Input data for the pipeline: reference databases, sample metadata, and raw-read QC reports.
| Path | Tracked? | Description |
|---|---|---|
silva-138-99-seqs.qza |
✅ (93 MB) | SILVA 138 99% reference sequences (QIIME 2 artifact) used as the reference for deblur denoise-other in 02.00. |
silva-138-99-seqs.qza.md5 |
✅ | MD5 checksum for the reference sequences. |
silva-138-99-nb-classifier.qza |
❌ (git-ignored) | Pre-trained SILVA 138 99% naïve-Bayes classifier for taxonomy. Download from the QIIME 2 data resources. Not tracked because of its size. |
raw-fastqs/metadata.tsv |
✅ | Sample metadata (see dictionary below). |
raw-fastqs/*_fastqc.html |
✅ | Per-sample FastQC reports for the raw reads (output of 00.00). |
raw-fastqs/multiqc_report.html |
✅ | Aggregated MultiQC report for the raw reads. |
Raw *.fastq.gz |
❌ (not in git) | Raw reads live on the lab server: https://owl.fish.washington.edu/nightingales/eDNA-yellow_island-2023/ |
QIIME 2-formatted sample metadata (tab-separated). 54 samples.
| Column | Description |
|---|---|
id |
Sample identifier, e.g. Site1a, Site5bb. Matches the SiteN... token in the FastQ filenames. |
site |
Collection site label, e.g. 1a, 5bb. First character is the site number (1 or 5). |
Sample counts: Site 1 = 28, Site 5 = 26.
Note
TODO — metadata enrichment. The current metadata has only id and site.
Consider adding collection date, latitude/longitude, replicate ID, sample
volume filtered, and any environmental covariates. Note also that 60 raw
sample sets were sequenced but 6 empty ones were dropped during 00.00/01.00,
leaving 54 — document which were removed and why.
00.00/01.00 rename samples from the raw
Yell_July18_Site1f_S92_L001_R[12]_001.fastq.gz form to the QIIME 2-compliant
Site1f_S92_L001_R[12]_001.fastq.gz form (Casava one-eight single-lane format)
before import in 02.00.
The 108 trimmed *.fastq.gz (~1.1 GB) in
output/01.00-trimming-fastp-fastqc-multiqc/ and the 75 MB
multiqc_data/multiqc_data.json have been untracked from git (they remain
on disk locally). They are now covered by .gitignore so they won't be
re-added. The FastQs can be regenerated by re-running
01.00 or re-downloaded from
the owl server; multiqc_data.json is regenerated whenever MultiQC runs, and
the human-readable multiqc_report.html summary is kept in git.
The .fastq.gz.md5 checksum sidecars are intentionally retained as a
provenance record for the (now server-hosted) trimmed reads.
Note
Untracking removes these files from future commits only; they still exist
in the packed history, so the .git directory does not shrink. If reducing
the on-disk clone size matters, rewrite history with
git filter-repo — coordinate
with collaborators first, since it changes every commit hash.
Kept in git: MultiQC summary reports, per-sample FastQC HTML reports,
metadata.tsv, the SILVA reference sequences, and the 02.00-qiime2
.qza/.qzv results.