q2-ancombc on qiime2-2021.11

[kindergarten]

Hi

I tried to install q2-ancombc as described here GitHub - mortonjt/q2-ancombc: qiime2 plugin for ANCOMBC.
R version 4.0.5 (2021-03-31)

if (!requireNamespace("BiocManager", quietly=TRUE))
install.packages("BiocManager")
BiocManager::install("ANCOMBC")
install.packages('tidyverse')

I noticed following errors

ERROR: dependency ‘nloptr’ is not available for package ‘ANCOMBC’
ERROR: dependency ‘ragg’ is not available for package ‘tidyverse’

I would appreciate if you could help me install q2-ancombc.

Thanks

[lizgehret]

Hi @kindergarten,

Thanks for your interest in using ANCOM-BC!

The package that you linked off to is not currently maintained - however, I developed a new implementation of ANCOM-BC that is currently available in the QIIME 2 core distribution within q2-composition. You can install the latest version of QIIME 2 using these instructions to utilize it! Once you have QIIME 2 installed, you can simply run qiime composition ancombc --help to examine the method parameters.

Hope this helps! Cheers

[kindergarten]

Thanks @lizgehret.
I will install latest qiime2.
Can I use output from version 2021.11 with the latest version or do I need to rerun the analysis pipeline with latest version?

[lizgehret]

Hi @kindergarten,

The input for ancombc will just take a FeatureTable[Frequency] so as long as that's what you've generated from your previous analysis in your 2021.11 environment, that's perfectly fine to use!

Cheers

[kindergarten]

Thanks @lizgehret .
I was able to run ancom-bc using the solution you suggested.
One more question. Bars are labelled with feature ids. Is there a way to add taxonomy labels (species, genus, etc).

[lizgehret]

Hi @kindergarten,

Yep! You can use qiime taxa collapse to generate a new feature table (from your original table) where all features are taxanomic annotations. You can then run this through ancombc to get your differential abundance data for use in da-barplot.

Cheers

[kindergarten]

@lizgehret Thanks for timely responses. I was able to perform ancom-bc on collapsed taxa using your suggestions.

[kindergarten]

Hi @lizgehret
I have few more questions. I would appreciate if you could answer these.

Differential Tables:
My understanding was that differential tables LFC values are relative to reference group (as set by --p-reference-levels)(see 3Q here GitHub - FrederickHuangLin/ANCOMBC: Differential abundance (DA) and correlation analyses for microbial absolute abundance data). However, qiime2view says "Positive numbers indicate enrichment relative to the intercept, negative numbers indicate depletion relative to the intercept." Could you please clarify.
Also, it is possible to download tables as tsv or csv files?

Structural Zeros:
As per GitHub - FrederickHuangLin/ANCOMBC: Differential abundance (DA) and correlation analyses for microbial absolute abundance data (see 2Q), taxa with structural zeros are not included in the primary results of ancombc or ancombc2. Does q2-ancombc excludes taxa with structural zeros? Is it possible to perform struc_zero analysis in q2-ancombc?

Trend:
Is it possible to perform trend analysis in q2-ancombc?

Thanks

[lizgehret]

Hi @kindergarten,

@kindergarten

My understanding was that differential tables LFC values are relative to reference group (as set by --p-reference-levels)(see 3Q here GitHub - FrederickHuangLin/ANCOMBC: Differential abundance (DA) and correlation analyses for microbial absolute abundance data). However, qiime2view says "Positive numbers indicate enrichment relative to the intercept, negative numbers indicate depletion relative to the intercept." Could you please clarify.

This is correct! The reference groups that you are specifying when running ancom-bc are used as the intercept(s).

@kindergarten

Also, it is possible to download tables as tsv or csv files?

Not directly from the visualization on q2view - but you can run qiime tools export on your summary table, and included in that output directory you'll receive a tsv file that contains all of the stats shown in the visualization.

@kindergarten

taxa with structural zeros are not included in the primary results of ancombc or ancombc2. Does q2-ancombc excludes taxa with structural zeros? Is it possible to perform struc_zero analysis in q2-ancombc?

We chose to exclude the struc-zeros parameter because the taxa associated with them are not further utilized within ANCOM-BC, and they are only useful when performing the global test (which we also excluded from the QIIME 2 implementation of ancombc). Huang has a good explanation of this in question 12 under Commonly asked questions.

Hope this helps! Cheers