TL;DR
VIBES is a new, automated workflow that efficiently identifies and visualizes viral sequences (prophages) integrated into bacterial genomes. It addresses the challenge of analyzing large bacterial genomic datasets by providing a user-friendly, containerized pipeline that integrates gene annotation and prophage detection, generating interactive visualizations for easier interpretation.
What Problem Does This Paper Address?
Bacteriophages, viruses that infect bacteria, can integrate their genetic material into their host's chromosome, becoming prophages. These prophages can have significant impacts on bacterial fitness, sometimes acting as parasites, other times conferring beneficial traits like increased virulence or resistance to other viruses. With the increasing availability of bacterial genome sequences, there's a growing need for tools that can accurately and efficiently identify these prophage integrations and understand their prevalence and characteristics across large datasets. Existing tools often require significant technical expertise for installation and execution on high-performance computing environments, and many focus solely on prophage annotation without providing broader genomic context or user-friendly visualization.
Methods
VIBES is built as a Nextflow-driven workflow, designed for easy execution across various computing environments (local, cluster, cloud). Key to its usability is the use of containerization (Docker/Singularity) for each step, ensuring all software dependencies are managed and eliminating complex manual installations. The core of the prophage detection relies on nhmmer, a sensitive tool for DNA sequence similarity searches, which is particularly useful for identifying diverged prophage sequences. To provide context, VIBES also integrates Prokka for bacterial gene annotation and BATH for viral protein-coding gene annotation, using the PHROGs database by default. The workflow is designed to run these annotation processes in parallel, maximizing efficiency. A significant contribution is the generation of interactive HTML visualizations using the SODA library, which allows users to explore bacterial chromosome plots, position occurrence plots of integrations, and detailed gene annotation tables.
Datasets
The primary dataset used for evaluation consisted of 1072 Pseudomonas spp. isolate genomes obtained from the Pseudomonas Genome Database (v21.1). These were analyzed against 178 Pf phage variants, with phage sequence coordinates sourced from a previous study on Pf phage lineages.
Key Insights and Ideas
- Unified Workflow for Prophage Analysis: VIBES consolidates prophage detection, bacterial gene annotation, and viral gene annotation into a single, automated pipeline, simplifying a complex analytical task.
- Interactive Visualization is Crucial: The generated HTML visualizations are not just supplementary but are presented as a key component for understanding prophage integration patterns, allowing users to explore genomic context, integration sites, and gene annotations interactively.
- Sensitivity Matters for Prophages: The choice of
nhmmerover faster but less sensitive tools likeblastnhighlights the importance of detecting highly diverged or mutated prophage sequences, which is critical for comprehensive analysis. - Flexibility Beyond Prophages: While designed for prophages, the underlying architecture of VIBES makes it adaptable for general-purpose sequence similarity searches and annotation projects, showcasing its potential as a versatile bioinformatics framework.
Key Results
VIBES successfully identified 517 full-length Pf prophage integrations and 51,386 partial hits across the 1072 Pseudomonas spp. genomes. After filtering for hits with at least two annotated viral genes, 13,850 likely integrations remained, with a mean length of 5095 nucleotides. The workflow demonstrated efficient parallel processing, with the prophage search component alone consuming over 13,500 CPU hours. When compared to other tools like DBSCAN-SWA, geNomad, and PHASTEST on reference genomes, VIBES showed comparable or superior performance in identifying known prophages, with a notable runtime of 880 seconds for both test genomes.
Limitations and Future Work
The authors acknowledge several limitations and suggest future directions:
- Att Site Annotation: The current release does not annotate attachment (att) sites, which are important signals of phage integration. Future work will include adding this functionality.
- Prophage Verification: VIBES currently relies on sequence similarity and filtering for verification. Future iterations aim to improve methods for distinguishing true prophage sequences from homologous non-phage regions, especially for novel or highly diverged prophages.
- De Novo Annotation Support: The workflow is primarily designed for targeted searches using user-provided queries. Enhancing de novo annotation capabilities, where unknown prophages are identified without prior knowledge, is a goal.
- Alternative Search Tools: While
nhmmeris sensitive, it can be computationally expensive. The authors suggest incorporating faster search tools likeLASTorblastnto offer users more options based on their specific needs for speed versus sensitivity.
Summary updated