Can Metabuli be used for the discovery of new viruses?

[zhangwenda0518]

Hello author, thank you for developing such an efficient species classification tool, I'm trying to use public transcriptome data to make virus discoveries on the plants I'm studying, and since most of the plants I'm interested in are non-model plants。

I'd like to ask Can Metabuli be used for the discovery of new viruses. Is it necessary to extract Metabuli's virus reads for megahit assembly, and further utilize virus identification tools such as Genomad to further identify and classify the assembly results.

Looking forward to your reply, thank you！

[jaebeom-kim]

Interesting question!
For each classified read, Metabuli reports a score reflecting the sequence identity.
If the score is low, it can be a signal of a new viral species.

In Supp. Fig. 4–7, I investigated the distribution of the score

when the query species’ genome is in the database and

when the query species is not in the database, but its sibling species’ genome is included in the DB.

For case 1), more than 80% of classifications had a score > 0.5.

Is it necessary to extract Metabuli's virus reads for MEGAHIT assembly, and further utilize virus identification tools such as GeNomad to further identify and classify the assembly results?

If you want more reliable classification than short-read classification, I’d like to suggest this approach:

Metabuli classification to get virus reads

Run an assembler to get contigs

Classify contigs with Metabuli again

You can skip step 1 if your sample is not large.
You can use GeNomad in step 3, but I have seen cases where Metabuli’s virus classification is better than GeNomad. So I think you can try both in step 3.