Synthesis Strategies and Computational Activity Screening of Pyrazolo [3,4-B] Pyridine, Quinoline, and 1,2,4-Triazolo [1,5-A] Pyrimidine Scaffolds

Abstract

The broad antimicrobial relevance of heterocyclic ring systems has been scarcely exploited, with three nitrogen-rich fused ring systems in particular – pyrazolo[3,4-b]pyridines (PYR), quinolines/polyhydroquinolines (QUIN), and 1,2,4-triazolo[1,5-a]pyrimidines (TAZ) – being highly underutilized. Recent synthetic and structure-activity relationship (SAR) literature pertaining to these scaffolds is reviewed and a machine-learning classification exercise that utilizes a descriptor-based approach is reported to supplement the traditional SAR approach. A set of 250 compounds from 4 different heterocyclic classes was processed with data auditing, exploratory analysis, feature engineering and comparison of logistic regression, decision tree and random forest classifiers. Logistic regression was able to achieve 86.0% held-out accuracy and 84.8% mean five-fold cross-validation accuracy, and the molecular weight and LogP were the most influential descriptors that were identified, confirmed by Kruskal-Wallis non-parametric testing. The results, in conjunction with the reported SAR values from the literature for the three scaffolds, are useful for rational prioritisation of new antibacterial and antifungal heterocyclic lead compound candidates for subsequent synthesis and biological evaluation, especially when rapid screening of the synthetic routes to the new compounds is needed, as would be the case in the pharmaceutical industry.