Datasets:

ClarusC64
/

genetics-population-transfer-integrity-v0.1

id stringclasses 8 values	genetic_context stringclasses 8 values	discovery_population stringclasses 8 values	evidence_pack stringclasses 8 values	claim stringclasses 8 values	target_population stringclasses 8 values	transfer_assumptions stringclasses 8 values	transfer_failures stringclasses 8 values	missing_transfer_evidence stringclasses 8 values	clinical_risk_if_applied stringclasses 8 values	notes stringclasses 8 values	constraints stringclasses 1 value	gold_checklist stringclasses 1 value
GPTI-001	PRS	European ancestry adults	PRS AUC 0.64 in UK Biobank. No non-European validation.	Accurate risk prediction.	Global populations	LD structure and effect sizes assumed stable.	PRS performance collapses outside European ancestry.	No multi-ancestry validation; no recalibration.	misclassification; inequitable care	Classic PRS transfer failure	Under 220 words.	assumptions+failures+risk
GPTI-002	rare_variant	Northern European families	Segregation in 3 families. Penetrance ~80%.	High penetrance pathogenic variant.	East Asian populations	Penetrance assumed invariant across ancestry.	Background modifiers differ; penetrance likely lower.	No ancestry-stratified penetrance data.	overdiagnosis; anxiety; unnecessary intervention	Penetrance transfer error	Under 220 words.	assumptions+failures+risk
GPTI-003	pharmacogenomics	European trial cohort	CYP variant linked to toxicity.	Dose adjustment required universally.	African ancestry patients	Allele frequency and haplotype assumed similar.	Variant rare; different functional alleles dominate.	No African pharmacogenomic data.	undertreatment or toxicity	Allele frequency ignorance	Under 220 words.	assumptions+failures+risk
GPTI-004	GWAS	European cohort	Index SNP associated; fine-mapping unresolved.	Causal variant identified.	Latino populations	Tag SNP assumed causal across ancestries.	LD differs; tag breaks.	No trans-ethnic fine-mapping.	false causality; wrong mechanism	LD transfer failure	Under 220 words.	assumptions+failures+risk
GPTI-005	cancer_risk	Ashkenazi Jewish cohort	Founder mutation high risk.	Same risk estimate applied.	General population	Founder effect ignored.	Risk vastly lower outside founder group.	No population prevalence adjustment.	overtesting; harm	Founder mutation misuse	Under 220 words.	assumptions+failures+risk
GPTI-006	monogenic_score	European neonates	Score predicts disease onset.	Used for universal screening.	African neonates	Score transport assumed.	Different variant spectrum.	No ancestry-specific calibration.	missed cases; false reassurance	Screening inequity	Under 220 words.	assumptions+failures+risk
GPTI-007	polygenic_trait	European adults	BMI PRS validated.	Predicts obesity risk.	South Asian populations	Environmental interaction ignored.	Gene–environment differs.	No interaction modeling.	misleading prevention advice	Context collapse	Under 220 words.	assumptions+failures+risk
GPTI-008	carrier_screen	European carrier data	Carrier frequency established.	Applied universally.	Middle Eastern populations	Carrier rates assumed similar.	Different pathogenic variants.	No regional panels.	missed carrier couples	Reproductive risk error	Under 220 words.	assumptions+failures+risk