Streamlining CRISPR spacer-based bacterial host predictions to decipher the viral dark matter

Area of Science:

• Microbiology
• Virology
• Bioinformatics

Background:

• Viral metagenomics has revealed numerous novel phages with diverse genomes.
• Identifying bacterial hosts is crucial for understanding phage ecological roles.
• CRISPR spacers offer a record of phage-bacteria interactions for host prediction.

Purpose of the Study:

• To establish standardized guidelines and develop tools for phage host prediction using CRISPR spacers.
• To create a comprehensive database and a program for large-scale viral dataset analysis.
• To provide an interpretable host prediction approach based on CRISPR-Cas biological principles.

Main Methods:

• Developed a computational toolset for phage host prediction from CRISPR spacers.
• Compiled a database exceeding 11 million CRISPR spacers.
• Implemented a host prediction algorithm incorporating biological criteria of CRISPR-Cas systems.

Main Results:

• Achieved a recall of 49% and a precision of 69% in host prediction for 9484 known phages.
• Demonstrated higher performance for phages infecting gut-associated bacteria.
• Successfully applied the method to large viral datasets.

Conclusions:

• The developed tools provide a standardized and interpretable method for predicting bacterial hosts of unknown phages.
• This approach is particularly effective for characterizing phages within gut viromes.
• Facilitates ecological and functional studies of newly discovered phages.