Research Association ForInter

Computational identification of ligand-receptor binding in single cell RNA-seq and application to neuron-microglia interaction


Experimental breakthroughs in single cell biology now enable profiling molecular variation at multiple molecular layers and in multiple tissues in the same individuals at large numbers, increasingly also in single cells, and are currently revolutionizing molecular biology, in particular stem cell biology. The increasing complexity of data gives rise to the need for better analysis methods, especially in terms of function as well as performance. Basic questions such as differential comparison become interesting again since instead of comparing bulk averages single cell differences can be both on mean expression as well as distribution level.

In this project, we focus on computational methods to analyze single-cell RNA-sequencing (scRNA-seq) data of neural stem cells in the developing human brain modeled by 2D and 3D organoids. In particular, we ask how to estimate the role of intercellular crosstalk via ligand-receptor interactions. Building upon our experience in machine learning and data analysis in the context of single cell genomics, we propose to first identify receptor-ligand interactions from published scRNA-seq data using a new model for analyzing dependencies of single cell gene expression profiles adapted to the particular noise model, which we want to denote as count graphical modelling.
Second, we will develop a differential expression profiling method based on generalized linear models with adapted noise model to identify new ligand-receptor pairs. Finally, in close collaboration with the labs of B. Winner and B. Treutlein in ForInter, we aim to apply these methods to data from the consortium in order to predict novel ligand-receptor interactions via genetic screens, followed by prediction using perturbation screens and single molecule FISH.

Thus, our project will support experimentalists in generating hypotheses on ligand-receptor interactions in human brain development and in parallel provide versatile computational tools for a reliable discovery of cell-cell communication in scRNA-seq.

Project lead:

Prof. Dr. Dr. Fabian Theis
Department of Mathematics, TU Munich

Project team:

Maren Buettner


Launching date