RareGap names five upstream sources itself; a broad scan finds 82 candidate scientific providers it could draw on
Asked:
“What are ALL the third-party scientific providers RareGap (raregap.ai) could use?”
There is no closed universe of providers RareGap “could use”, so no list can be globally exhaustive. This live-web scan assembled a researched landscape of 82 distinct candidate providers and resources across 15 source hosts — spanning data and knowledge resources, computational platforms, sequencing and omics, wet-lab and preclinical CROs, clinical and regulatory CROs, and regulatory or public infrastructure. RareGap’s own FAQ, license and paper explicitly name five common current inputs (PubMed/NCBI, ChEMBL, Open Targets, GTEx, Orphadata/Orphanet) — “and others”, so even its disclosed list is open-ended.
The candidate landscape, grouped by provider type
One tile per provider. Colour is evidence status, not vendor status — hover or tap a tile for the stated potential use and its source.
Named by RareGap as a current common input (5)
On RareGap’s methods page as a ranking experiment, not a confirmed production input (1)
Reported as an input in related research papers, not on a RareGap primary page (2)
Potential provider — plausible option only (74)
What the tiles cannot say
RareGap’s FAQ and license pages name PubMed/NCBI, ChEMBL, Open Targets, GTEx and Orphadata/Orphanet as ingested-and-harmonized inputs to its Atlas, adding “and others” —
raregap.aiHPO appears on RareGap’s methods page only as a phenotype-ranking experiment that “did not promote candidates and in several cases regressed performance” —
raregap.aiAll 13 clinical/regulatory-CRO candidates come from a single market-landscape page, so treat that segment as breadth of options, not vetted shortlist —
verifiedmarketreports.comOrphanet supplies the 11,460 Orphadata-derived rare-disease entities indexed by RareGap Atlas v1 and the ORPHA IDs keying its API endpoints —
raregap.aiRead the automated counts with care
An automated pass tagged 8 rows “confirmed current upstream input”, but that bucket is over-inclusive: only the five sources above are named on RareGap’s own pages; DisGeNET and Scopus come from third-party research papers and HPO from an experiment write-up. Treat every other candidate as potential, not a confirmed RareGap vendor or integration. Before adopting any provider, follow RareGap’s workflow — evidence ingestion, computational inference, experimental validation, translation — and check licensing, redistributability, privacy and validation quality.
Method: researched landscape of 82 candidate third-party scientific providers/resources for RareGap, one row per normalized provider, gathered in a broad live-web scan across 15 source hosts; the largest single URL supports 14 of 82 rows (17.1%). Status reflects evidence basis (RareGap primary pages vs third-party sources), not commercial relationships. Counts are per-provider; long use/capability texts are truncated for space. This is a researched sample, not every provider on Earth.