Qiime2 ConQuR Plugin for Batch Correction (or alternative)

[charlesalexandreroy]

Hi all!

I recently came across the ConQuR tool (paper, GitHub) for dealing with microbiome batch effects, and it looks pretty useful.

I was wondering whether a Qiime2 plugin exists for it, and if not, whether one might be in development?

If not, as described in their vignette (or manual), the tool starts from a n (samples, rows) by p (taxa, cols) counts matrix. Where in the Qiime2 pipeline would an output like that be generated (if anywhere) if I were to unzip a .qza file to process it with ConQuR?

I am also interested if anyone knows about any other tools that might be as good as (or better than) ConQuR.

[Nicholas_Bokulich]

Hi @charlesalexandreroy ,
Welcome to the forum!

A few years back there was a QIIME 2 plugin, q2-perc-norm, designed for batch correction. You can see a post about it here with some more useful links:

Case-Control Studies and Batch Correction: Community Tutorial

Dear QIIME2 Community, Christian Diener and myself taught a QIIME2 workshop at the University of British Columbia at Kelowna on December 3rd, 2018. Christian has included all of our workshop materials, including a step-by-step walk-through, input files, and output files on github: GitHub - Gibbons-Lab/ccmb_workshop: Data and materials for the CCMB workshop on Cloud Computing and Microbiome Bioinformatics. We hope you find our tutorial useful. It walks through analysis of several 16S colorectal…

I am not aware of a plugin for ConQuR, but it certainly sounds useful. I would recommend contacting the authors of ConQuR to discuss the idea — if you or they are interested in developing such a plugin, you can let us know on the forum if you need help getting started with plugin development.

@charlesalexandreroy

the tool starts from a n (samples, rows) by p (taxa, cols) counts matrix. Where in the Qiime2 pipeline would an output like that be generated

This would be a feature table, and more specifically a FeatureData[Frequency] artifact. Many different QIIME 2 plugins produce and work with these but the most basic steps to generate a feature table (and specifically one where the features are taxonomic groups) would be:

1. denoise and/or cluster your sequences, e.g., with q2-dada2, q2-deblur, or q2-vsearch (all of these will generate feature tables from demultiplexed sequence data; see the amplicon-docs for more usage details)
2. perform taxonomy classification of your representative sequences with q2-feature-classifier
3. use the q2-taxa plugin to collapse your feature table by taxonomy. The output will be a feature table (count matrix) of samples X taxa.

Good luck!

[charlesalexandreroy]

Edit:
Actually, scratch the below. It seems that the author had a deadline and wasn't able to respond initially, but they are back now. They said they would look into making a Qiime2 plugin and are otherwise helping me get ConQuR to work for my dataset.

---

Thanks @Nicholas_Bokulich!

In case anyone else ends up here, after trying to get ConQuR to work for a few days, I think I am ready to give up on it. Unfortunately, even though it was published in Nature less than 3 years ago, it appears to be abandoned. At least, I'm not hearing back from the author via GitHub or email, and I'm experiencing an endless series of bugs as I try to get it working.

I'll see if q2-perc-norm can do the job for me.