Fig. 3

Hierarchical clustering of the samples of the discovery cohort based on the abundance levels of the 50 most significantly differentially expressed proteins resulting from the 3 pairwise comparisons. The samples are represented on the x-axis, FTLD-tau samples in yellow, FTLD-TDP in purple, and NHC in green. Normalized protein abundance levels are represented on the y-axis. Shades of red represent the upregulated proteins, and shades of blue represent the downregulated proteins. p-values for every 3 pairwise comparisons are indicated to the left of the figure. Shades of blue indicate statistically significant values: the darker, the more significant, and grey indicates non-significant p-values. To check the robustness in a data set with a limited sample size, we have repeated the clustering using Spearman correlation (Fig. S18). No significant differences could be observed