Extended Data Fig. 5: Comparison of classical binned vs novel continuous analysis approach. | Nature Metabolism

Extended Data Fig. 5: Comparison of classical binned vs novel continuous analysis approach.

From: Single-cell spatial proteomics maps human liver zonation patterns and their vulnerability to disruption in tissue architecture

Extended Data Fig. 5

a) Left: Statistical significance comparison between continuous (LMM-based) and binned (ANOVA-based) analysis. For the binned analysis, protein intensities were divided into 20 equal-width spatial bins and analyzed using one-way ANOVA to test for expression differences between bins. Each dot represents a protein with its respective -log10(Q value). Numbers of proteins significant in both, either, or neither analysis and Spearman correlation coefficient are indicated. Right: Two-dimensional kernel density plot showing the distribution of -log10(Q value) with 20 density levels in the enlarged region (x: −3, 15; y: −8, 15). Individual proteins are shown as white scatter points overlaid on the density plot. Dashed lines indicate significance threshold (q < 0.05). b) Gene set enrichment analysis (GSEA) results from healthy human liver dataset using 3-bin ANOVA analysis (ranked by F statistic) using KEGG pathway database. Normalized enrichment scores (NES) are shown with statistical significance assessed using Benjamini-Hochberg correction for multiple testing (color gradient, q < 0.1). c) Analogue analysis to b using 20-bin ANOVA analysis representing higher-resolution binned approaches. Unless stated otherwise, only proteins with 70% data completeness are included (n = 1741, N = 14).

Back to article page