ParaPathology
A seven-year, 360-file collection of my own pathology notes published as a GitBook-style site — whose largest single section is statistics and bioinformatics, and which carries no licence at all despite asking for contributions.
Purpose
Personal notes on pathology and the tools around it, published openly for medical students, pathology residents and pathologists. The README states the intent plainly: “This booklet will contain my notes on pathology and the tools I use for my research.” Mostly English, some Turkish and German.
No local clone exists on this machine, so there is no local_path — unlike the other own repositories indexed here. This review was done from a scratch clone of the default master branch; repo: is the durable identifier and the only one that will keep working.
This is also the first public repository in the recent run of own-repo reviews. PinkKidney, PembeBobrek and PinkKidneyWeb are all private; this one is not, and that changes which questions matter. The two that do are licensing and third-party content, below.
Data used
No patient data, and this was checked rather than assumed — which matters because the previous review in this series did find an identifier mapping in a sibling repository.
Scanning the source markdown for Turkish and English identifier terms returns nothing real. The only hits are in statistics-bioinformatics-and-pathology/histopatoloji-calismalarinda-istatistik-icin-nasil-veri-hazirlanir.md (“how to prepare data for statistics in histopathology studies”), where TC Kimlik, Hasta No, Ad Soyad appears as a checklist item about which identifier fields to think about, next to “HIPAA kuralları” and “patient-based versus sample-based study”. That is teaching material about identifiers, not identifiers. MRN matched 1,662 files but zero markdown files — it is boilerplate inside vendored HTML. No CSV or spreadsheet is tracked anywhere.
Worth recording as a method note: the same column-name-first approach that caught the problem in PembeBobrek is what cleared this repository. Reading the context, not the pattern, is what distinguishes the two cases.
Methods
Not code — a content repository. 360 source markdown files organised into topic directories, assembled by a GitBook SUMMARY.md table of contents and published through Jekyll with the sighingnow/jekyll-gitbook remote theme (_config.yml). The live site at parapathology.com serves <title>Preface | (Para) Pathology Notes</title> with a GitBook (0.0.0) generator tag, matching the repository’s README.md.
R scaffolding sits alongside it — ParaPathology.Rproj, DESCRIPTION, renv.lock, and a GitHub Action — but the DESCRIPTION declares an R package that the repository is not.
The content inventory is the useful output here, because it does not match how this repo is described elsewhere in the wiki. Source markdown by section:
| Section | Files |
|---|---|
| statistics-and-bioinformatics | 64 |
| systemic-pathology | 43 |
| medical-school-lectures | 35 |
| pathology-residents-and-pathologists | 25 |
| writing-journal-articles | 22 |
| laboratory-management | 16 |
| pathology-tweets | 16 |
| bibliometrics | 13 |
| pathology-and-other-departments | 12 |
| computational-digital-and-mathematical-pathology | 11 |
| databases-and-pathology | 9 |
Roughly 100 of the 360 files sit in statistics, bioinformatics, computational pathology and bibliometrics — this wiki’s own core domains — which makes the largest cluster in the repository the one nothing in this wiki currently cites. See the correction noted under Patoloji Atlası.
Current state / open questions
Currency, read 2026-07-26 from the GitHub API. Created 2019-05-08, last pushed 2026-06-06 (Add pathology AI theory docs and infographic). No tags and no releases in seven years. 8 stars, 0 forks, 0 open issues. Active, in the sense that it is still being added to.
The licence question, which is the finding that matters
Three primary observations, all pointing the same way:
- The GitHub API reports
license: null. DESCRIPTIONstill carries the literal uneditedusethisplaceholder:License: `use_mit_license()`, `use_gpl3_license()` or friends to pick a license. The decision was never made, and the file records that it was never made.- The published site carries no licence, copyright or terms statement anywhere.
Against that sits an explicit invitation to reuse. The README displays a “contributions welcome” badge pointing at the issue tracker and says: “You may also comment on paragraphs on this page or make edit suggestions (fork) via GitHub.”
