OmicVerse Microbiome — 16S amplicon end-to-end

Goal

Take paired-end Illumina FASTQs from a 16S rRNA gene survey through to an analysis-ready AnnData (samples × ASVs, with 7-rank SINTAX taxonomy in var and the ASV centroid sequence stored alongside), plus the de-facto downstream stack: alpha-diversity (Shannon / observed OTUs / Simpson) → beta-diversity (Bray-Curtis distance matrix) → PCoA / NMDS ordination → differential abundance (Wilcoxon by default; pyDESeq2 / ANCOM-BC available).

Cover two backends behind the same wrapper:

• vsearch / UNOISE3 — the default; fast (single-threaded C, parallelised via jobs); slightly more reads retained per sample on noisy datasets.
• DADA2 — Bayesian denoising; longer to fit but produces tighter ASV-level resolution. Same wrapper, just backend='dada2'.

Plus the stepwise API for users who want to inspect or substitute individual stages (merge → filter → dereplicate → UNOISE3 → uchime3 → SINTAX → OTU table → AnnData).

Quick Workflow