Worked Example

Worked example: a public SRA dataset, start to finish

Want to see what qiime2-its produces before installing anything? This page walks through a real, production-scale run on public data -- 53 soil-fungi samples from NCBI BioProject PRJNA767765 -- and publishes everything: the scripts, the metadata, and the resulting report.

Open the full PDF report (37 pages) · all the files

A look inside the report

Alpha diversity per group, one page for each metadata column that tested significant.	PCoA for Bray-Curtis and unweighted UniFrac. Colorblind-safe colors, plus marker shapes when a column has more groups than the palette has colors.
UPGMA sample clustering from the same distance matrix, leaves colored like the PCoA.	Genus-level composition of every sample.

The report also carries the run's full provenance (exact command, parameters, tool and plugin versions, input files), DADA2 read retention per sample, sequence-length and classification-confidence distributions, rarefaction curves and the sample-classifier results -- see Advanced stats and the PDF report for what each page means.

The dataset

Source	BioProject PRJNA767765 -- fungal ITS amplicons, Illumina MiSeq, paired-end
Samples	the 53 field samples from Poland and Iran (the BioProject's laboratory microcosm samples are left out)
Size	106 fastq files, 3.3 GB
Metadata	country, environmental medium, local environment, collection date, elevation, replicate -- built from the samples' public BioSample records
Classifier	UNITE 2025-02-19, fungi, 99% clusters, no singletons

Reproduce it

Inside an activated QIIME2 environment with qiime2-its installed (pip install qiime2-its, see Installation) and a clone of the repository for the scripts, from an empty working folder with ~15 GB free:

EX=/path/to/QIIME2_ITS/examples/PRJNA767765

python3 $EX/01_fetch_metadata.py .   # optional: rebuilds metadata.tsv + download_manifest.tsv
bash $EX/02_download_reads.sh        # 3.3 GB from ENA into ./raw_reads, MD5-verified, resumable
bash $EX/03_train_classifier.sh      # UNITE classifier into ./classifier (slow; skip if you have one)
THREADS=40 PARALLEL=10 bash $EX/04_run_pipeline.sh   # -> ./output/report.pdf

The analysis itself is a single command, everything else left at its default:

qiime2-its -q rachis-qiime2-2026.7 \
    -i raw_reads -o output -m metadata.tsv -c classifier/unite-classifier.qza \
    -pe --extract-its2 --taxa Fungi -t 40 -p 10

The published run used qiime2-its 0.3.2 on QIIME2 2026.7 and took 3 h 28 min with 40 threads on a 64-core workstation. Expect small numerical differences if you re-run it (rarefaction and PERMANOVA permutations are random), not different conclusions.

Script	What it does
01_fetch_metadata.py	Builds the QIIME2 metadata file and the download manifest from two public APIs (ENA file report, NCBI BioSample); standard library only. Its output is committed, so this step is optional -- it exists so nothing in the example is hand-made.
02_download_reads.sh	Downloads each run's fastq pair from ENA under the Casava-style names QIIME2 requires, checking every file against ENA's MD5.
03_train_classifier.sh	Fetches UNITE with QIIME2's rescript plugin and trains the naive-Bayes classifier -- about 1 h 40 min, single-threaded (see also Building a classifier).
04_run_pipeline.sh	The qiime2-its command above.

metadata.tsv is also a realistic template for your own metadata file.

One disclosed edit

Before publishing, the user name, host name and the home-directory part of the program path were blanked in the run's run_metadata.json, and the PDF was rebuilt from the same output folder (qiime2_its.report.build_report()), so "Run by" reads "(not recorded)". Nothing else was changed. You can do the same before sharing your own reports.

Data and citation

The sequencing data belongs to its authors; the repository only holds scripts and metadata derived from the public archive records. If you use it, cite the study registered on the BioProject:

Okrasińska A. et al. (2022). Marginal lands and fungi -- linking the type of soil contamination with fungal community composition. Environmental Microbiology. https://doi.org/10.1111/1462-2920.16007

UNITE is distributed under CC BY-SA 4.0 -- see https://unite.ut.ee/cite.php.