A self-guided lab notebook · runs in your browser

The gut microbiome,under glass.

A nine-module curriculum in gut microbiome bioinformatics — covering microbial taxonomy, compositional data analysis, and the statistical methods applied to 16S rRNA sequencing data. Replicates a published cross-population study, implements alpha- and beta-diversity metrics from first principles, and surveys the computational pipelines — QIIME2, DADA2, GMWI2 — used to generate this kind of data in practice. Runs natively in-browser via a WebAssembly Python runtime; no local installation required.

Start the course →
Fig. 1 — Sample composition, relative abundance
0%25%50%75%100%
Other genera48.4%
Bacteroides23.7%
Prevotella15.1%
Faecalibacterium7.9%
Bifidobacterium3.6%
Escherichia1.3%
Relative abundance of six genera in one real sample (S03) from this course's teaching dataset. Notebook 02 reconstructs this exact chart; Notebook 01 covers what each genus does.

Specimen log — course order

№ 00 Welcome
How this course works, and how to run a cell
 → № 01 Gut Microbiome Biology Primer
Phyla, genera, the gut-brain axis, and what these CSVs actually measure
 → № 02 Pandas & Plotting Basics
Load data, explore it, make your first boxplot
 → № 03 P-values & False Positives
What a p-value actually means — and how it can mislead you
 → № 04 Multiple-Testing Correction
Why testing hundreds of bacteria at once needs a stricter bar
 → № 05 Real USA–Malawi Diet Study
A published dataset, and one of the field's most reproducible findings
 → № 06 Diversity Metrics, By Hand
Shannon, Simpson, and Bray-Curtis, built yourself with numpy
 → № 07 Bioinformatics Tools Walkthrough
How QIIME2 and DADA2 turn raw DNA reads into the tables you've been using
 → № 08 GMWI2 Wellness Index
A real 2024 Mayo Clinic tool that scores gut health from a stool sample
 →
Nothing to install. These notebooks run on Pyodide — a real Python interpreter compiled to WebAssembly — entirely inside this browser tab. The first load takes a few seconds while it downloads; after that, every import pandas and every plot is genuine, live code, not a recording.