RMSCNN: A Random Multi-Scale Convolutional Neural Network for Marine Microbial Bacteriocins Identification.

Area of Science:

• Microbiology
• Bioinformatics
• Drug Discovery

Background:

• Antibiotic resistance is a growing global health crisis, driven by the overuse of traditional antibiotics.
• Bacteriocins, antimicrobial peptides produced by bacteria, show promise as alternatives to conventional antibiotics.
• Marine environments are rich sources of novel microbial compounds, including marine microbial bacteriocins (MMBs).

Purpose of the Study:

• To develop an effective deep learning method for identifying potential marine microbial bacteriocins (MMBs).
• To explore the potential of MMBs in combating antibiotic resistance.
• To identify novel MMB candidates from marine microorganisms for drug development.

Main Methods:

• A novel random multi-scale convolutional neural network was proposed for MMB identification.
• The method incorporated a random model for scale setting to enhance extensibility.
• State-of-the-art classification techniques were used for performance comparison.

Main Results:

• The proposed deep learning method demonstrated superior classification performance compared to existing methods.
• Several potential MMB candidates were successfully predicted.
• Sequence analysis identified HNH endonucleases from marine bacteria as promising MMB candidates.

Conclusions:

• Deep learning provides an effective approach for identifying novel marine microbial bacteriocins (MMBs).
• The developed random multi-scale convolutional neural network method shows significant potential for MMB discovery.
• HNH endonucleases represent a novel class of potential bacteriocins derived from marine bacteria, offering new avenues for antibiotic development.