Fig. 1

Illustration of the method for cutoff point identification in descriptive high-throughput biological studies. Variable distribution (A, D, F) and biological categories enriched in shortlists identified using cutoff points (B, E, G) for the following datasets: genes expressed in the human cerebral cortex (A, B), genes sensitive to chemical exposures (D, E), and proteins expressed in the adult human heart (F, G). Figure C illustrates changes in the number of shortlisted genes identified by the described cutoff algorithm in relation to the number of genes in the dataset. Number of shortlisted genes is shown as percent of the total shortlisted genes identified for a complete dataset (16,353 genes). In graphs (A, D, F), A is a curve of the original data distribution, B is a linear shortcut connecting the first and the last points of A, and C is a family of linear functions perpendicular to B. Four C functions are shown in figure A. In figures C and D longest segments corresponding C functions are shown. Red vertical lines in figures A, D, F correspond to the cutoff points