/q2-ancombc

qiime2 plugin for ANCOMBC

Primary LanguagePython

q2-ANCOMBC

qiime2 plugin for ANCOMBC

Installation

Make sure you have qiime2 installed according to the installation instruction. Once you have the conda enviroment activated, open R by running

R

This will open the R prompt window in the terminal. R libraries installed in the terminal within your conda enviroment are the only ones qiime2 will see; if you wish to install ancombc in R studio or something similar, you will need to redo the installation there.

In the R terminal, install ANCOMBC locally:

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

Then install the qiime2 plugin via

pip install git+https://github.com/mortonjt/q2-ancombc.git
qiime dev refresh-cache

If you run the command,

qiime

you should see ancombc listed in the avaliable plugins.

Getting started

You can try running ANCOM-BC using the tutorial data in this repository. Start by making a tutorial directory and downloading the tutorial dataset:

mkdir ancom-bc-tutorial
cd ancom-bc-tutorial

# Downloads the files
wget https://github.com/mortonjt/q2-ancombc/raw/main/example_data/table.qza
wget https://github.com/mortonjt/q2-ancombc/raw/main/example_data/taxonomy.qza
wget https://raw.githubusercontent.com/mortonjt/q2-ancombc/main/example_data/metadata.txt

We can run ANCOM-BC using an R-style formula. In this case, we'll use the "labels" column:

qiime ancombc ancombc \
    --i-table table.qza \
    --m-metadata-file metadata.txt \
    --p-formula "labels" \
    --o-differentials differentials.qza

This will output a single results file, differentials.qza, which contains the parameters and p-value. You can visualize this and the taxonomy results using qiime:

qiime metadata tabulate \
 --m-input-file differentials.qza \
 --m-input-file taxonomy.qza \
 --o-visualization differentials.qzv

In the output table, you will find 5 columns from ANCOM-BC:

  • beta: The coeffecient for the taxonomic feature.
  • se: The standard error for the coeffecient
  • W: The test statistic, calculated as $\beta/se$
  • p_val: The p-value; p-value comes from a two-sided z-test using the W test statistics
  • q_val: The adjusted p-value after multiple hypothsis correction.

If you included the taxonomy artifact, you will find an additional 2:

  • Taxon: The taxonomic identifier for the features.
  • Confidence: How confident the classifier was with the taxonomic assignment. Most

Citation

Lin, H. and Peddada, S.D. (2020) "Analysis of compositions of microbiomes with bias correction." Nature Communications 11: 3514. doi: 10.1038/s41467-020-17041-7

Credits

These docs draw heavily off the documentation from the R repository by Frederick Huang Lin.