GenomeRiskLab PRS Model Card: {DISEASE_NAME}

RESEARCH USE ONLY — Not for clinical or diagnostic purposes.

Model Details

Field	Value
Disease	{DISEASE_NAME}
PGS Catalog ID	{PGS_ID}
PGS Version	{PGS_VERSION}
GWAS Source	{GWAS_SOURCE}
Genome Build	{GENOME_BUILD}
N Variants	{N_VARIANTS}
Training Ancestry	{TRAINING_ANCESTRY}

Intended Use

Intended use: Population-level research and educational demonstration of polygenic risk scoring methodology.
Out-of-scope: Clinical diagnosis, individual patient risk communication, direct-to-consumer health decisions.

Model Architecture

This is an additive linear PRS model:

PRS = Σ (dosage_i × effect_weight_i)

where dosage_i ∈ {0, 1, 2} is the count of effect alleles at variant i and effect_weight_i is the log-odds (or beta) weight from PGS Catalog.

Training Data

Weights were derived from a GWAS meta-analysis ({GWAS_SOURCE}). Full details at the PGS Catalog entry: https://www.pgscatalog.org/score/{PGS_ID}/

Calibration

Raw scores are calibrated against 1000 Genomes Project Phase 3 distributions (genomerisklab-calibration-1kg). Percentile and risk band are ancestry-specific.

Performance

Metric	Value	Population
AUC	{AUC}	{EVAL_POPULATION}
Nagelkerke R²	{R2}	{EVAL_POPULATION}

Limitations

PRS performance varies substantially across ancestry groups; models trained predominantly on European cohorts may underperform in other populations.
Raw scores are not probabilities and do not indicate absolute disease risk.
Environmental, lifestyle, and rare variant factors are not captured.
Completeness < 80% (fewer than 80% of model variants present in the input VCF) flags results as unreliable.

Ethical Considerations

Results carry mandatory "RESEARCH USE ONLY" labels at all layers.
Ancestry-matched calibration is used where available to reduce bias.
PRS and rare variant annotation are strictly separated to prevent conflation of population-level and individual variant risk.

Creator / Author

Dr Siddalingaiah H S Professor, Department of Community Medicine Shridevi Institute of Medical Sciences and Research Hospital, Tumkur, India

Citation

If you use this model, please cite the original GWAS and the PGS Catalog:

@article{lambert2021pgscatalog,
  title={The Polygenic Score Catalog as an open database for reproducibility and systematic evaluation},
  author={Lambert, Samuel A and others},
  journal={Nature Genetics},
  year={2021},
  doi={10.1038/s41588-021-00783-5}
}

Disclaimer

This model card template is provided for informational purposes. Actual model performance must be validated on independent cohorts before any use. This software is not a medical device and has not been evaluated by any regulatory authority.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support