Quantitative Structure–Activity Relationship Models of Chemical Transformations from Matched Pairs Analyses

The concepts of activity cliffs and matched molecular pairs (MMP) are recent paradigms for analysis of data sets to identify structural changes that may be used to modify the potency of lead molecules in drug discovery projects. Analysis of MMPs was recently demonstrated as a feasible technique for...

Full description

Saved in:
Bibliographic Details
Published in:Journal of chemical information and modeling 2014-04, Vol.54 (4), p.1226-1234
Main Authors: Beck, Jeremy M, Springer, Clayton
Format: Article
Language:English
Subjects:
Chemical compounds
Chemistry
Feasibility Studies
Models, Chemical
Molecules
Quantitative Structure-Activity Relationship
R&D
Research & development
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The concepts of activity cliffs and matched molecular pairs (MMP) are recent paradigms for analysis of data sets to identify structural changes that may be used to modify the potency of lead molecules in drug discovery projects. Analysis of MMPs was recently demonstrated as a feasible technique for quantitative structure–activity relationship (QSAR) modeling of prospective compounds. Although within a small data set, the lack of matched pairs, and the lack of knowledge about specific chemical transformations limit prospective applications. Here we present an alternative technique that determines pairwise descriptors for each matched pair and then uses a QSAR model to estimate the activity change associated with a chemical transformation. The descriptors effectively group similar transformations and incorporate information about the transformation and its local environment. Use of a transformation QSAR model allows one to estimate the activity change for novel transformations and therefore returns predictions for a larger fraction of test set compounds. Application of the proposed methodology to four public data sets results in increased model performance over a benchmark random forest and direct application of chemical transformations using QSAR-by-matched molecular pairs analysis (QSAR-by-MMPA).
ISSN:1549-9596
1549-960X
DOI:10.1021/ci500012n