A five-book series on how biologic medicines are made, how their data is managed, and how to build the open-source stack that carries that data — written to be read from scratch, and translated in full into every locale the site ships.
Live site: https://bioprocess-guide.pages.dev
| Book | Route | What it covers |
|---|---|---|
| Biologic Drug Manufacturing | / |
How a monoclonal-antibody biologic is physically made, stop by stop |
| Data Management in Biopharmaceutical Manufacturing | /data-management |
The records, systems, and integrity rules around the process |
| Open-Source Bioprocess Data Systems | /open-source |
Building the plant data platform from open-source parts, with a runnable companion stack |
| Ontologies for Biopharmaceutical Manufacturing | /ontology |
The ontology-engineering lifecycle applied to one mAb campaign, with an executable requirements catalogue |
| Machine Learning & AI for Biomanufacturing | /ml-ai |
Soft sensors, release prediction, MPC, LLMs and agents — with a runnable ML suite |
Every book is written for a reader starting with no background in the subject and ends where a practitioner works. Chapter claims carry numbered citations to a per-chapter references page.
content/<locale>/ book prose (Markdown/MDX), one tree per locale — full translations
i18n/<locale>/ Docusaurus translation plumbing (docs dirs point into content/<locale>)
static/img/diagrams-* hand-coded SVG figures, one hero + one flow per page
static/img/photo-materials photographs (Wikimedia Commons; credited inline on each page)
examples/ the in-tree companion repo: Docker Compose data stack, datasets,
ontology (+ SHACL/competency-question validator), ML/PyTorch suite
scripts/ authoring and figure-audit tooling used to produce the books
Figures are hand-authored SVG — never raster or generated images, never live Mermaid. CLAUDE.md
documents the figure standard and the verification protocol they are held to.
npm install npm start # dev server (default locale) npm start -- --locale <locale> # dev server for one translated locale npm run build # builds every configured locale; must finish [SUCCESS] with no broken links npm run serve # serve the production build locally
The configured locales are declared in docusaurus.config.js (i18n.locales).
LOCALIZATION.md documents the directory layout, the parity rules a translation
must satisfy, and how to add a language.
The site is a static build deployed to Cloudflare Pages by direct upload:
npm run build npx wrangler pages deploy build --project-name=bioprocess-guide
(The npm run deploy script inherited from the Docusaurus template targets GitHub Pages and is not
used by this project.)
examples/ is a self-contained repo-in-repo. The Book 3 stack comes up with Docker Compose under the
project name bioprocess-data-stack; the Book 4 ontology validates with
examples/platform/ontology/validate.py (competency questions run as PASS/FAIL acceptance tests);
the Book 5 suite runs with examples/platform/ml/run_all.py. See examples/README.md.
Two licenses, because the repository holds two different kinds of work:
| What | License |
|---|---|
The book — content/, i18n/, the hand-authored figures in static/img/diagrams-*/, the logo and social card |
CC BY-SA 4.0 — see LICENSE-CONTENT |
The software — examples/, scripts/, src/, build config |
MIT |
Reuse the prose or a figure freely, including commercially: credit the source, link the license, and license your adaptation alike.
Two exceptions worth knowing before you reuse anything:
- Photographs under
static/img/photo-materials/andstatic/img/photo-cell/are third-party works from Wikimedia Commons, each under its own license (public domain, CC0, CC BY 2.0/3.0/4.0, or CC BY-SA 4.0). Photographer, license, and source are credited in the caption below each image; reuse follows that license. - Cited standards, specifications, and papers belong to their publishers. Titles, clause numbers, and short quotations appear here for citation and commentary only.
Prose, figures, and companion code are the author's own work, created with AI assistance where the per-figure attribution lines say so.