Discovery of new antibiotics using bioinformatics and machine learning methods

Abstract

Antibiotic resistance is a serious global health threat that causes approximately 1.27 million deaths worldwide each year and is expected to reach 10 million by 2050. New antibiotic development is exceptionally challenging, typically requiring 10-15 years and approximately $1.5 billion investment. In this process, genomic and metagenomic analyses play a critical role by revealing the genetic potential of unculturable microorganisms and identifying new antibiotic-producing microorganisms. Additionally, deep learning models analyze molecular structures to identify new compounds with antibacterial activity, and virtual screening techniques analyze large molecular databases to determine potential active compounds. It has been shown that models developed using deep learning
can predict antibiotic biosynthesis gene clusters with over 90% accuracy. Alongside these approaches, the identification of antibiotic
combinations and the prediction of synergistic effects allow for the development of more effective treatment strategies against multidrug resistance. These methods contribute to the development of proactive approaches in managing antibiotic resistance and optimize the discovery of new antibiotics and the effective use of existing ones. This review examines the discovery of new antibiotics using bioinformatics and machine learning methods.