Prediction of Metabolic Gene Biomarkers for Neurodegenerative Disease by an Integrated Network-Based Approach

Figure 3

(a) The −log 10 based value of enrichment of predicted key enzyme-coding genes of PD and HD in DrugBank enzymes for the two diseases, respectively. (b) The flowchart of how to calculate the mean shortest path lengths between predicted enzyme-coding genes and known disease genes from public databases. (c) and (d) The mean shortest path lengths of predicted enzyme-coding genes for PD and HD with disease genes from DrugBank (c) and HGMD (d) (the bars, compared with random), and the corresponding distributions of the mean shortest path lengths of random selected genes (the line plots above the bars). The red lines marked the actual mean shortest path lengths for predicted enzyme-coding genes with known disease genes. The black lines represented the distributions of the mean shortest path lengths of random selected genes with known disease genes.

(a)

(b)

(c)

(d)