The growing threat posed by bacteria resistant to antibiotics in recent years has highlighted the need for innovative antimicrobial agents. The Community for Open Antimicrobial Drug Discovery (CO-ADD), a non-profit organization established at The University of Queensland, aims to address this issue by encouraging the public sharing of chemical structures and data on antimicrobial action. The CO-ADD database presents a multi-label classification problem that adds complexity to the traditional machine learning methods for classification therefore demanding the use of new approaches and algorithms for efficient data analysis. Several multi-label classification challenges are tackled in this paper, with the primary question explored being the debate between using multiple classifiers versus employing a single one that can simultaneously classify multiple items. Additionally, the efficacy of feature selection is examined to determine if it’s equally effective across both cases. Furthermore, a comparative analysis is conducted to determine whether the Harmony Search algorithm outperforms the more popular population-based optimization algorithms, like the Genetic algorithm and Particle swarm optimization algorithm.