Dimension Reduction by Spatial Components Analysis Improves Pattern Detection in Multivariate Spatial Data
We introduce a multivariate statistical approach for pattern recognition in spatial transcriptomics data. Our algorithm (SPACO) constructs a low-dimensional projection of the data maximising Morans I, which mitigates non-spatial variation and outperforms PCA for pre-processing. Our method also provides a calibrated, powerful test of spatial gene expression that excels in robustness and specificity.
bioinformatics↗