Evagene adds Related Concepts — an educational correlation graph
Evagene already catalogues diseases, traits, clinical-test results, allergies, and genetic markers. This release connects them. A hand-curated correlation graph records educational associations between those entities, and a new Related concepts panel surfaces them in the editor for whatever is recorded on an individual. It is reference data — in the same family as the catalogues it links — and it is not risk analysis, not diagnosis, and not a claim about any person.
Audience: educators and students, researchers, genetic counsellors, and clinical geneticists.
What shipped
The correlation graph is a catalogue of educational associations between the five entity kinds Evagene already knows about. Each association is an edge joining two endpoints, carrying a relationship term from a fixed eight-term vocabulary, a short neutral educational note, an optional strength (strong / moderate / weak), and a direction flag. The full model is described on the Related Concepts Explorer page.
1. Two sources: curated and marker-derived
The graph is assembled from two sources at start-up:
- Curated edges — over 1,100 hand-authored associations, each with its own educational note: biomarkers and findings, associated conditions, shared features, sequelae, and risk associations.
- Auto-derived genetic edges — Evagene reads the marker catalogue and generates two kinds of edge on the fly, tagged
marker-derived: agenetic_associationedge between a gene / marker and each disease, trait, or allergy it is linked to, and ashared_geneedge between any two entities linked to the same marker. See gene and shared-gene associations.
2. Status-qualified clinical-test edges
A clinical-test result means different things depending on whether it is low or high, so a clinical-test endpoint can carry a status (abnormal_low or abnormal_high). The graph surfaces different associations for each direction. A low ferritin is associated with iron-deficiency anaemia and restless legs syndrome; a high ferritin is associated with hereditary haemochromatosis. A test recorded without a low/high mark surfaces only direction-independent associations. The biomarker and condition associations page works the iron studies through in full.
3. The Related concepts panel
With an individual selected, the panel opens from the dock button between Genetics and Risk analysis. It lists concepts associated with the items already recorded on that person, grouped per recorded item (for example, "Because Ferritin (low) is recorded:"). Ticking a suggestion records it on the individual; unticking removes it. Nothing is ticked by inference — a concept shows as "recorded" only when a user has genuinely recorded it. An "explore any concept" search browses associations for any catalogue entity without recording it on anyone. An educational-use banner is shown by default and is suppressible per user from the Account tab.
4. API and MCP access
The same data is available through the open GET/POST /api/correlations* endpoints. Because the graph is static reference data that produces no risk number, these endpoints are not gated by the intended-use acknowledgement. For AI agents, the MCP server gains a related_concepts tool — the fifteenth tool on the server — which returns related concepts for an anchor and always carries the educational-use disclaimer. See the MCP server and platform pages.
What Evagene does not claim
- The correlation graph does not diagnose and asserts nothing about any individual.
- It produces no risk number and runs no inheritance analysis.
- A
shared_geneedge is a catalogue relationship, not a claim of shared risk, mechanism, or inheritance. - Notes are neutral and educational; they describe an association and carry no screening, referral, or treatment instruction.
- Nothing is recorded on a person by inference — the user does the asserting.
- Evagene is an academic, research, and educational pedigree modelling platform — not a medical device, not clinical decision support, and not a diagnostic or screening tool.