Fig. 4
From: Machine intelligence-driven framework for optimized hit selection in virtual screening
Identified hits by the CS-driven stacked ensemble framework from small independent validation dataset. The stacked ensemble of chemical-space A-HIOT led to the best identification of 35 hit molecules for a particular target (in case, CXCR4 receptor). The trained CS module of A-HIOT tested on small independent validation datasets (56 molecules) predicted most of the hits belonging to the aromatic ring system, in particular, imidazole ring which is the prime inhibitor of CXCR4 receptor justified its good performance measures as compared to individual machine learning algorithms