SNVstory

SNVstory infers sub-continental genetic ancestry from genome sequencing data using machine learning to provide objective, quantifiable measures of ancestry.

Key Features:

• Sub-Continental Ancestry Inference: Three independent machine learning models determine sub-continental ancestry from sequencing data.
• Simulation of Individual Samples: Individual samples are simulated from aggregate allele frequencies of known populations to support inference.
• Feature-Importance Scheme: A feature-importance scheme identifies and tracks ancestral signals associated with specific genes or loci.
• Evaluation on Clinical Exome Data: Performance was evaluated on clinical exome sequencing datasets across 36 populations, demonstrating high accuracy.

Scientific Applications:

• Personalized Medicine: Inform clinical decisions such as genetic testing, health screenings, and medication dosing based on detailed ancestry.
• Reducing Ancestry Misclassification: Provide objective ancestry estimates to mitigate reliance on self-reported ancestry and reduce related health disparities.

Methodology:

Three independent machine learning models, simulation of individual samples from aggregate allele frequencies of known populations, and a feature-importance scheme were used, with evaluation on clinical exome sequencing datasets.