A spectral dimension reduction technique that improves pattern detection in multivariate spatial data

TL;DR

This paper introduces a new method for analyzing spatial gene expression data that improves pattern detection by reducing non-spatial noise.

Contribution

A novel dimension reduction technique that optimizes spatial dependency for multivariate spatial transcriptomics.

Findings

The method outperforms PCA in capturing spatial patterns in gene expression data.

The projection reduces noise and enhances the detection of spatially variable genes.

The algorithm requires no parameter tuning and provides a calibrated statistical test.

Abstract

We introduce a statistical approach for pattern recognition in multivariate spatial transcriptomics data.

Our algorithm constructs a projection of the data onto a low-dimensional feature space which is optimal in maximizing Moran’s I, a measure of spatial dependency. This projection mitigates non-spatial variation and outperforms principal components analysis for pre-processing. Patterns of spatially variable genes are well represented in this feature space, and their projection can be shown to be a denoising operation. Our framework does not require any parameter tuning, and it furthermore gives rise to a calibrated, powerful test of spatial gene expression.

The algorithm is implemented in the open source software R and is available at https://github.com/IMSBCompBio/SpaCo.

Figures