Not sure if you have seen this new tool <vclust>. ...
# taxprofiler
l
Not sure if you have seen this new tool vclust. https://www.nature.com/articles/s41592-025-02701-7
👍 2
j
@Joon Klaps you have this in #C0905QCFTPE right?
j
Nope not yet but its an open issue, other stuff took priority.
Got my attention again through this tool. vconcat3
Wouldn’t vcontact3 also be an interesting tool for taxprofiler?
j
That's a genome clustering thing isn't it?
(the description is a bit vgue)
j
Yes for genomes not reads though that’s a key part
l
It seems quite good at discovering the new viruses.
j
(from an older version, the newer version is more senstive to more viruses))
👍 1
j
Yeah, wouldn't fit in taxprofiler then really if it's genome level
l
@Joon Klaps If you consider including vclust to viralmetagenome, will it be an alternative to BLAST or which step would you add it?
j
Would be for contig clustering, currently i’ve mmseqs, vsearch-cluster, cdhit-est, vrhyme and mash supported
l
Aha, I see
Do you think it could potentially replace the BLAST since BLAST is quite slow if there are many viral queries
j
You mean in the step before when identifying potential references for scaffolding?
If so, not certain it will be of a lot of benficial value, it will do a all vs all search as it’s trying to cluster. Within that subworkflow from my own analysis runs, i’ve never had blast being a bottleneck performance-wise. it would always be one of the downstream cluster algorithms that would take much longer for more complex sampels. That said, we can always switch to using mmseqs for that step.
👍 1
💯 1