concept · intermediate · partly written
Hetionet
Hetionet is a 2017 heterogeneous network of 47,031 nodes of 11 types and 2,250,197 edges of 24 types, built to rank drugs for repurposing.
Also called: Rephetio, Project Rephetio
Short answer
Hetionet integrates 29 resources into 47,031 nodes and 2,250,197 edges. The Rephetio model ranked drugs by degree-weighted path counts. Its authors printed its flaws: a degree-only prior scored 97.9% on training positives but 54.1% on external indications. 12
Reported, not yet replicated
Measured results more
AUROC was 97.4% on the 755 known treatments, 85.5% on DrugCentral indications not used in training and 70.0% on clinical-trial indications; the degree-only prior scored 97.9%, 54.1% and 62.5% on those sets. Two curators disagreed on 444 of 1,388 indication calls (kappa 49.9%). The authors close by saying additional studies will be needed to test the predictions and list 3,394 repurposing candidates above 1% probability. 1
Reported, not yet replicated
Its one evolutionary edge type more
Gene-covaries-Gene has 61,690 edges, based on evolutionary rate covariation (whether two genes' evolutionary rates move together across species). The authors call its predictive contribution "weak yet appreciable" and see synergy between evolutionary genomics and disease biology. It is rate covariation, not orthology. Hetionet's `binds` edge hides agonist versus antagonist: 15 of the top 100 epilepsy predictions were ictogenic. 1
Reported, not yet replicated
Grounding deep
The authors mapped by gene symbol only in a few cases and call that method disfavoured. The mapping method as a whole was not read. 1
Reported, not yet replicated
Still open here
- The Hetionet edge threshold for evolutionary rate covariation is given as 0.75 by one report and 0.7 by another; the edge count 61,690 agrees.
- The Rephetio 97.4% looks like in-sample performance; the authors' own wording on this was not seen.
- Source-database licences: the content is released CC0, but some upstream databases had no licence or forbade redistribution; this was not checked.
Not yet written: Plain answer, Why it matters, Equation, Real numbers.
Linked from
- Biomedical knowledge graphs: shapes, edges and what has worked
- How a link-prediction benchmark is built and how it leaks
- Purpose as structure: scoring a target differently by indication
- The QIAGEN Knowledge Base (behind Ingenuity Pathway Analysis)
- Is orthology or evolutionary information an edge type anywhere, and does it help prediction?
- Has any top prediction from a public biomedical graph been tested prospectively?
- What does an edge type connote to a practitioner beyond its definition?
- What fraction of a graph benchmark score survives a degree-only baseline and a by-disease split?
Sources
- Himmelstein DS, Lizee A, Hessler C, et al.. Systematic integration of biomedical knowledge prioritizes drugs for repurposing. eLife 6:e26726 (2017). paper; opened 2026-11-18 all records citing this
- Himmelstein DS, et al.. Hetionet repository and metagraph. GitHub (hetio/hetionet) (2017). database; opened 2026-11-18 all records citing this
Where it sits on the map
History of this record (5)
This record has been changed 5 times, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.
- Status
- partly written
- Last updated
- 2026-11-19
- Record
- content/concepts/hetionet.json