Novel anti-tuberculosis drug designs by data mining for similarity in substituent substitution and structure modification

Mycobacterium tuberculosis (TB) is among the most common of infectious diseases that cause death, and as many as one-third of the world’s population may be infected. This work presents 17 novel hydrazide agents formed by focused in silico data mining utilizing search parameters restricted to substituent replacement only. Substituent substitution has been highly successful in design of novel antibacterial and antiviral drugs. This diverse set of hydrazide constructs possess molecular properties indicating favorable bioavailability with excellent intestinal absorption for oral administration. All agents have zero violations of the Rule of 5, indicating favorable druglikeness. Important pharmaceutical properties including polar surface area, Log P, and formula weight were determined and compared to that of the parent structure of isoniazid by hierarchical cluster analysis and discriminant analysis. The average Log P with range is -0.258 and -2.165 to 1.373, respectively. The average polar surface area (PSA) with range is 75.19 A2 and 55.121 A2 to 94.036 A2, respectively. The diverse range of PSA and Log P, with other descriptors, portend a versatile group of hydrazide drugs having substantial potential to expand the application and effectiveness for clinical treatment of multi-organ infected TB patients. Analysis of similarity indicated that all 17 agents are significantly similar to isoniazid, however discriminant analysis and hierarchical cluster analysis are able to differentiate isoniazid based upon molecular properties. Molecular weight and number of atoms were highly correlated by Pearson r (r > 0.9000), with Log P moderately correlated (r > 0.5500) to number of atoms, molecular weight, and volume. Seventeen hydrazide compounds (success rate of approximately 38%) having diverse pharmaceutical properties resulted from substituent data mining with potential for clinical application.