Role of moving average analysis for development of multi-target (Q)SAR models

In modern drug discovery era, multi target- quantitative structure activity relationship [mt- (Q)SAR] approaches have emerged as novel and powerful alternatives in the field of in-silico drug design so as to facilitate the discovery of new chemical entities with multiple biological activities. Amongst various machine learning approaches, moving average analysis (MAA) has frequently exhibited high accuracy of prediction of diverse biological activities against different biological targets and experimental conditions. Role of MAA in developing (Q)SAR models for prediction of single/dual or multi target activity has been briefly reviewed in the present article. Subsequently, MAA was successfully utilized for developing mt-(Q)SAR models for simultaneous prediction of anti-Plasmodium falciparum and anti-Trypanosoma brucei rhodesiense activities of benzyl phenyl ether derivatives. The statistical significance of models was assessed through intercorrelation analysis, sensitivity, specificity and Matthew's correlation coefficient. Proposed MAA based models were also validated using test set. High predictability of the order of 80% to 95% amalgamated with safety (indicated by high value of selectivity index) of proposed mt-(Q)SAR models justifies use of MAA in developing models in order to obtain more realistic and accurate results for prediction of anti-protozal activity against multiple targets. Active ranges of the proposed models can play a significant role in the development of novel, potent, versatile and safe anti-protozoal drugs with improved profile in terms of both anti-Plasmodium falciparum and anti-Trypanosoma brucei rhodesiense activities.