A Bayesian nonparametric method for jointly clustering multiple spatial transcriptomic datasets and simultaneous gene selection

Let \({\textbf{Y}}_{j}\in {\mathbb {R}}^{n_j\times p}\) denote gene expression for \(n_j\) cells and p genes in group/donor \(j=1,\dots , J\) and let \({\textbf{X}}_j\in {\mathbb {R}}^{n_j\times 2}\) denote the matching spatial coordinates of the cells. Our goal is to cluster cells into unknown cell types from multiple spatial transcriptomic datasets \(({\textbf {Y}}_j,{\textbf {X}}_j)_{j=1}^J\) and, simultaneously, select genes that are differentially expressed across cell types. We call our method, Multiple Spatial Sparse Clustering…