[AMBER] Segmentation fault in MMPBSA.PY with two parts receptor

From: Amor San Juan <amorsanjuan.yahoo.com>
Date: Tue, 20 Jul 2010 03:16:57 -0700 (PDT)

Thank you Jason for the reply.

After checking the spaces at "/" as well as checking the compatibility of the topology files, the mmpbsa.py finished the job. The problem I face now is the 'segfault' error from sander9 when it calculates the ligand coordinates (which in this case is a peptide with 12 residues).

The lsf.out show:

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source
sander             00000000004304C0  Unknown               Unknown  Unknown
sander             0000000000734566  Unknown               Unknown  Unknown
sander             00000000004B076B  Unknown               Unknown  Unknown
sander             00000000004A3935  Unknown               Unknown  Unknown
sander             00000000004A10E6  Unknown               Unknown  Unknown
sander             000000000049C928  Unknown               Unknown  Unknown
sander             00000000004038AA  Unknown               Unknown  Unknown
libc.so.6          0000003FAFC1C3FB  Unknown               Unknown  Unknown
sander             00000000004037EA  Unknown               Unknown  Unknown

When I checked the output file:

| Run on Tue Jul 20 17:17:32 SGT 2010

|Input file:
|MMPBSA.py input file for running GB
|   startframe=5000, endframe=5050,
|   verbose=2, interval=1, strip_mdcrd=1,
|#  entropy=1,
|   igb=2, saltcon=0.0072
|Complex topology file:           ./ABC.prmtop
|Receptor topology file:          ./AC.prmtop
|Ligand topology file:            ./B.prmtop
|Initial mdcrd(s):                ./complex.mdcrd
|Best guess for receptor mask:   ":1-187:200"
|Best guess for  ligand  mask:   ":188-199"

|Calculations performed using 51 frames.
|All units are reported in kcal/mole.

The input file generated by the mmpbsa.py shows:
 Here is the input file:

File generated by MMPBSA.py
   ntb=0, idecomp=0, cut=999.0, nsnb=99999,
   imin=5, maxcyc=0, ncyc=0, igb=2,
   saltcon=0.0072, intdiel=1, extdiel=78.3,
   gbsa=2, surften=0.0072,

I do not think there is an error from this input file generated by mmpbsa.py.

Next, I checked the *complex_gb.mdout file:

  NSTEP       ENERGY          RMS            GMAX         NAME    NUMBER
      1       4.1603E+11     2.0384E+11     1.3011E+13     CB        977

 BOND    = *************  ANGLE   =   373559.8688  DIHED      =    10267.5296
 VDWAALS = *************  EEL     =     -913.1742  EGB        =   -28522.9796
 1-4 VDW =  3271043.7734  1-4 EEL =      -66.7265  RESTRAINT  =        0.0000
 ESURF   =      830.3698
minimization completed, ENE= 0.41603215E+12 RMS= 0.203844E+12
minimizing coord set #     3

The program sander was not able to calculate the BOND and VDWAALS terms. Why ? The values of the other terms are quite big ! For the *receptor_gb.mdout and *ligand_gb.mdout, values are strangely the same as above.

Could it be that the something wrong happened during the generation of snapshots? But as one can see in the output file, the mmpbsa.py automatic guesser was able to guess or assign correctly the receptor and ligand. When I viewed by vmd the *_avgcomplex.pdb file shows erratic with bonds broken all over upside down !

I am at lost here. If the only cause of the problem is segmentation fault, even I issued a command at console for 'ulimit -s unlimited' it didnt have any effect. And since the generated snapshots were incorrect, how to fix it knowing mmpbsa.py has assigned well the ligands, receptor.

Please be patient to this post ! I believe that the solution is out there. Thanks a lot for taking time.


AMBER mailing list
Received on Tue Jul 20 2010 - 03:30:03 PDT
Custom Search