" parts have no
> problem since the v0 and v1 are still the "same" complex and only the
> charges are changing. The "vdw" part with soft-core caused problem. In the
> initial minimization step (SD with only 2 processors), the "sander.MPI"
> process hangs without producing any results. Both processes did not crash,
> just running without updating output files or producing the "mdinfo" file.
> I put the two output files in attachments.
>
> The command I started the run was:
>
> openmpirun_wrapper sander.MPI -ng 2 -groupfile group_mini_L0.5
>
> "groupfile group_mini_L0.5" contains:
> -O -i VDW_mdin_mini_v0_L0.5 -o VDW_mdout_mini_v0_L0.5 -p T4_comp.prm -c
> T4_comp.rst -r T4_comp_mini_v0_0.5.rst
> -O -i VDW_mdin_mini_v1_L0.5 -o VDW_mdout_mini_v1_L0.5 -p T4_V111T_comp.prm
> -c T4_V111T_comp.rst -r T4_V111T_comp_mini_v1_0.5.rst
>
> The difference in the two input parameter/coordinate files are only at the
> residue 111. The ligand, water and counter ions remain the same. The
> following are the difference in pdb format. Of course the atom numbering
> is different after residue 111.
>
> "original"
> ATOM 1736 N VAL 111 -6.428 0.036 -6.083 1.00 0.00
> ATOM 1737 H VAL 111 -6.147 -0.063 -5.118 1.00 0.00
> ATOM 1738 CA VAL 111 -6.477 -1.161 -6.915 1.00 0.00
> ATOM 1739 HA VAL 111 -5.773 -1.044 -7.739 1.00 0.00
> ATOM 1740 CB VAL 111 -6.057 -2.343 -6.088 1.00 0.00
> ATOM 1741 HB VAL 111 -6.796 -2.538 -5.310 1.00 0.00
> ATOM 1742 CG1 VAL 111 -5.954 -3.559 -7.009 1.00 0.00
> ATOM 1743 1HG1 VAL 111 -5.215 -3.365 -7.786 1.00 0.00
> ATOM 1744 2HG1 VAL 111 -5.650 -4.430 -6.429 1.00 0.00
> ATOM 1745 3HG1 VAL 111 -6.924 -3.750 -7.469 1.00 0.00
> ATOM 1746 CG2 VAL 111 -4.691 -2.036 -5.519 1.00 0.00
> ATOM 1747 1HG2 VAL 111 -4.748 -1.140 -4.901 1.00 0.00
> ATOM 1748 2HG2 VAL 111 -4.354 -2.875 -4.911 1.00 0.00
> ATOM 1749 3HG2 VAL 111 -3.986 -1.871 -6.334 1.00 0.00
> ATOM 1750 C VAL 111 -7.821 -1.439 -7.544 1.00 0.00
> ATOM 1751 O VAL 111 -7.926 -1.788 -8.734 1.00 0.00
>
> "V111T"
> ATOM 1736 N THR 111 -6.428 0.036 -6.083 1.00 0.00
> ATOM 1737 H THR 111 -6.147 -0.063 -5.118 1.00 0.00
> ATOM 1738 CA THR 111 -6.477 -1.161 -6.915 1.00 0.00
> ATOM 1739 HA THR 111 -5.773 -1.044 -7.739 1.00 0.00
> ATOM 1740 CB THR 111 -6.057 -2.343 -6.088 1.00 0.00
> ATOM 1741 HB THR 111 -5.819 -2.008 -5.078 1.00 0.00
> ATOM 1742 CG2 THR 111 -7.209 -3.362 -6.026 1.00 0.00
> ATOM 1743 1HG2 THR 111 -6.905 -4.219 -5.426 1.00 0.00
> ATOM 1744 2HG2 THR 111 -7.456 -3.694 -7.034 1.00 0.00
> ATOM 1745 3HG2 THR 111 -8.084 -2.895 -5.573 1.00 0.00
> ATOM 1746 OG1 THR 111 -4.900 -2.956 -6.667 1.00 0.00
> ATOM 1747 1HG THR 111 -5.124 -3.277 -7.641 1.00 0.00
> ATOM 1748 C THR 111 -7.821 -1.439 -7.544 1.00 0.00
> ATOM 1749 O THR 111 -7.926 -1.788 -8.734 1.00 0.00
>
> I tried to run this with different memory setup 1.75Gb or 3.5Gb. There is
> no difference. Also, I tried to bypass minimization and started MD
> directly. Similar problem encountered. The "sander.MPI" tested was rebuilt
> on Feb 24, 2009 with latest patches.
>
> I can provide the files to run the jobs if someone would like to take a
> look at the problem.
>
> Thanks in advance!
>
> Regards,
>
> Jeff
>
> Xiaohui (Jeff) Jiang, Ph.D.
> Research Scientist
> Phone: (301) 619-1986
> Fax: (301) 619-1983
>
> DoD Biotechnology HPC Software Applications Institute
> BHSAI/MRMC, ATTN: MCMR-TT
> Building 363 Miller Drive
> Fort Detrick, MD 21702-5012

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber