Re: [AMBER] MMPBSA.py nmode maxcyc problem

From: Jason Swails <jason.swails.gmail.com>
Date: Fri, 11 Jun 2010 00:19:47 -0400

Hello Niel,

I've attached the patch that will prevent MM/PBSA from increasing the number
for maxcyc. Apply this by moving mmpbsa.patch to $AMBERHOME and running the
command

patch -p0 < mmpbsa.patch

Then recompile MM/PBSA, but make sure that parallel nab has not been built
(this will not work with MMPBSA.py.MPI, since the MPI in nab tends to get
confused with the MPI in the python script. One other thing I didn't
mention is that the minimization stops when the gradient is less than the
specified tolerance, in case you didn't know that already.

All the best,
Jason

On Thu, Jun 10, 2010 at 8:01 PM, Jason Swails <jason.swails.gmail.com>wrote:

> I forgot to say this last time, but I think 1000 cycles is probably a
> little too small... I would suggest sticking with the default (10 000) or
> maybe going even a little higher (of course that depends on your system
> size...). It may take a while to converge to the default gradient (more
> than 1000, perhaps).
>
> All the best,
> Jason
>
>
> On Thu, Jun 10, 2010 at 7:24 PM, Jason Swails <jason.swails.gmail.com>wrote:
>
>> Hello,
>>
>> On Thu, Jun 10, 2010 at 10:51 AM, Niel Henriksen <niel.henriksen.utah.edu
>> > wrote:
>>
>>> I've been running gb, pb, and nmode analysis on an RNA/ligand system
>>> using MMPBSA.py.MPI on two 8-core nodes (16 CPUs total).
>>>
>>> my input file is:
>>>
>>> &general
>>> startframe=1, endframe=999999999, interval=1,
>>> receptor_mask=':1-38', ligand_mask=':39', strip_mdcrd=0,
>>> verbose=1, keep_files=1, entropy=1,
>>> /
>>> &gb
>>> igb=1, gbsa=1, saltcon=0.200,
>>> /
>>> &pb
>>> istrng=0.2, fillratio=4.0,
>>> /
>>> &nmode
>>> nmode_igb=1, nmode_istrng=0.2, nminterval=200, maxcyc=1000,
>>> /
>>>
>>> The gb and pb calculations are fine. Most of the nmode calculations are
>>> okay.
>>> But it looks like one of them won't finish. It reached the the maxcyc
>>> number of
>>> iterations and then increased maxcyc by a factor of 10. All the rest of
>>> the
>>> nmode tasks are finished except for this one and 10 hours later it is
>>> still running.
>>> I have had several jobs do this and they keep running indefinitely.
>>>
>>> First, why have a maxcyc if the program just increases it if it is
>>> reached?
>>>
>>
>>> Second, can you tell from the output what is going on with the
>>> calculation? It
>>> looks like the CG minimization is dying. Is it probably just a bad
>>> structure that
>>> won't minimize nicely? See out put at the bottom of this email.
>>>
>>
>> I can't tell... I don't have much experience with xmin output through nab
>> programs... The output is described on page 209 of the AmberTools 1.4
>> manual.
>>
>>
>>>
>>> Third, how do I know which pdb was used for the calculations that
>>> produced
>>> _MMPBSA_complex_nm.out.5? (I guess I could search through the code, but
>>> I'll try asking here first.) Is it the sixth pdb produced by ptraj? In
>>> my case
>>> _MMPBSA_complex_nm.pdb.1001 ?
>>>
>>
>> It depends. The suffix on the .out specifies which processor is creating
>> the file, not specifically which frame it is. However, it is fairly simple
>> to determine which frame corresponds to which processor. The number of
>> frames is divided up equally amongst all processors. For n frames and m
>> processors, each processor takes n/m frames. If n/m is not an integer, then
>> for the number of remaining frames, that many processors (from rank 0 to