So the repository asks for contributions it grants no rights to accept or redistribute. Without a licence, default copyright applies and readers have no permission to fork, adapt or reuse the material the site invites them to fork. That is a conflict between two things the project itself says, not a conclusion drawn from outside it, and it is the owner’s call to resolve — nothing here should pick a licence on his behalf. It is also, mechanically, the cheapest thing on this page to fix.
Third-party full-text articles in a public repository
Clippings/ holds seven files saved with the Obsidian Web Clipper (frontmatter carrying title, source: URL, author list, publisher abstract, tags: clippings). Four are short stubs. Three are verbatim full text of published articles, which was verified rather than inferred — they retain the publishers’ own inline reference anchors and numbered reference lists:
| Clipping | Words | Source |
|---|---|---|
| The hallmarks of cancer immune evasion | 42,096 | 592 numbered references |
| Artificial intelligence in digital pathology — time for a reality check | 17,683 | Nature Reviews Clinical Oncology, s41571-025-00991-6 |
| Regression modeling of competing risk using R | 4,202 | Bone Marrow Transplantation |
This is a copyright question rather than a data-protection one, and it belongs in a different register from the identifier finding in PembeBobrek — nobody is put at risk. But the repository is public, and two of the three come from subscription journals. Options are .gitignore-ing the folder, keeping clippings in a private vault, or reducing them to citation plus notes; which one is the owner’s decision.
Deliberately not done during this review: none of that text was copied into sources/papers/. Doing so would reproduce the same problem inside a repository that also publishes. The two clippings genuinely relevant here — the AI reality-check Perspective, and the competing-risks regression guide that bears directly on Competing Risks — should be read from the journals and written up as ordinary source notes with citation only.
Smaller things
- A stale
mainsits beside the activemaster.masteris the default and was last committed 2026-06-06;mainlast moved 2023-07-02 and has diverged. Anyone cloning and assuming the modern default name gets three-year-old content. - Windows clone caveat. The longest path in the tree is 156 characters, so a clone into a deep target directory fails partway with
Filename too long, leaving 482 of 2,578 files. It is not broken —git clone -c core.longpaths=truechecks out all 2,578 — but it is worth knowing before assuming a failed clone means a corrupt repository. patoloji-hakkindais vendored here as rendered output, 1,662 HTML files with their own.github/workflows/patoloji-hakkinda.ymlAction. That accounts for most of the repository’s ~93 MB, and it also explains theMRNfalse positives above.
Open questions
- How is parapathology.com actually served? Three observations that do not fully reconcile: there is no
CNAMEfile in the tree,llms.txtis served at the domain (HTTP 200) but does not exist anywhere in the repository, and the Pages configuration endpoint returns 404 to an unauthenticated caller. The site’s content clearly matches this repository’sREADME.md, so it is served from here in some arrangement — but which branch, and where the extra file comes from, cannot be determined from outside. Only the repository owner can see the Pages settings. [unverified] - Is the divergence between
mainandmasterintentional, or ismainan abandoned artefact that should be deleted to prevent a confusing clone? - Does the
DESCRIPTION/renv/.Rprojscaffolding still do anything, or is it left over from when this was structured as an R project?
Related: Patoloji Atlası — the sibling publishing effort, and the page whose description of this repository this review corrects.
Related: PembeBobrek — the immediately preceding own-repo review, and the contrast worth holding: a private repository with a real data-protection problem against a public one that is clean on data and unresolved on rights.
Related: Competing Risks — one of the saved clippings is a competing-risks regression guide, which is a topic this wiki already covers and a paper worth reading properly from source.
Related: Bibliometrics — Who Cites Who — this repository holds a 13-file bibliometrics section that the bibliometrics project page does not reference.
Derived from: scratch clone read 2026-07-26 at master tip 30a860f (2026-06-06) — README.md, SUMMARY.md, _config.yml, DESCRIPTION, .github/workflows/patoloji-hakkinda.yml, the Clippings/ files, the full git ls-tree inventory, git ls-remote for branch state, the GitHub REST API for visibility, currency and licence, and https://www.parapathology.com/ for the published site and its absent licence statement.