Traditional ML
Missense Variant Classification
ML pipeline classifying missense SNVs as pathogenic or benign for TEKNOFEST 2026.
A machine-learning pipeline that classifies missense single-nucleotide variants (SNVs) as pathogenic or benign for hereditary disease genes (PAH, CFTR and a hereditary cancer gene panel). Developed for the TEKNOFEST 2026 Health in AI competition.
Highlights
- Integrates ClinVar ground-truth labels, Ensembl VEP and dbNSFP in-silico risk scores (SIFT, PolyPhen, CADD, REVEL), GRCh38 sequence context and UniProt amino-acid context.
- End-to-end Jupyter pipeline covering data acquisition, preprocessing, feature extraction and model training.
- Documented feature logic and dataset processing rules for reproducibility.
- Bioinformatics
- Machine Learning
- TEKNOFEST
- ClinVar
- VEP
- Python