This lecture is open to RSU doctoral students, researchers or other interested participants. It will provide an overview of unsupervised machine learning approaches for biomedical data analysis, with a focus on applications in cell biology and medicine.
As high-dimensional datasets such as single-cell omics, imaging, and clinical measurements become increasingly common, methods for extracting meaningful patterns without predefined labels are essential. The talk will cover key dimensionality reduction techniques, including Principal Component Analysis (PCA), t-distributed Stochastic Neighbor Embedding (t-SNE), and Uniform Manifold Approximation and Projection (UMAP), highlighting their roles in visualization, feature extraction, and interpretation of complex biological data. In addition, clustering strategies such as hierarchical clustering, partitioning methods, density-based approaches, and graph-based algorithms will be discussed as tools for identifying cellular subpopulations, disease subtypes, and hidden biological structure. Examples from cell biology and medical research will illustrate how these methods support discovery and improve our understanding of biological systems and disease mechanisms.
The lecture will be held in English.
About the lecturer
Nikolay Oskolkov is a bioinformatician and Group Leader of the Metabolic Research Group at the Latvian Institute of Organic Synthesis (LIOS). With a PhD in theoretical physics (2007, Moscow State University / University of Ulm), he transitioned to life sciences in 2011 and has since applied statistical and machine learning methods to biomedical data. His research background spans medical genomics, diabetes, and cell and evolutionary biology at Lund University and NBIS SciLifeLab, Sweden.
