Project title: Topic modeling for multimodal single-cell data

The project aims to develop a new topic modeling algorithm able to compress data from multimodal single-cell sequencing into a form that is drastically smaller in size and easily interpreted. To the best of our knowledge no one has yet attempted to model multimodal topics for single-cell sequencing data. Using the complete multimodal information to create topics will unlock much more detailed analyzes that are impossible with current methods. Piotr will apply modern techniques from the field of Bayesian probability theory to carry out individual research tasks. He will propose different model architectures based on existing well-functioning models and adapt them to single-cell data. He will improve the models iteratively. The resulting models will be programmed and made available in a public repository as an open-source software package. The project will produce valuable insights into topic modeling and single-cell genomics by producing novel tools for data analysis. The total funding of the project is 139 568 PLN.