Figure - PMC

Skip to main content

An official website of the United States government

Here's how you know

Here's how you know

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

View full-text article in PMC

. 2018 Oct 10;7:e37059. doi: 10.7554/eLife.37059

Search in PMC
Search in PubMed
View in NLM Catalog
Add to search

© 2018, Halu et al

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

PMC Copyright notice

Figure 2. — (A) The entire literature curated PPI network with co-abundance edges from all three stimulation conditions, providing a global view of the distribution and connectivity of co-abundance edges and drug targets (Figure 2—figure supplement 1 for a quantification of related degree distributions and largest connected component (LCC) sizes). A force-directed layout algorithm was used to visualize the networks. Edge colors follow the same convention as Figure 1. (Inset) Toy network depicting the drug target prioritization scheme: For each candidate protein (green node), the shortest path length to each CVD drug target (blue nodes) is calculated and the proximity score PS(c) is calculated (see Materials and methods). Shortest paths between the candidate and CVD drug targets are denoted with the thicker edges and may consist of both PPI and co-abundance edges. (B) The ROC curves for iCTNet CVD drug targets on PPI network only and PPI network with added co-abundance edges, with AUROCs shown in the legend. (C) Violin plots showing the distribution of percentile ranks of iCTNet CVD drug targets on PPI network only and PPI network with added co-abundance edges. Dashed lines indicate the 2nd quartiles (medians) and dotted lines indicate the 1st and 3rd quartiles. (D) AUROC value comparison between PPI and co-abundance edges (indicated by triangles) and PPI and the same number of randomly added edges as the co-abundance networks (circles with error bars). Randomization was repeated for 20 realizations using degree-preserving randomization (see Materials and methods). All empirical p-values between co-abundance and random case are less than 0.05. The AUROC of the PPI network is indicated by the grey dashed line. (E) ROC curves and prediction percentile rank violin plots for external drug target databases: DrugCentral, TCRD Clinical and TCRD Chemical. Violin plots show the distribution of percentile ranks of CVD drug targets from these databases for each case. Dashed lines indicate the 2nd quartiles (medians) and dotted lines indicate the 1st and 3rd quartiles. (F) ROC curves and prediction percentile rank violin plots for inflammation (Inflammatome), innate immune response (InnateDB) and coronary artery disease (CADGene) signatures. Violin plots show the distribution of percentile ranks of proteins implicated in these datasets for each case. Dashed lines indicate the 2nd quartiles (medians) and dotted lines indicate the 1st and 3rd quartiles.

Figure 2—figure supplement 1. — (A) The entire literature curated PPI network with co-abundance edges from all three stimulation conditions, providing a global view of the distribution and connectivity of co-abundance edges and drug targets (Figure 2—figure supplement 1 for a quantification of related degree distributions and largest connected component (LCC) sizes). A force-directed layout algorithm was used to visualize the networks. Edge colors follow the same convention as Figure 1. (Inset) Toy network depicting the drug target prioritization scheme: For each candidate protein (green node), the shortest path length to each CVD drug target (blue nodes) is calculated and the proximity score PS(c) is calculated (see Materials and methods). Shortest paths between the candidate and CVD drug targets are denoted with the thicker edges and may consist of both PPI and co-abundance edges. (B) The ROC curves for iCTNet CVD drug targets on PPI network only and PPI network with added co-abundance edges, with AUROCs shown in the legend. (C) Violin plots showing the distribution of percentile ranks of iCTNet CVD drug targets on PPI network only and PPI network with added co-abundance edges. Dashed lines indicate the 2nd quartiles (medians) and dotted lines indicate the 1st and 3rd quartiles. (D) AUROC value comparison between PPI and co-abundance edges (indicated by triangles) and PPI and the same number of randomly added edges as the co-abundance networks (circles with error bars). Randomization was repeated for 20 realizations using degree-preserving randomization (see Materials and methods). All empirical p-values between co-abundance and random case are less than 0.05. The AUROC of the PPI network is indicated by the grey dashed line. (E) ROC curves and prediction percentile rank violin plots for external drug target databases: DrugCentral, TCRD Clinical and TCRD Chemical. Violin plots show the distribution of percentile ranks of CVD drug targets from these databases for each case. Dashed lines indicate the 2nd quartiles (medians) and dotted lines indicate the 1st and 3rd quartiles. (F) ROC curves and prediction percentile rank violin plots for inflammation (Inflammatome), innate immune response (InnateDB) and coronary artery disease (CADGene) signatures. Violin plots show the distribution of percentile ranks of proteins implicated in these datasets for each case. Dashed lines indicate the 2nd quartiles (medians) and dotted lines indicate the 1st and 3rd quartiles.