Sini Kerminen

KANN estimates genetic ancestry profiles using k-nearest neighbor regression on principal components, showing strong agreement with haplotype-based methods in large-scale genomic data.

The paper develops metabolomic risk scores from half a million blood samples to predict 12 common diseases, showing they outperform polygenic scores and benefit from repeat measurements.

This study evaluates polygenic risk score methods across diverse biobanks and ancestries, finding that European reference panels often yield better performance, while PRS-CS outperforms P+T for high-heritability traits.

This study examines geographic variation and bias in polygenic scores for complex traits in Finland, finding that population genetic structure may cause apparent geographic patterns rather than direct genotype-phenotype associations.

The paper investigates the genetic contributions to intellectual disability, finding rare variants are more common in severe cases and common variants contribute to all severity levels.

This study uses haplotype sharing among Finns to reconstruct recent population history and identify regions with increased rare disease variant sharing.