Interface Tabs | Description |
---|---|
Load data and select algorithms | The load data and selection of algorithms panel allows users to load custom data or example datasets in matrix format. Users need to upload the matrices binary drug-target bipartite network, drug –drug similarity and protein – protein similarity along with algorithm and parameters of choice. |
Network Properties | The network properties several different properties of the bipartite graph such as the degree centrality of two types of nodes, density of the network, betweenness of two types of nodes, total number of interactions, count of each type of nodes. |
Network Modules | Bipartite network modules are computed using the lpbrim algorithm [49] the tab shows the bipartite nodes as tables and module network. Modules are dynamically updated based on input data. |
Prediction Results | For a given dataset to compute the results one can select any one of the algorithms. The results are shown using the jquery Data Tables library. The table shows the drugs, targets, pvalues, outcome (True/predicted interaction). |
Network Plot | The network plot tab plots the computed predicted network. It uses visNetwork package which uses the vis.js library to generate network. The predicted interactions are marked in dashed lines and true interactions are marked with bold lines. Drop downs are provided to select specific nodes and groups. |
Statistical Testing | The statistical testing tab tests the performance of the model based one the random removal of true links from the network based on the frequency of the drug-target associations. It measures the auac, auc, auctop(10%), bedroc and enrichment of links. Based on these scores we select which algorithm to use. |
Permutation Testing | In permutation testing significance of the associations are calculated by, randomly permuting the matrices and and compute the significance using, standard normal distribution. |
Search Drugbank | This tab allows users to search predicted drug target associations from the drugbank database from 5970 drugs and 3797 proteins from a total of 316645 predicted and 14167 true interactions. |
Ontology and Pathway Search | This tab allows users to search for enrichment of Ontologies and Pathways using a given set of genes. |