AEPMA: peptide-microbe association prediction based on autoevolutionary heterogeneous graph learning.

Area of Science:

• Biochemistry and Computational Biology
• Drug Discovery and Development
• Microbiology and Infectious Diseases

Background:

• Antibiotic resistance is a growing global health threat, necessitating novel therapeutic strategies.
• Antimicrobial peptides (AMPs) show promise as alternatives to conventional antibiotics.
• Current computational methods for AMP discovery often lack specificity for target microbes.

Purpose of the Study:

• To develop a computational framework, AEPMA, for targeted prediction of peptide-microbe associations.
• To address the limitations of existing methods by focusing on specific microbial targets.
• To accelerate the discovery of novel antimicrobial peptides.

Main Methods:

• Construction of a peptide-microbe-disease heterogeneous network (PMDHAN).
• Development of an autoevolutionary information aggregation mechanism for representation learning.
• Utilizing a heterogeneous graph-based approach for peptide-microbe association prediction.

Main Results:

• AEPMA demonstrated superior performance compared to five state-of-the-art methods on multiple datasets.
• The framework exhibited robust modeling capabilities and strong generalization ability.
• Identification of novel peptides effective against Staphylococcus aureus and Escherichia coli.

Conclusions:

• AEPMA provides an efficient and targeted computational approach for antimicrobial peptide discovery.
• The framework contributes valuable insights for developing new antimicrobial drugs and combating antibiotic resistance.
• This study highlights the potential of autoevolutionary heterogeneous graphs in drug discovery.