Journal: Communications Biology

Article Title: SEPAR enables spatial metagene discovery and associated molecular pattern characterization in spatial transcriptomics and multi-omics datasets

doi: 10.1038/s42003-025-09340-w

Figure Lengend Snippet: a SEPAR is a framework based on graph-regularized NMF designed to identify spatially aware metagene patterns using both spatial location and gene expression as input. Spatial location data is leveraged to construct weighted graph regularization capturing the spatial relationships of the spots or cells. Sparsity regularization and dissimilarity regularization are applied to ensure distinct spatial metagene patterns. b SEPAR supports efficient and robust downstream analyses of SRT data, including metagene expression pattern recognition, pattern-specific gene analysis, SVG identification, spatial domain delineation, gene expression denoising and spatial multi-omics data analysis.

Article Snippet: Beyond demonstrating robust performance in diverse spatial transcriptomics technologies (10 × Visium, Stereo-seq, osmFISH and MERFISH), where SEPAR consistently identified biologically meaningful spatial patterns and revealed tissue-specific expression programs across different resolution scales and measurement principles, SEPAR effectively handles multi-omics spatial molecular data.

Techniques: Gene Expression, Construct, Expressing, Biomarker Discovery

Journal: Briefings in Bioinformatics

Article Title: Spatial information matters: are traditional imputation methods effective for spatial transcriptomics data?

doi: 10.1093/bib/bbag027

Figure Lengend Snippet: Clustering benchmark results for the Stereo-seq dataset across three gene selection levels (2000 genes, 5000 genes, and all genes). Eight methods—including the base (no imputation) and seven imputation approaches—are evaluated using (a) ARI, (b) NMI, (c) AMI, and (d) HOMO scores. MAGIC method consistently performs well across all metrics

Article Snippet: Stereo-seq (Spatiotemporal Enhanced Resolution Omics-sequencing) is a high-resolution spatial transcriptomics technology that captures genome-wide gene expression with spatial fidelity using DNA nanoball (DNB)-patterned arrays [ ].

Techniques: Selection