Prophage Prediction

Led by Thomas Sicheritz-Pontén and Kimmo Sirén , we developed a new tool for the rapid identification of prophages within bacterial genomes. The full paper is here

Benchmarking against other common prophage prediction tools including VIBRANT, VirSorter, Prophage Hunter and PHASTER , PhageBoost identified the most validated prophage regions (231). PHASTER identified 224, VirSorter 221, VIBRANT 192 and Prophage Hunter 173.  In terms of sensitivity VirSorter was highest (0.85) followed by PhageBoost (0.82) , Prophage Hunter (0.66), PHASTER (0.78) and VIBRANT (0.73).

Speed comparison

On a dataset of 5537 SAGs PhageBoost was clearly the fastest, with predictions completing in ~33 hrs. Which was 9× faster than VIBRANT, and ∼19× faster than VirSorter. PHASTER took >2 weeks to return results from a remote server.

Some of these tools do things beyond which PhageBoost was designed to do, and also do not require surrounding bacterial DNA to predict prophages as PhageBoost does. The aim of PhageBoost was not to replace these tools but add a high-speed and accurate prophage detection, that is suitable for high-throughput sequencing projects.

PhageBoost can be accessed here or downloaded from github