Quality Control
Distributions of key QC metrics across body sites and modalities.
Metrics are derived from a stratified sample of 2,000 samples;
synthetic read-depth and alpha-diversity values are illustrated for demonstration —
replace with real per-sample metrics after computing from the parquet files.
How to compute real QC metrics:
Shannon diversity from the genus/species relab parquet files using
scipy.stats.entropy; read depth from sequencing run metadata;
unclassified fraction from the UNCLASSIFIED column in MetaPhlAn output.
Alpha Diversity (Shannon H′) by Body Site
Violin + box — urogenital shows widest range due to Lactobacillus dominance vs. dysbiosis states
Read Depth Distribution by Sequencing Modality
Histogram (overlapping) — NovaSeq shotgun samples are generally deeper
Detected Taxa per Sample by Body Site
Box plot with outliers — stool has highest alpha diversity; milk has fewest taxa
Unclassified Read Fraction
Histogram across all samples — values >40% may indicate host contamination
QC Summary by Body Site
| Body Site | N samples | Mean Shannon H′ | Median Shannon H′ | Mean Read Depth | Mean Taxa/Sample |
|---|
Based on 2,000-sample stratified subset with synthetic QC values for illustration.