Dear all,
Just as a follow up on this ...
I did process the entire trajectory with cpptraj ... Significantly
faster than ptraj ...
The results however appear a bit better but still far off ... Now I get
a dS around -1 kcal/mol*K ... Which still translates in a TdS of about
+300 kcal/mol ..
With these values, the complete dG I calculate with MMPBSA is around
+210 kcal/mol (linear PB, inp=2) or +280 (non-linear PB, inp=2) ...
These values although better than the +700 I got with 8000 frames are
better but still don't make too much sense ...
Below is a output snapshot ... I am guessing I am still far from
convergence ... Also, maybe the fact that I am using a single trajectory
might also be the problem .. Although intuitively I would expect the
free receptor and ligand to have even higher values for the entropy if
simulated separately ... This would make the dG estimation even more off ...
Best wishes
Vlad
*******************
- Thermochemistry -
*******************
temperature 298.150 kelvin
pressure 1.00000 atm
molecular mass (principal isotopes) 50699.61600 amu
principal moments of inertia (nuclei only) in amu-A**2:
12814035.26 33604568.14 37705176.96
rotational symmetry number 1
Warning-- assumption of classical behavior for rotation
may cause significant error
rotational temperatures (kelvin) 0.00000 0.00000 0.00000
zero point vibrational energy 67824687.0 (joules/mol)
16199.88034 (kcal/mol)
25.8325510 (hartree/particle)
Warning-- 8185 vibrations have low frequencies and may represent hindered
internal rotations. The contributions printed below assume
that these
really are vibrations.
warning: setting vibrational entropy to zero for mode 19056 with vtemp
= -8049467.209940
warning: setting vibrational entropy to zero for mode 19057 with vtemp
= -6114803.608574
warning: setting vibrational entropy to zero for mode 19058 with vtemp
= -4648026.928337
warning: setting vibrational entropy to zero for mode 19059 with vtemp
= -395044.096191
freq. E Cv S
cm**-1 kcal/mol cal/mol-kelvin cal/mol-kelvin
--------------------------------------------------------------------------------
Total 57372.075 -nan 23444.731
translational 0.888 2.979 58.245
rotational 0.888 2.979 62.369
vibrational 57370.299 -nan 23324.117
1 0.178 0.592 1.986 16.008
2 0.242 0.592 1.986 15.392
3 0.349 0.592 1.986 14.665
4 0.411 0.592 1.986 14.345
5 0.609 0.592 1.986 13.562
6 0.701 0.592 1.986 13.282
7 0.854 0.592 1.986 12.891
8 0.940 0.592 1.986 12.699
9 0.980 0.592 1.986 12.616
10 1.110 0.592 1.986 12.369
11 1.202 0.592 1.986 12.211
12 1.265 0.592 1.986 12.110
13 1.302 0.592 1.986 12.053
14 1.379 0.592 1.986 11.939
On 10/29/2013 04:53 PM, Vlad Cojocaru wrote:
> Thanks Jason, Thanks Daniel,
>
> I was indeed talking only about ptraj processing .. Simulation is run
> already .. 150 ns = 50000 frames.
>
> Of course, I understand that its only one matrix to diagonalize
> regardless of the number of frames .. However, processing 8K frames
> versus 40K frames does make a huge difference in ptraj. with 8K frames
> my job was done ovenight (8-9 hours), whereas when I submitted the same
> job with 40 K frames, after 2 days I was processing only about 25 % of
> the frames. So, the bottleneck is not in the digonalization, but I guess
> in the generation of the matrix ...
>
> I will switch to cpptraj ... I used ptraj initially just because
> MMPBSA.py does it ..
>
> Best,
> Vlad
>
> On 10/29/2013 04:05 PM, Jason Swails wrote:
>> On Tue, Oct 29, 2013 at 9:50 AM, Vlad Cojocaru <
>> vlad.cojocaru.mpi-muenster.mpg.de> wrote:
>>
>>> yes, this was a test with 8000 frames ..
>>>
>>> The complex has 6353 atoms
>>> The receptor has 3788 atoms
>>> The ligand has 2565
>>>
>>> My trajectory has in total 50000 frames .. I wanted to exclude the first
>>> 10000. So I'd be left with 40000 frames .. This would make about
>>> 6*Natoms of the largest system. However this is extremely expensive for
>>> a 1 core job ...
>>>
>>> Maybe I could give it a try with the parallel ptraj.
>>>
>> I don't understand here, are you talking about the original simulation with
>> sander/pmemd or analyzing 40K frames with ptraj? From the sounds of it,
>> you're talking about ptraj? Parallel ptraj will be of no help here (I
>> don't think this action is even parallelized, anyway).
>>
>> With a quasi-harmonic approximation you are creating and diagonalizing only
>> ONE matrix, regardless of how many frames you use. Using 40K frames in a
>> quasi-harmonic analysis vs. 8K frames will not make a huge difference to
>> the processing time, I don't think.
>>
>> Good luck,
>> Jason
>>
--
Dr. Vlad Cojocaru
Max Planck Institute for Molecular Biomedicine
Department of Cell and Developmental Biology
Röntgenstrasse 20, 48149 Münster, Germany
Tel: +49-251-70365-324; Fax: +49-251-70365-399
Email: vlad.cojocaru[at]mpi-muenster.mpg.de
http://www.mpi-muenster.mpg.de/research/teams/groups/rgcojocaru
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Oct 31 2013 - 02:30:02 PDT
