|
|
| Re: Lilly Medchem Rules [message #869 is a reply to message #859] |
Thu, 23 April 2020 19:29   |
nbehrnd
Messages: 224
Registered: June 2019
|
Senior Member |
|
|
The publication you cite (ACS' Author Choice, thus open access) was published in 2012,
J. Med. Chem. currently is aware of 141 publications citing this work; some you might
know. The history of commits in the repository notes for Dec 14, 2017 «First working
version for version 2», too.
Are there additions / modifications to the set of 275 MedChem rules of the 2012
publication which you would like to suggest to consider today? The approach
suggested indeed is complementary to the approach of druglikeness in DW,[1] in turn
borrowed from the earlierr OSIRIS Property Explorer.
[1] http://www.openmolecules.org/properties/properties.html#drug score
[Updated on: Thu, 23 April 2020 20:32] Report message to a moderator |
|
|
|
|
|
| Re: Lilly Medchem Rules [message #883 is a reply to message #877] |
Sun, 03 May 2020 21:21 |
IanWatson
Messages: 1
Registered: May 2020
Location: United States
|
Junior Member |
|
|
I am one of the authors of the Lilly papers. Indeed the Medchem Rules paper was all about undesirable functional groups. There were other tools for property calculations.
The rules seem to have stood the test of time well, and are in widespread use. I have fielded lots of questions from people about building and using the software. I do not have a good feel for whether or not people have made substantial modifications to the rule set.
One of the important innovations was the concept of a demerit. Most rule sets are either 'in' or 'out'. We introduced the idea of a demerit. This arose because the team of medicinal chemists could not agree on Nitro groups. Some did not mind them, some did not like them at all. When we suggested the demerit idea, that was something they could all embrace, because while having one might be OK, most chemists were united in considering two such groups undesirable. From there, we ended up with lots of demerited, rather than rejected, functional groups.
The rules were developed at a time when there were a great many molecules available for purchase, so they tend to be fairly strict. I am open to ideas about rules that might be overly harsh, or other motifs that should be flagged.
The work we did with PAINS was to investigate how they worked - well for the assay formats for which they were developed, less so elsewhere. A byproduct of that was an implementation in our own query file format - rather than SLN. We did extensive checking of these results compared to Sibyl, but some things about SLN, and the results their software produced, never made sense to me.
Both these projects required sophisticated substructure search concepts, many of which would be hard to express in standard smarts notation. We made some interesting extensions to both smarts and matching concepts in order to express the ideas or chemists were describing.
So, back to the original question, would it be a useful annotation for molecules, I would say yes. The software is widely used, over 100 citations, and many people wold be familiar with the results. I would be happy to help with any tooling needed.
Ian
|
|
|
|
|
|
|
|
Search
Help
Members
Register
Login
Home
on a related topic, but it seems to be a distinct set of PAINS rules. So perhaps the 2012 paper is the best guide.
