Clustering spatial transcriptomics data