#40 · Proteomics

Spatial statistics and dimension reduction resolve molecular patterns in omics data

The common thread is statistical restructuring of high-dimensional omics measurements to recover weak or spatially organised effects before testing. A spectral method maximizing Moran's I denoises spatial transcriptomics and yields a calibrated test of spatial structure, while SAKURA uses genes of interest to recover rare single-cell populations and separate similar ones. Fisher's method aggregates peptide-level p-values into protein-level statistics for target identification in chemical proteomics, and Fuzzifier* applies fuzzy categorization with consensus and path-specific features to multi-omics differential analysis. The set is heterogeneous: only two items address spatial data directly, and the others operate on peptide, multi-omics or single-cell data. What is new is the emphasis on aggregation and dimensionality reduction as the locus of improved detection, with calibrated or knowledge-guided formulations.