Activity and Property Landscape Modeling is at the Interface of Chemoinformatics and Medicinal Chemistry

Property Landscape Modeling (PLM) methods and, more specifically, Activity Landscape Modeling (ALM) methods -where the ‘property’ is a biological activity- are emerging approaches to mine systematically structure-property/activity relationships (SPRs/SARs). ALM studies are relatively fast to execute and easy to interpret (provided that the appropriate structural representations are used). While it is well-accepted that the presence of activity cliffs in data sets has a detrimental effect in predictive computational modeling, the activity cliffs also have a ‘dual face’ in medicinal chemistry; activity cliffs have a beneficial aspect because of its high information content. However, cliffs may interrupt the ‘SAR-by-design’ practice that often relies on the similarity principle. To date, most of the published PLM/ALM studies have been developed by chemoinfomatitians using benchmark or public data sets in a retrospective manner. However, it is desirable if more medicinal chemists integrate the identification (if any) and interpretation of activity cliffs. Depending on the size of the data set and the complexity of the screening data (e.g., number of endpoints), an experienced medicinal chemist could rapidly identify all the property/activity cliffs. Alternatively, especially for large screening data sets, computational approaches can greatly help to identify such cliffs in a systematic manner. Either by visual inspection or through the use of computational analysis, the recommendation is to analyze the presence of activity cliffs. Such practice will enhance the understanding of the SAR, the development of local SAR models that later can be translated into global SAR models. Future Medicinal Chemistry, Vol. 7, No. 9. Pages 1197-1211. (http://dx.doi.org/10.4155/fmc.15.51)

Go to the profile of José L. Medina-Franco
Jul 24, 2015
1
0
Like 1 Comment
Future Medicinal Chemistry, Vol. 7, No. 9. Pages 1197-1211. (http://dx.doi.org/10.4155/fmc.15.51)
Go to the profile of José L. Medina-Franco

José L. Medina-Franco

Research Professsor, National Autonomous University of Mexico (UNAM)

No comments yet.