Fig. 6

Concordance between the sub- and whole dataset differential abundance analysis based on the prostate cancer proteomic data. The FDR threshold was 0.05. The unshaded bar indicates the number of differentially abundant features from the sub-dataset analysis which were also identified by the whole dataset analysis, and the shaded bar indicates the number of differentially abundant features from the sub-dataset analysis which were not identified by the whole dataset analysis. Results were averaged over 100 replicates. Upper panels: sub-sampling 10% of the data; lower pannels: sub-sampling 20% of the data. Left panels: all features were considered; Right panels: only non-normal features (Shapiro-Wilk test p-value <0.01 for at least one of the two groups) were considered