Abstract
Chemoinformatics play a vital role in drugs discovery process in which it screened out compounds with high probability of failing from the drugs discovery pipeline. Hence, more drugs could be produced in less time and at lower cost. The simplest method in Chemoinformatics that could do as such is the Similarity Searching. Starting with just a simple one target per search, similarity searching is now enhanced to be capable at handling multiple targets in one search. This enhancement is achieved by Turbo Similarity Searching (TSS) that has been proven to increase the recall (i.e. the active recovery rates). TSS incorporates Similarity Searching and Group Fusion in its procedure hence the factors that influence both should also influence TSS. In this paper, we present the concept of TSS investigation on factors related to the usage of different set of similarity measures combination at each phase of TSS and the effect of using different fusion rules on TSS with various descriptors. These are the factors that have been identified to affect the performance of Similarity Searching and Group Fusion. Our initial results indicate that there is a strong influence shown by one of the descriptor (SRECFC) when used at any stages of TSS which returned a high recall. We also observed the poor performance of TSS when another descriptor (ECFC) is used at the second stage of TSS. The second investigation revealed that there are certain preference of descriptors towards fusion modes that gave a recall below random.