Cell Migration Lab website

Source of the Cell Migration Lab website, https://cellmig.org (the old address https://cellmigrationlab.github.io/website/ redirects there).

The site is built with Zensical (like the doctoral programme pages) from Markdown and YAML files. Every push to main rebuilds the site and publishes it with GitHub Pages; pull requests get a downloadable preview.

Papers, software and datasets update themselves from the things_done ledger — see integrations/things_done. MAINTAINING.md lists what is automatic and what needs a person (new papers, people, grants, pictures, the token).

Most pages are generated from the files in data/. Open the file on GitHub, click the pencil, edit, commit. Most pages also have an edit button (top right) that opens their source file; the home page and the pages made only from things_done (publications, datasets, lab in numbers) have none.

Who is in the lab and their roles come from things_done (ledger/profile/lab_members.yaml): add a person there with their role (and earlier roles in previous_roles), and set status: alumni when they leave. Current members show their current role; alumni show all their roles. No dates are shown. The website only adds the look:

• Photo: data/photos/<id>.jpg, where <id> is the person's things_done roster id without member- (member-ivan-hidalgo-cenalmor → ivan-hidalgo-cenalmor.jpg; the id stays the same if the spelling of the name is corrected), or photo: in data/members/<id>.yaml pointing to an image under docs/ (one or the other, not both). Photos are cropped to a square automatically; without a photo the initials are shown.
• Optional data/members/<id>.yaml with links and a one-liner:

orcid: 0000-0000-0000-0000     # also: email, github, bluesky, scholar, website
bio: Filopodia and cancer cell invasion
photo_position: center         # crop the photo from the centre instead of the top
now: Postdoc at …              # alumni: where they are now

Only the keys above are allowed, and every file in data/members/ and data/photos/ must match someone in the roster: a typo or a leftover file stops the build with a message instead of being silently ignored (after a name change in the roster, rename the files too).

Members can send a photo and links with the Lab member profile issue form (.github/ISSUE_TEMPLATE/lab-member.yml: New issue → Lab member profile; it is not linked from the public site).

Pictures can go anywhere under docs/ (e.g. docs/assets/images/); refer to them by their path from docs/, e.g. image: assets/images/new-figure.jpg. Large originals are fine: the build makes resized WebP copies. A .gif is shown as a looping video when an .mp4 of the same name exists in docs/assets/media/gif/ (much smaller than the GIF).

The files in docs/wp-content/uploads/ come from the old WordPress site and keep their old addresses, so existing links to them keep working.

pip install -r requirements.txt
python scripts/build_pages.py   # generates the pages from data/
zensical serve                  # http://localhost:8000, reloads on changes
# the publish workflow then runs: zensical build --strict && python scripts/write_sitemap.py

Generated pages (listed in .gitignore) are rebuilt every time; do not edit them, edit data/ instead.

Tests (also run by the publish workflow):

PYTHONPATH=scripts python -m unittest discover -s tests

1. GitHub Pages: repository Settings → Pages → Source: GitHub Actions, Custom domain: cellmig.org, Enforce HTTPS ticked.
2. The domain cellmig.org is registered at WordPress.com, which also runs its DNS (Domains → cellmig.org → DNS records):
   • four A records for @: 185.199.108.153, 185.199.109.153, 185.199.110.153, 185.199.111.153; www → CNAME cellmigrationlab.github.io;
   • TXT _github-pages-challenge-CellMigrationLab (the organisation's domain verification, Organisation settings → Pages, against takeovers);
   • the MX records are the lab's Google Workspace email: do not delete them;
   • the domain and the old WordPress plan are separate subscriptions: cancelling the plan does not cancel the domain, whose own renewal keeps going (WordPress.com → Purchases; never press Delete on the domain).
3. Old addresses keep working: every WordPress page (/lab-members/, /software/, /portfolio/<name>/, images under /wp-content/uploads/, the RSS feed at /feed/) exists at the same path. Removed pages (the old news posts, archives) redirect (see redirects in mkdocs.yml).

Visits are counted with GoatCounter (no cookies, no personal data). The statistics are at https://cellmig.goatcounter.com/ (log in with the lab's GoatCounter account). The script is added to every page by overrides/main.html from extra.goatcounter in mkdocs.yml (the site code, cellmig); remove that line to turn the counting off. The Licensing page tells visitors about it (content/licensing.md).

data/*.yaml ──┐
data/things_done/*.yaml ◀── things_done (automatic)
              ├─ scripts/build_pages.py ─▶ docs/*.md ─┐
content/*.md (hand-written pages) ────────────────────┼─ zensical build ─▶ site/ ─▶ GitHub Pages
overrides/ (layout), docs/assets/ (CSS, JS, fonts) ───┘

• scripts/build_pages.py — turns data/ into pages, makes thumbnails, the RSS feed and the footer. The code lives in scripts/cellmig/, one module per concern (see its __init__.py for a map; one module per page family). There are no silent fallbacks: a missing file, a DOI that is not in the ledger, an unknown key (or a leftover color), or a picture that does not exist stops the build with a message naming the file to fix.
• scripts/write_sitemap.py — after zensical build, writes site/sitemap.xml (every page, including the featured-paper pages that are not in the menu) and site/robots.txt pointing to it, both from site_url in mkdocs.yml.
• For search engines and LLMs the build also writes, from the same data: schema.org JSON-LD (the lab and its leader on every page; each featured paper as a ScholarlyArticle; publications, software, datasets and people as lists), Google Scholar citation_* tags on paper pages, and /llms.txt + /llms-full.txt (the whole site as Markdown).
• tests/ — unit tests for the generator.
• scripts/sync_things_done.py — copies the public part of the ledger into data/things_done/ (run by things_done after each update).
• overrides/ — page layout (header, footer, home page).
• .github/workflows/publish.yml — build and deploy.

• Code (scripts, templates, style sheets): MIT, see LICENSE.
• Images and videos are not covered by the MIT licence. Their rights are recorded one file at a time in data/media.yaml (lab/personal work with its creators, third-party material with its source, or unknown until established); credits shown on the site come from there, and the build stops if an image the site shows has no entry. Public facts only in that file. The public summary is the Licensing page (content/licensing.md).
• Fonts: Inter and Bodoni Moda, SIL Open Font License, self-hosted (no requests to Google).