PREDICTION OF SMALL ANTIMICROBIAL PEPTIDES USING MACHINE LEARNING

Abstract

Antimicrobial resistance (AMR) is a concern to public health, prompting the development of novel strategies for combating AMR. While the use of machine learning (ML) to AMR is in its infancy, it has made significant progress as a diagnosis tool, owing to the growing availability of phenotypic/genotypic datasets and much faster computational power. While applying ML in AMR research is viable, its use is limited. it has been used to predict antimicrobial susceptibility genotypes/phenotypes, discover novel antibiotics, and improve diagnosis when combined with spectrosocopic and microscopy methods. ML implementation in healthcare settings has challenges to adoption due to concerns about model interpretability and data integrity. The focus of this thesis is to outline the significant benefits and drawbacks of ML in AMR with emphasis on models built for the prediction of antimicrobial peptides, along with the salient trends reported in recent studies.