Combine multiple samples into unified results for comparison : Taxonomic profiling

[ppreshant]

• Currently each sample runs through the pipeline independently and produces separate outputs.
• Unifying them into a single output is essential for comparitive analysis / downstreamtasks
• This is relevant for both tabular outputs as well as html ones like krona plots

tsv tables

agm/16S/gap: after all of the samples in the samplesheet have been processed we should run emu combine-outputs to create a singular OTU table of the abundances / taxon calls for the samples in the users dataset. ( probably a separate emu module)

.. the combine-outputs command is already an option within emu so we should just call that from the results dir or wherever the *rel-abundance.tsv files are

PK: I can make that into a separate module like you suggested then - I think staging all the individual .tsv files will take a bit of work, so will mark it for later/next week (+v1).

• Staging should be straightforward and supported by nextflow by collapsing channel content ; see the right word for this

Krona plots

PK: On the same note of collating multiple samples' outputs for taxonomic profiling, Todd mentioned that taxburst / krona can handle multiple samples and present them in the same html as separate tabs or something?
Do you have any experience with this and is it robust/ finicky?
Do you suggest implementing this? I think it will be pretty handy if it works reasonably!

agm, ..: I have never used krona plots in that way. I don’t even know if I’ve ever seen him being done, but Todd would probably know best from first hand experience with the original authors of krona plot.

Plan

• Start by making a new module emu_combine_samples for the tsv compilation
• Test it on 3 samples (if it's too long for the 3rd file, test with 2 current ones)
  • for getting the 3rd file, let's convert that documentation procedure into a standalone .sh script? don't waste time on this now

I think it would be even best is that interactive HTML like what ampliseq makes and we can incorporate something for a visualization of multiple samples into that

• Check what this means and leave some info and a screenshot here

[microbemarsh]

Emu has this capability already, just another module we'll need to make for it (use use whatever the other nextflow too currently does).

I think what would be important is the usage of multiple different tools for the ensemble profiling, this is a tricky thing to do tho.

• See discussion on tools from bakeoff for ensemble profiling here 5/Mar/26 - Lab meeting presentation feedback #96