From:	Tom Halgren <halgren[at]schrodinger.com>
Date:	Wed, 28 Apr 2004 17:56:08 -0400 (EDT)
Subject:	Re: CCL:Validating docking codes

Hi, Dr. Juhos -

I'd like to point out that we did not use a subset of the ACD in our
general database enrichment studies for Glide.  It is true that we did
base our ligand database on a set of commercially available compounds that
we assembed.  But as Table 10 in the first Glide paper (J. Med. Chem.
2004, 47, 1739-1750) shows, we selected out a subset of average MW 400
that matched the properties of a subset of the WDI to which we did have
access for some 15 properties, including such things as the average
numbers of first and higher-row atoms, rotatable bonds, total rings and
heteroaromatic rings, charged and neutral hydrogen bond donors and
acceptors, and amide hydrogens. The effect of this selection can best be
judged by comparing the properties shown in the table for the "puchasable
compounds MW 350" and "drug-like compounds MW 360" sets.

Your comment actually refers to a very small part of our work -- namely,
to the comparisons we made in the second paper (pp. 1750-1759) to the
Bissantz work on thymidine kinase and the estrogen receptor.  In this
work, as stated, we intentionally used the Bissantz/Rognan ACS database
preparation, kindly supplied by Dr. Rognan.  We agree that this decoy
ligand set is not the most representative, and indeed these ligands
were not prepared in the manner we would prefer, but we used them so that
we could directly compare Glide to FlexX, GOLD and Dock.  Thus, these
comparisons are "evenhanded".

Actually, I might note, these comparision were not in the originally
submitted manuscript at all.  Rather, they were added to meet the
insistence of a referee that comparisons to other methods be included, and
this is what we were able to do, with Dr. Rognan's help, on short notice;
we don't, after all, have access to the competitive docking codes
ourselves.

In our own work, we are now preparing to switch our testing protocol to
use decoy sets extracted from the WDI, which we now license, and we will
be using both 1000-ligand and 10,000-ligand sets.


				Tom Halgren

--------------------------------------------------------------------------
Thomas A. Halgren
Chief Technical Officer                    voice: 646-366-9555 x 106
Schrodinger, Inc.                          fax:   646-366-9550
120 W 45th Street                          e-mail: halgren[at]schrodinger.com
New York, NY 10036                         www.schrodinger.com
--------------------------------------------------------------------------


> Hi all,
>
>   As you may have noticed, Szilveszter Juhos wrote to CCL today that we
> used the ACD ligand database in our enrichment studies using Glide.  This
> is incorrect.  Should somebody write to him?
>
> Mee Shelley
>
> ---------- Forwarded message ----------
> Date: Wed, 28 Apr 2004 09:59:28 +0200
> From: Szilveszter Juhos 
> To: Mark Thompson , chemistry[at]ccl.net
> Subject: CCL:Validating docking codes.
>
> On Tue, Apr 27, 2004 at 09:18:10PM -0700, Mark Thompson wrote:
> > One area where I'm weak is having good ligand libraries for testing database
> > enrichments.  I could construct these libraries myself, but then direct
> > comparisons with other docking/scoring methods cited in the literature might
> > be ambiguous.
>
> You are right, however, in the recent Glide paper (J. Med. Chem. 2004, 47,
> 1750-1759 by  Halgren et al.) they are using the same ACD set that was
> used by Bissantz et al. (J. Med. Chem. 2000, 43, 4759-4767) as far as I
> see. My opinion ACD is really not the best source of lead/drug-like
> entities. The FlexX people (J. Med. Chem. 2001, 44, 1035-1042)  used a
> subset of WDI. Having WDI (means having $$$) you can make some
> comparison though it is only the library creation process not the
> entities you can read from the article.
>
> Likely the best thing you can do is to generate your own libs
> in the same way as it was mentioned in the papers. OTOH, testing with
> 1000 or so non-native ligands is not necessarily enough: using vHTS usually
> the top 1% or less of the whole library is purchased and that is the
> first ten if you are using a 1K lib :/
>
> Cheers:
> Szilva
>
> -= This is automatically added to each message by the mailing script =-
> To send e-mail to subscribers of CCL put the string CCL: on your Subject: line
> and send your message to:  CHEMISTRY[at]ccl.net
>
> Send your subscription/unsubscription requests to: CHEMISTRY-REQUEST[at]ccl.net
> HOME Page: http://www.ccl.net   | Jobs Page:
http://www.ccl.net/jobs
>
> If your mail is bouncing from CCL.NET domain send it to the maintainer:
> Jan Labanowski,  jlabanow[at]nd.edu (read about it on CCL Home Page)
> -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>
>
>
>
>
> _______________________________________________
> Impact-l mailing list
> Impact-l[at]www.schrodinger.com
> https://www.schrodinger.com/mailman/listinfo/impact-l
>




Raw Message Text