Prenatal exposures that produce phenotypes resembling Mendelian disorders. Each entry gives
the phenotype the exposure produces, the conditions it is mistaken for, the features that
separate them, and any shared mechanism.
An exposure appears here when it has all four of: a phenotype profile, a Mendelian
lookalike with an ontology identifier, a citation that resolves against PubMed, and a
clinical synopsis. A mechanism is not required. Exposures missing any of the four stay
in the dataset and are withheld from this page.
Phenotypes
From the HPO annotation release of 2026-09-02, which for these entries is Orphanet
curation, with this project's own annotations where marked. Frequency bands are Orphanet
estimates; measured rates are given as n/m with the cohort stated.
Evidence
Two grades, kept separate. Teratogenicity asks whether the exposure causes the
embryopathy in humans, on Shepard's criteria:
establishedemerginglimiteddisputed.
Mechanism asks whether the shared-mechanism claim is supported, on a ClinGen-style
rubric:
definitivestrongmoderatelimited.
An exposure can be an established teratogen while its resemblance to a given gene
disorder remains a hypothesis.
Curation
Every record names its curator, coded agent: or human:, so a
model's judgement can be told from a person's. Each annotation must quote the sentence it
came from, and a mechanical check rejects those that do not.
Sources
Phenotype annotations come from the HPO release, which draws on Orphanet, OMIM and
DECIPHER. This project's own curation is CC BY 4.0 and the code is MIT.
Download
teracopy.json, the whole dataset in one file,
including the entries withheld from this page and why. It also holds computed fields
the page does not show.
Limitations
Draft. Not clinical guidance and not citable. Discriminating features are curated from
the exposure literature rather than computed, and are absent where it does not support
them.