AMBER Archive (2007)

Subject: Re: AMBER: Positive binding free energy!

From: David Cerutti (dcerutti_at_mccammon.ucsd.edu)
Date: Tue Feb 13 2007 - 18:40:35 CST

It's perfectly possible that you get a positive binding free energy from
MM-PBSA analysis. It's even possible to obtain more negative binding free
energies from a three-trajectory approach (simulate A, simulate B,
simulate complex AB) than from a one-trajectory approach (see Gohlke and
Case, J. Comp. Chem. 25.2:238-250, page 246 center left hand column).

The MM-PBSA estimator is very imprecise and, perhaps unfortunately, gives
results to within 5kcal/mol of the correct binding energy about 30-40% of
the time. What this can translate into is investigators doing one study
(as you've probably seen, they're a fair bit of work) and publishing a
successful result or declining to submit a negative result. We tried
running a series of mutations from the same protein system, starting from
known (perhaps more importantly *different*) crystal structures for almost
all of our simulations, and obtained a 30% success rate, no correlation in
the answers we got, and unsatisfactory convergence even after 1500
snapshots over 30ns of simulation on each system.

It's a good exercise for teaching molecular dynamics and some aspects of
the solvation problem, but without a significant enhancement (which may in
fact exist) it's no magic bullet.

Dave

On Tue, 13 Feb 2007, Rima Chaudhuri wrote:

> Hi All,
>
> I ran some mm_pbsa calculations, and am getting positive deltaG(bind) i.e.
> the PBTOT = -30.69 and the TSTOT= -36.97, which is not possible, because the
> Ki for the ligand is about 6nM. I am getting similar positive values for
> some four other protein-ligand complexes. I saw a similar message posted,
> but that was on a docked complex and might have had ligand conformation
> problem. I am running the calculation on the MD performed on crystal
> structures, using MM=1, PB=1, MS=1 over 100 snapshots from the last 200ps of
> the 1ns simulation. I used 10 snapshots of those 100 for entropy
> calculations. Any suggestions on what might have gone wrong? and what might
> be a reasonable explanation..
>
>
> # DELTA
>
> # -----------------------
>
> # MEAN STD
>
> # =======================
>
> ELE -64.04 14.86
>
> VDW -60.27 3.27
>
> INT -0.00 0.00
>
> GAS -124.31 14.98
>
> PBSUR -6.72 0.10
>
> PBCAL 100.35 14.04
>
> PBSOL 93.63 14.01
>
> PBELE 36.30 4.41
>
> PBTOT -30.69 4.46
>
>
>
> # DELTA
>
> # -----------------------
>
> # MEAN STD
>
> # =======================
>
> TSTRA -13.43 0.00
>
> TSROT -11.61 0.01
>
> TSVIB -11.93 19.26
>
> TSTOT -36.97 19.27
>
>
>
> Thanks.
>
>
>
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu