From owner-chemistry@ccl.net Fri Sep 12 05:48:00 2008
From: "Vincent Leroux vincent.leroux]![loria.fr" <owner-chemistry%%server.ccl.net>
To: CCL
Subject: CCL: comparison between 2 different structures
Message-Id: <-37744-080912045649-5282-8teP1pLOtMmKbNL5aCd0IA%%server.ccl.net>
X-Original-From: Vincent Leroux <vincent.leroux\a/loria.fr>
Content-Transfer-Encoding: 8bit
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Date: Fri, 12 Sep 2008 10:56:46 +0200
MIME-Version: 1.0


Sent to CCL by: Vincent Leroux [vincent.leroux!A!loria.fr]
Hi Vivek,

You are trying to get the absolute low energy state of two similar 
molecules, this means you try to find their state in vacuum at 0K. I 
suppose you are in a protein-ligand context, bear in mind your small 
molecules will act on their target at 300K, your system being 
solvated... You do not have a single conformation then, but an ensemble 
of thermodynamically-accessible states (landscapes as you say). You are 
right being cautious as using only the (assumed) lowest energy state to 
describe the global "stability" of your molecules is a bit oversimplified.

As you mention protein-ligand complexes I suppose you are in the drug 
design context. Rather than performing complex minimizations / 
conformational searches it could be more useful to compute basic 
properties to check Lipinski's rule-of-five, try toxicity prediction 
algorithms if you have them, make sure the chemists can actually make 
the compounds, and so on... No molecular modelling at this point - you 
just want to check that there is nothing stopping you from discarding 
your candidates before going further. Drug designers do not want to 
waste their time on useless compounds  ;-)

When you study protein-ligand complexes, bear in mind that the free 
energy of binding is what describes best the strength of the 
protein-ligand interaction, the interaction energy being only a small 
part of it (the simplest to compute indeed)... Calculating free energy 
of binding rigorously is very difficult - get the "Free energy 
calculations" book edited by C.Chipot and A.Pohorille if you want to try.

You can compare interaction energy between two ligands bound to the same 
protein, this is similar to ranking compounds according to docking 
scores after a virtual screening experiment, but...
- is your single protein-ligand conformation enough to describe the 
entire energy landscape of the system?
- is the complex stable at 300K and solvated?
- is the forcefield used to compute the interaction energy (= nonbonded 
terms) okay?
- how flexible the protein active site can be upon ligand binding 
(induced fit effects)?
- are you sure you located the active site correctly?
- is the protein geometry reliable? (resolution of experimental data? 
homology modelling details?)
- is it possible that discrete water molecules actively take part in the 
protein-ligand binding?
- are the free energy of binding terms you do not consider of the same 
order between the two ligands?
- ...

Performing at least 1 ns of MD (CHARMM or AMBER FF) of the two 
protein-ligand complexes in explicit solvent, 300K, no constraint, can 
answer many questions (but not all - given in many cases it can answer 
***enough*** questions, this is fine). You should search for papers at 
this point to gain knowledge, before planning such an heavy calculation. 
Remember that at best you will be able to state that ligand A has a 
significant probability of being more potent to target X than ligand B. 
You can always be wrong - there are so many aspects of protein-ligand 
binding that are not taken into account at this stage...

Good luck  :-)
VL


Vivek Sharma vivek.viv.sharma::gmail.com a �crit :
> Sent to CCL by: "Vivek  Sharma" [vivek.viv.sharma ~ gmail.com]
> Dear CCL people,
>
> I have one conceptual question to ask, may be its trivial but all your comments will be very helpful in understanding more about this field and will obviosuly correct some doubts.
>
> Lets say I have two different ligand molecules (different substituents on some common core). I sketch them, do conformational analysis on those and minimise the energy of both the systems separately and finally some values of total potential energy are obtained. And, if one value is lower than the other, can I say that this molecule (with lower potential energy) is more stable than the other one?
> I have mixed thoughts about it. I would assume if there have been a single compound, one can compare the potential energies of the compound in two different tautomeric states for example. But in case of two different compounds, which will have different potential energy landscapes, etc, it may not be possible.
> My this query also scales-up to a docking study where a similar question arises, i.e. can I compare the Binding Energy or Interaction Energy between two different Protein Ligand complexes (where protein is same in both, but ligands are chemically different).?
>
> thanks in advance,
>
> Vivek
> IMM, Pune, India.
>
>