RE: AMBER: QM-MM output interpretation

From: Ross Walker <>
Date: Mon, 30 Jul 2007 21:11:50 -0700

Hi Francesco,

> PM3ESCF= -259.7765
> PM3ESCF= -257.1894

> Judging from QM-MM PM3ESCF, a difference of 2.59 kcal/mol
> between the two
> conformers is not too far from the experimental value (NMR)
> of ca 1.9. Though -
> after what you explained in detail for this type of system -
> is that average
> /\PM3ESCF an estimate on which to rely?

This is a tricky one. The key is whether the contributions to the ESCF
energy cancel out so that what you recover when you take the difference of
the averages is the difference between the average conformer one energy and
the average conformer 2 energy within solvent. I.e. does all the 'noise'
from the water etc cancel out. I'm really not sure on this as it would take
working through all the equations to determine this for sure. Maybe others
on the list can chime in here. One key point if you wanted to use such
analysis would be to write to the output file more often and calculate a
cummulative average that you then plot. This would tell you if the average
was converging or not. At the moment you have no idea if you have done
enough sampling or not - I.e. take the cumulative average every 100 steps or
so and see what the fluctuation in this average is with time.

Do you know what the experimental NMR value was actually measuring?

Ultimately it might be better to do some form of thermodynamic integration
where you convert one conformer into the other or make one vanish and the
same with the other.

> I am at Amber just to have dynamics. I have already carried
> out DFT at high
> functional and basis set level, including vibrational
> analysis and temperature
> correction, all in vacuum.

Okay, so this makes sense then since here you are obtaining solution phase
data - I would certainly start by running longer simulations - you likely
want at least 100ps or so I would guess although as above it would be
important to look at the average energy as a function of time. You might
want to try reposting this message along more general lines about what you
can learn by comparing average potential energies since a number of people
who might be able to offer more insight might be skipping this thread
thinking it is specific to QM/MM but really the same argument here would
apply to a classical simulation although the "noise" on EPTOT may be so
large that you need a significant amount of sampling to obtain an accurate
difference. This is the "noise" that may be cancelling out when you just
look at the ESCF value.

Good luck...


|\oss Walker

| HPC Consultant and Staff Scientist |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- |
| | PGP Key available on request |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Aug 01 2007 - 06:07:30 PDT
Custom Search