AMBER Archive (2007)

Subject: Re: AMBER: High Standard Deviation for nmode (TSVIB) mm_pbsa

From: Rima Chaudhuri (rima.chaudhuri_at_gmail.com)
Date: Wed Feb 28 2007 - 16:25:54 CST

hello Dr. Case, As per your suggestion (please see email below), I ran the
nmode calculations separately for 5 different snapshots instead of using the
mm_pbsa-nmode script, in the hope of getting better/lower STD.
The results from the initial nmode calculation was:

# COMPLEX RECEPTOR LIGAND
# ----------------------- -----------------------
-----------------------
# MEAN STD MEAN STD
MEAN STD
# ======================= =======================
=======================
TSTRA 17.47 0.00 17.46 0.00 13.89
0.00
TSROT 17.83 0.01 17.81 0.00 12.00
0.01
TSVIB 2778.59 12.75 2721.74 13.33 47.17
1.63
TSTOT 2813.90 12.75 2757.01 13.33 73.06
1.63

# DELTA
# -----------------------
# MEAN STD
# =======================
TSTRA - 13.87 0.00
TSROT -11.98 0.02
TSVIB 9.69 16.63
TSTOT -16.17 16.64
~
In an attempt to lower the STD, post-calculation results of the individual
snapshots are as follows; ofcourse, I converted the S(cal/mol-K) reported in
the nmode.out to Kcal/mol and made the following table for the 6 individual
snapshots:

  Complex(TStotal)

STD

Receptor

STD

Ligand

STD

T deltaS

STD

2843.482

9.158574

2795.477

7.9831

69.56

0

-24.37

6.521333

2825.911

2783.296

69.56

-36.13

2816.364

2782.934

69.56

-26.94

2823.572

2778.389

69.56

-21.55

2830.805

2784.739

69.56

-23.49

2824.158

2771.2

69.56

-16.6

 mean

2827.382

2782.673

69.56

-24.84667

I checked the nmode output file for all of them, the first 6 modes are all
close to 0. I visualized the eigenvectors in IED plugin of VMD, nothing
seems abnormal (either the receptor alone or the complex). If I run the
nmode calculation using mm_pbsa, the standard deviation is high due to the
contribution of TSVIB only, whereas on performing the individual
calculations, the result is not the same, I am very confused, since I am
using the same snapshots for both the calculations.
Also, although I can lower the STD of NM by this method from about 16 to 6,
the overall deltaG(bind) still remains to be positive, i.e my TdeltaS is
always a higher negative number if not almost equal to the PBtotal value for
all the different ligand-protein complex systems.....any ideas why? From
experimental results, I know the inhibitors binds in the nanomolar range.

I ran the simulation of the solvated system for 5ns (about 3700 atoms w/o
water). The protein has a Ca2+ ion and a bound uncharged inhibitor. 200
snapshots from the last 2ns were used for MM_pbsa, whereas the last 6 frames
were used for nmode calculation. Please let me know what kind of details
about the system might help you in making a judgement.
Any suggestion would help a lot....

Thanks
Rima

On 2/17/07, David A. Case <case_at_scripps.edu> wrote:
>
> On Sat, Feb 17, 2007, Rima Chaudhuri wrote:
>
> > I looked at the structure in vmd and nothing looks weird
> > post-simulation, hence I doubt if there is an outlier.
>
> I think you need to look at the nmode outputs themselves, study the
> distribution of low-frequency eigenvalues, and see if you can spot
> anything.
> Not knowing anything about your system, I can't say from my vantage point
> what might be happening. Of course, what you see could be real, as well.
>
> ...dac
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> 

-- 
-Rima

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu