Projects & Grants
The HRS4R human resources management strategy for the scientific sector was funded through the project „HR Excellence in Research na Ostravské univerzitě“, Reg. No.: CZ.02.2.69/0.0/0.0/16_028/0006225.
| Vývoj genomových referenčních panelů pro vizualizaci populační struktury pomocí vysoce optimalizovaného manifold learningu | |
|---|---|
| Project Id | RRC/16/2024 |
| Main solver | M.Sc. Pavel Flegontov, Ph.D. |
| Period | 6/2026 - 5/2028 |
| Provider | Katedra biologie a ekologie, OP ST (Spravedlivá transformace) |
| State | solved |
| Anotation | Principal component analysis (PCA) is a standard first step in archaeogenetic studies, but low-dimensional plots of genetic data can be misleading. They are sensitive to sampling bias and outliers, compress complex patterns into only two or three axes, and often fail to capture fine-scale population structure. These limitations are especially important for ancient DNA, where low sequencing coverage, missing data, and heterogeneous laboratory protocols complicate interpretation. Our recent work indicates that carefully optimized manifold-learning approaches can recover genetic gradients and migration patterns more clearly than standard PCA, but their performance depends strongly on the quality and design of reference datasets. This project will create improved reference panels for visualizing genetic structure in humans, with a focus on Eurasia. We will compile at least three complementary panels of individuals from high-quality present-day whole-genome shotgun data, ancient whole-genome shotgun data, and high-coverage ancient capture data, and compare them with a widely used present-day panel of SNP array data. A key innovation is the inclusion of rare variants and ancient reference data, which should improve the resolution of closely related populations and better reflect past genetic diversity. We will evaluate these panels by combining simulations with analyses of real datasets and by testing both PCA projections and deeply optimized manifold-learning workflows. Performance will be assessed through preservation of genetic distances and local neighborhoods, resolution of known population structure, and separation of related groups. The expected result is a robust, unsupervised framework for clearer and more informative visualization of ancient and present-day genetic structure. The project will deliver new reference resources, improve analytical practice in archaeogenetics, and enable finer-scale studies of mobility, admixture, and population history in Eurasia. |
| Total Costs | 3 466 848 CZK |
