Advisor: Joaquin Dopazo
Single cell RNA-seq data provides valuable insights into cellular heterogeneity which may significantly improve our understanding of biology and human disease. Nonetheless, there are some challenges to identify new cell types and cell states. The common methods to address this issue are unsupervised approaches, based on clustering methods together with methods to reduce the huge dimensionality of data. Although they are widely used, the inability to capture the complex patterns behind single cell data can lead to poor performance or misleading interpretations. In this project, neural network architectures are used to reduce the dimension of data and to find complex patterns in a supervised framework. To avoid the lack of interpretability of networks (also called black box), several types of prior biological information which are protein interactions and pathway are integrated in neural network architectures. The goal of this project is to integrate biological information to create an interpretable neural network.
- Deep learning approach to combine patient imaging and health records
February 2021 – April 2021Postponed due to Corona
- Kernel methods for network completion
Planned date: March 2022 – May 2022
Preprints and Reports
Pelin Gundogdu, Joaquín Dopazo, Isabel A. Nepomuceno-Chamorro, & Carlos Loucera. (2020, June 30). Integrating prior knowledge in neural network methods to reduce the dimensions of single-cell RNA-Seq Data. Zenodo. http://doi.org/10.5281/zenodo.3924254