The published Gene Knowledge Graph exactly as it stood on 2026-10-12. It has not changed since and will not.
That means everything published up to 2026-10-12. A new version is frozen every Monday, and each is bought once and kept.
The working catalogue is curated continuously and is always ahead of any sealed version — which is the whole reason this page exists. What is recorded here is what was published on 2026-10-12, and it will still say so in ten years.
+27,150 variants, +12 conditions, +46,914 links.
Counts only — added, corrected and removed rows all net out the same way here. GKG-Snapshot-6-2026-10-05 is its own citable version, unchanged by anything since.
One gzip-compressed UTF-8 JSON document, mygenelog-GKG-Snapshot-7-2026-10-12.json.gz
— 5287 KB compressed,
81.0 MB opened. A single object, not NDJSON and not
CSV, so one json.load reads all of it. The field names are the same ones the live
API uses, so code written against one reads the other.
{
"snapshot": { "version", "generated_at", "source", "note" },
"license": "…", "license_note": "…",
"counts": { "variants", "conditions", "drugs", "senses", "topics" },
"variants": [ { "rsid", "slug", "url", "gene", "trait_name", "risk_tier",
"genotypes", "guideline_note", "source_note", "updated_at" } ],
"conditions": [ { "slug", "name", "url", "category", "summary", "prevalence_note",
"inheritance_pattern", "external_codes", "pgx_notes", "source_notes",
"reviewed_at", "updated_at", "license", "license_note",
"faq_count", "open_questions", "content" } ],
"drugs": [ { "slug", "name", "kind", "summary", "genes", "url" } ],
"senses": [ { "slug", "name", "tagline", "summary", "variants", "url" } ],
"topics": [ { "slug", "label", "label_ko", "hook", "why_now", "source",
"terms", "confirmed_variants", "confirmed_note", "url" } ]
}genotypes is the per-genotype
reading of a variant, external_codes the MONDO / OMIM / ICD-10 identifiers, and
confirmed_variants the rsIDs a topic is actually backed by. Opening it needs
nothing installed:
gunzip -c mygenelog-GKG-Snapshot-7-2026-10-12.json.gz | jq '.counts'
python3 -c "import gzip, json; d = json.load(gzip.open('mygenelog-GKG-Snapshot-7-2026-10-12.json.gz')); \
print(d['counts'], d['variants'][0]['rsid'])"If somebody hands you this file, that digest is how you know it is the one we published:
shasum -a 256 mygenelog-GKG-Snapshot-7-2026-10-12.json.gz
Cite the version rather than the live API. The API answers with today's data by design, so a figure taken from it cannot be checked later; this page can.
MyGeneLog™ dataset snapshot GKG-Snapshot-7-2026-10-12. MyGeneLog™. 2026. https://doi.org/10.5281/zenodo.23314933
Registered with Zenodo, CERN's open-science archive — a DOI is the identifier a paper's own reference list expects, and it resolves on its own even if mygenelog.com ever does not. Use 10.5281/zenodo.23314933 to cite this exact version; 10.5281/zenodo.22776633 always resolves to whichever version is newest, for anyone who means "the current snapshot" rather than this specific one. The Zenodo record itself is closed access — it is the citation and the digest, not a second way to download the file; the file stays here.
5287 KB of gzipped JSON: every published variant, condition, drug, chemosensory group and topic as they stood on 2026-10-12 — with the genotype interpretations, the MONDO/OMIM/ICD-10 codes, every cited source and the date each page was last fact-checked. Joined rather than five separate lists: each condition carries the rsIDs linked to it, each sense group and topic the variants behind it, each drug its genes — the part a scrape of the pages does not give you. Yours to keep, with no expiry and no limit on how often you download it.
This is a file, not live access. A new version is frozen every Monday, and each is a separate purchase — but almost nobody needs more than one: you buy the version your work cites, and it stays valid for as long as the work does. A dated version does not go stale; the date is what makes it worth citing. If what you need is queries against today's data, that is API Pro instead.
An account is needed before paying rather than after, because the file is unlocked for an account. Create one — it is free and takes a minute.
The record above stays public whether or not you buy anything — the version, the counts and the digest are what a reviewer checks, and a citation nobody can check is not a citation.
Structured fields are CC BY 4.0 — free to use, including commercially, with attribution. Not covered: the written sections, and the links between variants, conditions, drugs and topics.