This research investigates how normalization methods affect spatial domain identification in spatial transcriptomics data. It challenges the direct adoption of single-cell RNA-sequencing (scRNA-seq) tools, demonstrating that library size is associated with tissue structure. Normalizing these effects out can negatively impact spatial domain identification, urging caution when adopting scRNA-seq algorithms for spatial data. By analyzing spatial molecular data, the study reveals that library size correlates with tissue structure. Applying scRNA-seq normalization methods designed to remove library size effects can distort the spatial information, leading to inaccurate identification of spatial domains. The findings suggest that spatial data should not be specifically corrected for library size prior to analysis and that algorithms designed for scRNA-seq data should be applied with care. This study highlights the importance of appropriate normalization techniques for spatial transcriptomics data analysis.
This study on spatial transcriptomics data analysis is well-suited for Genome Biology, given its focus on genetics and genomics. The research provides critical insights into data normalization techniques, contributing to more accurate and reliable analyses in this field.