Figure 1. Overview of LINCS data-driven therapeutic discovery.
The working principle of ‘signature reversion’-based computational approach. A disease signature representing discordant expression pattern needs first to be identified (G1, G2, and G3 stand for upregulated genes while G4, G5, and G6 stand for down-regulated genes in disease state). With this signature, pharmacologic perturbation data sets can be queried to find compounds with the ability to reverse disease expression pattern (suppress expression of G1, G2, and G3 and induce expression of G4, G5, and G6). After determining the candidate compounds, experimental and clinical validation are required to translate computational findings to clinical applications. LINCS, Library of Integrated Network-based Cellular Signatures.
Figure 1—figure supplement 1. A summary of potential factors influencing the accuracy of signature reversion-based computational approach.
Within the framework of this approach, there are mainly three components: compound signature, disease signature, and signature matching methods. Each component is likely to be influenced by several factors. In addition to the brief descriptions illustrated in this figure, we also discussed these factors in more detail in Supplementary Discussion.
Figure 1—figure supplement 2. An overview of compound-induced expression profiles in LINCS.
(A) The distribution of compound profiles of different perturbation times (upper) and concentrations (lower) across all the compound experiments in LINCS data set. (B) The profile count distribution of all 71 cell lines in LINCS. Each bar represents the number of available compound profiles per cell line. The nine most profiled cell lines were labeled in the figure. (C) Heatmap integrating annotation of the cell lines with perturbation time and concentration. The specific values have not been displayed if there are less than 2000 profiles in the combination of cell line and experimental conditions. LINCS, Library of Integrated Network-based Cellular Signatures.