[AMBER] MMPBSA.py.MPI: questions about output warnings and discrepancies

From: Kasprzak, Wojciech (NIH/NCI) [C] <"Kasprzak,>
Date: Fri, 11 May 2018 13:59:40 +0000

      Dear Amber Users,

 I have untangled my earlier problem with matching topologies to trajectories while trying to run
MMPBSA,py.MPI ptotocol on 3 PME trajectories for an RNA/ligand, RNA alone and ligand alone.
(No takers on my post from May 8)

I have extracted the subset of the whole system (RNA interface and ligand) into three trajectories
with all the solvent and excess RNA removed, and matched them to ante-MMPBSA.py-generated
topology files, effectively using the same topologies as input for all the solvated/vacuum input pairs
(-sp & -cp, -srp & -rp, and -slp & -lp).

MMPBSA.py.MPI (Amber 16) calculating the ENTROPY of the system runs to apparent completion
(progress.log and FINAL-RESULTS.dat file look normal) but with a warning in the batch error output file:

make_trajs.py:631: UserWarning: Solvated topology 0 has IFBOX == 0,
followed by a lot of messages of the type:
Line minimizer aborted: step at upper bound 0.005123168 (actual number vary, of course)

Question 1:
Since my topology files have no solvent and no Periodic Box sections, the IFBOX == 0 seems to make
sense, but is it a serious issue and is it related to all the line minimizer messages? What do they mean:
convergence of fatal error?

Question 2:
Also, I accidentally generated a set of input trajectories with the PBox info left in them (by not
using the nobox option in cpptraj output) and ran the MMPBSA.py.MPI with the topology files without the PBox info.
That run completed with with same normal-looking output files and identical warning in
the error output file, but with a ~35% difference in the total ENTROPY, with the largest difference due
to the vibrational component of the RNA/ligand complex entropy. I am puzzled why this run worked,
and I'd like to verify that the results it generated are incorrect, or hear more about the need or no-need for
the PBOX information in the MMPBSA.py.MPI calculations.

Hope to clarify the remaining confusion about the binding free energy calculations,
Voytek Kasprzak

Wojciech (Voytek) Kasprzak (Contractor)
Analyst Programmer
Frederick National Laboratory for Cancer Research
Leidos Biomedical Research, Inc.
Post Office Box B
Frederick, Maryland 21702
Phone: 301-846-5537

AMBER mailing list
Received on Fri May 11 2018 - 07:30:02 PDT
Custom Search