Computational MOdeling Lab

Accessibility to genomic-scale information increased exponentially in the last decades. Quantitative readouts of gene expression and protein concentration available with previously unprecedented abundance and generation of such data becomes increasingly faster as molecular techniques are becoming even more precise and cheap. This situation gradually enables us to develop novel analytic tools capable of extracting additional layers of information from high-throughput measurements. Our research focuses primarily on machine learning and graph theory inspired method development for comparative analysis of human brain aging and neurodegeneration omics datasets.

Functional connectivity methods create models for measuring interactions between distant brain regions based on the synchronized timing of neural activity. Graph theoretical analysis of these functional networks can identify local or global information processing strategies of our brain. We apply sophisticated techniques for modelling the direction of the information flow, the circularity of the connections and the re-organization of these networks in different clinical states such as epilepsy, or aging.

Questions usually build on preconceptions. However, investigators cannot be sure, whether their preconceptions lead to the best questions to understand a psychological phenomenon. Free word association technique - on the contrary – allows relatively unrestricted access to mental representations, it is easy to apply and provides fast and cheap data collection. So far the identification of prominent opinions from a large amount of diverse associations was a drawback of the word-association method. We develop and validate techniques for mapping freely expressed opinions and cognitive schemes from multiple word associations.