Hi,
in the last year I have worked with three tools for finding prophages in bacterial genomes: Prophinder, Phast and PhiSpy.
There are a few things I have noticed when comparing the results of all three and that you should be aware of. First of all it really depends from what kind of environment your bacteria come from. Mine are from hydrothermal vents, where a lot of bugs are just unknown. Same for their phages.
All three are database driven since they rely on annotated genomes. Those annotations are nowadays allways made using reference databases. PhiSpy looks at protein length and uses the extracted annotations from the genbank file to determine those. So it is database based.
All three tools behave very differently when it comes to detecting prophages. I have seen that sometimes, all three detect a prophage in the same genome, but give it a different size which sometimes hardly overlap. Meaning the start and stop of the phage are not easily identified. At other times, it is maybe only one of the tools finding a prophage. PhiSphy tends to overshoot, so it often indicates multiple regions in the genome that are prophages, while often all are false positives. In my experience I trust prophinder the most, but it is not a given.
I always manually inspect the results and determine if I agree with them or not. Bioinformatic tools are nice, but with respect to prophage detecting, I think most of the tools have a hard time detecting really unfamiliar prophages, that were integrated in the distant past. Recently, a paper was published on the dark mater (prophages and the like) in all present Microbial genomes, that completely missed the phylum on which I am working. With their novel tool, they did not detect any virus in that phylum. So how good are those results then...
So don't trust the accuracy of these tools at all.
Thank you for your comments, I really appreciate it.