B1711
Title: Clustering spatially resolved genes at a spot level in spatial trascriptomics with SpaRTaCo
Authors: Andrea Sottosanti - University of Padova (Italy) [presenting]
Davide Risso - University of Padua (Italy)
Abstract: Spatial transcriptomics is a modern sequencing technology that allows measuring the activity of thousands of genes in a tissue sample and map where the activity is occurring. The increasing popularity of such advanced technology has grown the interest for the so-called spatially expressed genes, i.e. genes whose expression in a cell affects the expression in the surrounding ones. Comprehending the functions and the interactions of such genes across multiple cells is of great scientific interest because it might lead to a deeper understanding of several complex biological mechanisms. Another relevant aspect in the analysis of tissues is the classification of the cells. Distinguishing, for example, a tumor cell from a stromal or an immune cell is vital. We present SpaRTaCo, a new advanced statistical method for the analysis of spatial transcriptomic experiments. We investigate the properties of the matrix variate distributions to infer the latent block structure at the base of the data, dividing both the genes and the cells into clusters. This procedure, known in statistical literature as co-clustering, allows us to classify the nature of cells and to detect groups of spatially expressed genes only in some specific areas of the tissue sample.