GOPhage: protein function annotation for bacteriophages by integrating the genomic context

Area of Science:

• Microbiology
• Bioinformatics
• Virology

Background:

• Bacteriophages (phages) are vital in microbial ecology, but their protein functions are often unannotated.
• Metagenomic sequencing yields many new phages, exacerbating the annotation challenge due to phage protein diversity and scarcity of known functions.

Purpose of the Study:

• To develop an advanced computational tool, GOPhage, for accurate bacteriophage protein function annotation.
• To address limitations of existing methods in annotating diverse and novel phage proteins.

Main Methods:

• Leveraging the modular genomic structure of phage genomes.
• Utilizing embeddings from advanced protein foundation models and Transformer architecture.
• Capturing contextual information between proteins within phage genomes.

Main Results:

• GOPhage significantly improves annotation of diverged proteins (6.78%) and proteins with uncommon functions (13.05%) compared to state-of-the-art methods.
• The tool successfully annotates proteins lacking homology search results, crucial for new phage genomes.
• Identified 688 potential holins in phages with high structural conservation.

Conclusions:

• GOPhage enhances the functional annotation of bacteriophage proteins, particularly for novel and diverse sequences.
• The tool aids in characterizing newly discovered phages and understanding phage biology.
• GOPhage holds potential for expanding our knowledge of microbial ecosystems through phage discovery.