In trials involving different biological targets, usually the template CoMFA methodology has automatically generated 3D-QSAR models that predict all ChEMBL-recorded potencies for that target, with SDEPs of roughly a log unit. For these studies, the requisite training 2D SAR tables were random subsets of these ChEMBL data, the prospective predictions then being carried out on the remaining data. The requisite 3D templates were either extracted from X-ray structures (for such targets as factor Xa or PTP1B) or, with surprisingly high effectiveness, from small sets of overlaid known ligands (for such targets as D2A, COX2, or estrogen receptor).  Virtual screening and off-target applications, which seek efficient selection of putative actives rather than the potency forecasts useful in lead series identification and development, have led to a means of generalizing the SAR scopes of these models.<br/>

General applicability of template CoMFA to prospective bioactivity prediction

249th National Meeting (2015)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.