Using lotus2 with input ASVs

[Amanda-Biocortex]

Hi,

I have ASVs across a set of samples- can I run these through lotus 2 lambda using the taxOnly flag?

My guess is I need to create a fasta from the ASVs but then I wont have any quality scores because they are ASVs instead of reads. Can I run these through taxonomy assignment without quality scores?

How would you recommend proceeding?

Many thanks,
Amanda

[hildebra]

Hey, yes you can run the .fasta with the ASVs nt sequences directly through the -taxOnly flag options, no need for quality values at this point :)
best, Falk

[Amanda-Biocortex]

thanks!

And how to I capture ASV frequency so that -taxOnly outputs the 'normal' lotus2 output of abundance per taxa phyloseq?

[hildebra]

LotuS2 doesn't support this. It will only give you the taxonomic annotation for each ASV. However, given an ASV matrix you could then calculate higher taxonomic levels in R

[Amanda-Biocortex]

I need to be able to set.seed() before DADA2 so that my results are reproducible. Is there a way to set.seed() in lotus so that a sample rune through my lotus2 pipeline (DADA2+lambda) will always produce the same result when rerun?

If setting seed for DADA2 in lotus2 is not possible, then I was thinking I could use DADA2 in R to output ASVs, and then assign taxonomy to the ASVs using taxOnly in lotus2. Would that be a correct use of lotus2?

[hildebra]

Hey, you can edit directly the lotus2/bin/R/ASV_clustering.R file in your own copy of lotus2 and set.seed there.
(better to install via github in this case)
Please be aware that these ASVs will likely differ from the ASVs that you produce using a vanilla dada2 run - we have benchmarked this in the LotuS2 paper and seen quite some differences in number of ASVs etc, as lotus2 has some pretty strict filters for DNA quality.