Clustering and detection of rare cells in single cell open chromatin sequencing

• Applicant: Professor Dr. Ivan Gesteira Costa Filho
• Subject Area: Bioinformatics and Theoretical Biology
• Term: since 2017
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 388802535

The combination of open chromatin protocols with single cell sequencing (scATAC-seq) allows the dissection of regulatory features of thousands of single cells from whole tissues. The detection of groups of cells is a crucial task in the analysis of scATAC-seq. However, the high dimensionality and sparsity of single cell open chromatin data imposes great methodological and technical challenges in cluster analysis. A yet overseen problem is the detection of rare cells, which is crucial in experiments measuring diseased tissues. Particularly challenging is the fact that methods for detection of the feature spaces of open chromatin matrices, i.e. genomic regions with open chromatin regions, are likely to miss regulatory regions specific to rare cells, as these have very low read coverage. This project has as objective the investigation of the impact of dimensionality and sparsity of scATAC-seq data in tasks associated to clustering as feature space definition, dimension reduction, proximity indices and clustering algorithms. In particular, we will focus on clustering methods for detection of rare cells. For this, we will propose novel approaches for selection of genomic regions, which are discriminative of rare cells, and explore density based clustering methods for detection of rare cells. Proposed methods will be applied to novel scATAC-seq from collaborative efforts, i.e. the identification of rare gut epithelial cells in neonates and fibrosis-causing cells in heart and kidney.

• DFG Programme: Research Grants