Taxonomy Assignment with QIIME2

This workflow performs taxonomic assignment for an input OTU/ASV (Operating Taxonomic Unit/Amplicon Sequence Variant) Feature Table and set of amplicon Representative Sequences using QIIME2. The workflow creates and trains a QIIME2 classifier for an input reference taxonomic database (e.g. HOMD, SILVA, PR2, UNITE) and uses the trained classifier to assign taxonomy to OTUs/ASVs in the input Feature Table.

Inputs

• Feature Table [biom2]: OTU table with ASVs in rows and samples in columns containing ASV counts per sample (can be the output from previous QIIME2 or DADA2 processing)
• Representative Sequences [fasta]: File containing sequences for all ASVs detected in the samples (can be the output from previous QIIME2 or DADA2 processing)
• Reference sequences [fasta]: File containing sequences for the chosen reference taxonomy database
• Reference taxonomy [tabular]: File containing taxonomy assigned to sequences in the reference taxonomy database
• Forward Primer Sequence [text]: Forward primer sequence (5' -> 3'). Required for qiime2 feature-classifier extract-reads (input parameter: f_primer). Default value is the primer for the 16S rRNA V3-V4 region: CCTAYGGGRBGCASCAG.
• Reverse Primer Sequence [text]: Reverse primer sequence (5' -> 3'). Required for qiime2 feature-classifier extract-reads (input parameter: r_primer). Default value is the primer for the 16S rRNA V3-V4 region: GGACTACNNGGGTATCTAAT.

Outputs

• OTU/ASV table [csv]: Modified version of input Feature Table with additional column with the assigned taxonomy and assignment confidence for each ASV as additional columns.
• Feature Table Summary [html]: QIIME2 visualisation of the input OTU/ASV Feature Table biom2 dataset
• Representative Sequences Summary [html]: QIIME2 visualisation of the input Representative Sequences fasta dataset
• Taxonomy Bar Plot [qzv]: Interactive bar plot for OTU/ASV table with assigned taxonomy (this dataset cannot be viewed in Galaxy; to view this dataset, export it from Galaxy and use a dedicated QIIME2 viewer, e.g. https://view.qiime2.org/).

Processing Steps

1. Convert all input datasets to QIIME2 artefacts using qiime2 tools import.
2. Summarise the Feature Table and Representative Sequences datasets using qiime2 feature-table summarize and qiime2 feature-table tabulate-seqs, respectively, and extract html visualisations that can be viewed in Galaxy using QIIME vizualisation extractor.
3. Extract reference sequences (for the input reference taxonomy database) for training the feature classifier using qiime2 feature-classifier extract-reads.
4. Train the QIIME2 Naive Bayes feature classifier on the reference database (extracted sequences and taxonomy) using qiime2 feature-classifier fit-classifier-naive-bayes.
5. Run the classifier on the amplicon Representative Sequences to assign taxonomy to the sample ASVs using qiime2 feature-classifier classify-sklearn.
6. Create a summary visualisation of the taxonomy assigned to the sample dataset using qiime2 taxa barplot.
7. Export the assigned taxonomy table from a QIIME2 artefact to tsv using qiime2 tools export.
8. Convert the output tsv table to tabular format using Convert tabular to CSV and Convert CSV to tabular (requires intermediate conversion to csv).
9. Convert the biom2 Feature Table to tabular using Convert between BIOM table formats.
10. Combine the ASV taxonomy assignment and Feature Table tabular datasets using Remove beginning and Join.
11. Convert the combined OTU/ASV table from tabular to csv format using Convert tabular to CSV.

Sample Data

Sample input datasets for this workflow are available on Zenodo: https://doi.org/10.5281/zenodo.18263330

Contribution

Workflow created by Tristan Reynolds (provided Galaxy and bioinformatics expertise) and Amy Loughman (provided scientific requirements and expertise).