search

ablab / viralVerify

viralVerify: viral contig verification tool
MIT License
62 stars 11 forks source link

Plasmids classified as viruses #6

Open asl opened 3 years ago

asl commented 3 years ago

Consider the attached FASTA file. It contains two sequences that match a plasmid from RefSeq at 99% IDY. However, there are classified as virus by viralVerify. Looks strange. ic5-metaplasmidspades.plasmid.fasta.zip

mikeraiko commented 3 years ago

Interesting. They both predicted as viruses by other tools, such as VirSorter, and match phages P1 and P7 with IDY 96-98% and span 86%. And this is fine because these phages are somewhat special - they can exist as a plasmids in the cell, and have both plasmid- and phage-specific genetic features.

asl commented 3 years ago

FWIW Platon (https://github.com/oschwengers/platon) predicts it as plasmid.

Dmitry-Antipov commented 3 years ago

I'm pretty sure that plasmidVerify would also classify it as plasmid. For the sequences that similar to both P and V and not similar to C a plasmid prediction tool would say they are plasmid, and a viral prediction tool would say they are viral

Current logic for viralVerify is to compare V vs P+C and then only if P+C compare P vs C. It is vulnerable to such cases, but I suppose it is not a common situation..

Dmitry-Antipov commented 3 years ago

And these contigs are plasmids and phages in the same time.

So, the right answer for such contigs should be virus_or_plasmid that is not currently supported by viralverify.