Figure 1
From: Revealing protein networks and gene-drug connectivity in cancer from direct information
Workflow of direct coupling identification on cancer data. The proteomic data is obtained from the TCPA28, while pharmacogenomic data is acquired from CCLE8. The analysis on these data has been divided into 3 steps: (1) Expression and IC50 data is discretized to representative letters to form matrices with samples in rows and proteins, genes mutation statuses/expression levels, and IC50 values in columns, obtaining 3 types of matrices, the protein expression matrix, gene mutation and drug response matrix, and gene expression and drug response matrix. (2) Computation and subsequent descending ranking of DI values by using DCA and the matrices as input for each protein pair and gene-drug pairs. (3) Performance evaluation based on evidence searched from literature and classification of high confident pairs ranked top 100 based on the presence of evidence and strength of interaction.
