Fig. 8

Model prediction sensitivity analysis with respect to missing genes. 20 random sets of N genes were selected from transcriptomic datasets (N = 8, 12, 16, 20, 24, 28). 85 wound biopsies were tested. The percentage of correctly predicted points was calculated for each set of genes. The resulting means and standard deviations are plotted. The magenta bar represents the original prediction percentage in 69 test points (shown in Fig. 6.) based on all available genes