>> remainder - 1) does 1 extra frame. This is all done in the order that ptraj
>> writes out files. Moreover, processor 0 will take frames 0 - n/m (+1 if
>> there are remainders).
>>
>>
>>> Fourth, any suggestions for how to prevent the problem or at least
>>> terminate the program when the problem occurs?
>>>
>>
>> I will create a patch that does not increase the number of maxcyc if it's
>> reached and send that to you soon. You can either kill the job if it's in a
>> batch system or send it a termination signal (ctrl-C in unix, press it
>> repeatedly to exit all processes).
>>
>> Hope this helps,
>> Jason
>>
>>
>>> Thanks for the help
>>> --Niel
>>>
>>>
>>> OUTPUT FROM _MMPBSA_complex_nm.out.5:
>>>
>>> Reading parm file (com.topo)
>>> title:
>>>
>>> mm_options: ntpr=10000
>>> mm_options: diel=C
>>> mm_options: kappa=0.147010203727
>>> mm_options: cut=1000
>>> mm_options: gb=1
>>> mm_options: dielc=4.0
>>> mm_options: temp0=298.15
>>> scaling charges by 0.500
>>> iter Total bad vdW elect nonpolar genBorn
>>> frms
>>> ff: 0 -805.90 1966.92 -388.57 718.31 0.00 -3102.56
>>> 1.99e+01
>>> ________________________________________________________________
>>> MIN: It= 0 E= -805.90 ( 19.856)
>>> CG: It= 5 ( 0.413)q :-)
>>> LS: i= 1 lhs_f= -759.27688 rhs_f= -0.15304093
>>> lhs_g= 18.930947 rhs_g= 1377.3683
>>> LS: step= 1 it= 1
>>> MIN: It= 1 E= -1565.18 ( 5.645)
>>> CG: It= 3 ( 0.473)q :-)
>>> LS: i= 1 lhs_f= -99.283686 rhs_f= -0.019840063
>>> lhs_g= 0.17934537 rhs_g= 178.56056
>>> LS: step= 1 it= 1
>>> ....
>>> ....
>>> ....
>>> ....
>>> MIN: It= 9702 E= -254.26 ( 0.013)
>>> CG: It= 1 (999.999)q :-((
>>> LS: i= 1 lhs_f= 0.0010889731 rhs_f= -8.8553167e-08
>>> lhs_g= 0.0021649181 rhs_g= 0.0007969785
>>> LS: i= 2 lhs_f= 3.2271245e-05 rhs_f= -1.4808629e-08
>>> lhs_g= 0.00073717487 rhs_g= 0.0007969785
>>> LS: i= 3 lhs_f= 1.2717392e-06 rhs_f= -2.5536537e-09
>>> lhs_g= 0.00085998718 rhs_g= 0.0007969785
>>> LS: i= 4 lhs_f= 1.1181794e-07 rhs_f= -5.1091314e-10
>>> lhs_g= 0.00088042224 rhs_g= 0.0007969785
>>> LS: i= 5 lhs_f= 1.7803018e-08 rhs_f= -1.0507513e-10
>>> lhs_g= 0.00088448091 rhs_g= 0.0007969785
>>> LS: i= 6 lhs_f= 4.0517989e-09 rhs_f= -2.1713693e-11
>>> lhs_g= 0.00088531453 rhs_g= 0.0007969785
>>> LS: i= 7 lhs_f= 1.4933903e-09 rhs_f= -4.4752754e-12
>>> lhs_g= 0.00088548692 rhs_g= 0.0007969785
>>> LS: i= 8 lhs_f= 9.0960839e-10 rhs_f= -9.0468856e-13
>>> lhs_g= 0.00088552262 rhs_g= 0.0007969785
>>> LS: i= 9 lhs_f= 7.8819085e-10 rhs_f= -1.6832494e-13
>>> lhs_g= 0.00088552999 rhs_g= 0.0007969785
>>> LS: i=10 lhs_f= 7.7579898e-10 rhs_f= -2.1976357e-14
>>> lhs_g= 0.00088553145 rhs_g= 0.0007969785
>>> LS: i=11 lhs_f= 7.8921403e-10 rhs_f= -8.4068087e-16
>>> lhs_g= 0.00088553166 rhs_g= 0.0007969785
>>> LS: i=12 lhs_f= 7.6067863e-10 rhs_f= -1.4793792e-18
>>> lhs_g= 0.00088553167 rhs_g= 0.0007969785
>>> LS: i=13 lhs_f= 7.5692697e-10 rhs_f= -4.7951204e-24
>>> lhs_g= 0.00088553167 rhs_g= 0.0007969785
>>> LS: i=14 lhs_f= 7.5692697e-10 rhs_f= -8.8553167e-28
>>> lhs_g= 0.00088553167 rhs_g= 0.0007969785
>>> LS: step= 1e-20 it=14
>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>>
>>
>>
>> --
>> Jason M. Swails
>> Quantum Theory Project,
>> University of Florida
>> Ph.D. Graduate Student
>> 352-392-4032
>>
>
>
>
> --
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Graduate Student
> 352-392-4032
>



-- 
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032



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


Received on Thu Jun 10 2010 - 21:30:05 PDT
Custom Search