CoolFace
Modelpublic

opentargets/locus_to_gene_26.03-test-new_training_set_dark_matter_1

sourceHugging Facemitupdated 4mo agoView on Hugging Face
0likes
Model Card

Locus-to-Gene (L2G) Model

The locus-to-gene (L2G) model prioritises likely causal genes at each GWAS locus based on genetic and functional genomics features.

Model Description

This is a Gradient Boosting Classifier (XGBoost) trained to predict causal genes at GWAS loci.

Limited to protein-coding genes with available feature data.

Key Features:

  • —Distance: proximity from credible set variants to gene
  • —Molecular QTL Colocalization: evidence from expression/protein QTL studies
  • —Variant Pathogenicity: VEP (Variant Effect Predictor) scores

Usage

python
from gentropy.method.l2g.model import LocusToGeneModel
from gentropy.common.session import Session

# Load model from Hugging Face Hub
session = Session()
model = LocusToGeneModel.load_from_hub(
    session=session,
    hf_model_id="opentargets/locus_to_gene"
)

# Make predictions on your L2G feature matrix
predictions = model.predict(your_feature_matrix, session)

Training

  • —Architecture: XGBoost Gradient Boosting Classifier
  • —Training Data: Curated positive/negative gene-locus pairs from Open Targets
  • —Evaluation Metric: Area under precision-recall curve (AUCPR)

Citation

If you use this model, please cite:

bibtex
@article{ghoussaini2021open,
title={Open Targets Genetics: systematic identification of trait-associated genes using large-scale genetics and functional genomics},
author={Ghoussaini, Maya and Mountjoy, Edward and Carmona, Maria and others},
journal={Nature Genetics},
volume={53},
pages={1527--1533},
year={2021},
doi={10.1038/s41588-021-00945-5}
}

More Information