Qiime2 Conversion Script for Valencia

[Fmadere (Ferra Madere)]

Hello All,

A few months ago I ran some 16S sequencing data through Qiime2. I am now attempting to do some further analysis of the data using Valencia, a nearest-centroid based algorithm for the classification of human vaginal microbial community into state types based on their taxonomic composition. On the Valencia Github page, there is a section that gives instructions on how to convert Qiime2 output files to a format that is suitable for input to Valencia (listed below). These instructions ask for the ASV taxon names key and the ASV read count table. I believe that I should be able to find/export the read count table but am unfamiliar with the ASV taxon names key. Is anyone familiar with these files and can give me some clues on how to find and/or generate them? Many thanks for any help that you can provide!

qiime2 conversion script

This script takes ouput from qiime2 and converts it to a format suitable for VALENCIA. The two files expected as input are the ASV taxon names key and the ASV read count table. The expected output is a new table which contains the samples as rows and the condensed taxa as columns. This file can be used as input to VALENCIA. The second file provides a link between the original taxa names and the condensed names.

The script is run as follows:

python3 /path/to/convert_qiime.py /path/to/taxon_key.csv /path/to/asv_count_table.csv

[colinbrislawn (Colin J Brislawn)]

Hello Ferra,

Welcome to the forums!

Valencia sounds like a need tool! It's great to find domain-specific tools that are open source.

For my own notes, here's the GitHub page:

GitHub - ravel-lab/VALENCIA: VAginaL community state typE Nearest CentroId...

And here's the qiime2 conversion script you mentioned:

python3 /path/to/convert_qiime.py \
  /path/to/taxon_key.csv \
  /path/to/asv_count_table.csv

My guess is that taxon_key.csv and asv_count_table.csv are exported from a Qiime2 .qza artifact file. Exactly which files are unclear to me too, and they don't recommend a pipeline for building these ASVs or assigning taxonomy (like dada2 + sklearn or deblur + vsearch).

Have you considered asking the author(s) about how to use Qiime2 to make the input files for Valencia?

Please contact the author if you have any questions, suggestions, or if would like to contribute data to further develop VALENCIA. Thank you and Enjoy!

I think the Valencia authors will be more helpful here, but let us know what you find!