Binary formal inference-based recursive modeling using multiple atom and physicochemical property class pair and torsion descriptors as decision criteria

Analysis of a large amount of information, typically generated by high-thro ughput screening, is a very difficult task. To address this problem, we hav e developed binary formal inference-based recursive modeling using atom and physicochemical property class pair and torsion descriptors. Recursive par titioning is an exploratory technique for identifying structure in data. Th e implemented algorithm utilizes a statistical hypothesis resting, similar to Hawkins' formal inference-based recursive modeling program, to separate a data set into two homogeneous subsets at each splitting node. This proces s is repented recursively until no further separation can occur. Our implem entation of recursive partitioning differs from previously reported approac hes by employing a method to extract multiple features at each splitting no de. The method was examined for its ability to distinguish random and real data sets. The effect of including a single descriptor and multiple descrip tors in the splitting descriptor set was also studied. The method was teste d using 27 401 National Cancer Institute (NCI) compounds and their pGI50 (- log(GI(50))) against the NCl-H23 cell line. The analyses show that partitio ning using multiple descriptors is advantageous in analyzing the structure- activity relationship information.