maximum target sequences parameter

[panosioannidis]

Hello,

I'm trying to understand how blobtools works and I've been reading the "getting started" section.

https://blobtoolkit.genomehubs.org/blobtools2/blobtools2-tutorials/getting-started-with-blobtools2/

Why do you suggest running diamond blastx with the --max-target-seqs 1 parameter? Wouldn't this mean that you would get hits against only one target sequence for each query sequence? So wouldn't this, in turn, mean that in the case of very large contigs (many Mbp) you would only get one hit? I'm asking because if this contig contained a contaminant (due to misassembly) you wouldn't know that this contig is chimeric. Or is it that blobtools isn't really "interested" in detecting chimeric contigs?

Thanks!

[rjchallis]

These instructions are in need of an update - apologies - so they aren't up to date with the optimal ways of running BlobToolKit.

If possible, it is best to run BlobToolKit via the NextFlow pipeline, or the Snakemake pipeline that it replaces. These pipelines break long sequences up before blasting, focus similarity searches on gene-rich regions and return results along the length of any longer sequences so are quicker to run and more robust to small regions of a long sequence having hits to non-target taxa. Such hits may be due to misassembly, HGT or just limitations of the taxonomic annotation in the sequence databases and while BlobToolKit doesn't really have an opinion on which is which, running the pipeline gives more data to drill down into in the viewer to see the context of any of the hits.

[panosioannidis]

Okay, this makes sense! I will try the nextflow pipeline. Thanks for the reply!