Bipartite network properties are calculated by transforming the network in to one-mode networks (contain one set of nodes) called projection of the network in which a bipartite network of drugs and proteins two drugs are connected if they share a single protein similarly two proteins are connected if they share a single drug molecule. Using the two-projected network of drugs and proteins we compute degree centrality, betweenness, total number of interactions, total number of each of the nodes and distribution of the drug and target nodes shown in Figure 2. WE have implemented the visualization of cousts and betweenness histograms using the rCharts R package \cite{rchart}. Bipartite network modules are computed using the lpbrim algorithm \cite{bib27} for which lpbrim R package is used\cite{lpbrim2}. The algorithm consists of two stages. First, during the LP phase, neighboring nodes (i.e. those which share links) exchange their labels representing the community they belong to, with each node receiving the most common label amongst its neighbors. The process is iterated until densely connected groups of nodes reach a consensus of what is the most representative label, as indicated by the fact that the modularity is not increased by additional exchanges. Second, the BRIM algorithm (2) refines the partitions found with label propagation. HeatS and network based inference compute (NBI) recommendations using a bipartite graph , where a two phase resource transfer Information from set of nodes in A gets distributed to B set of nodes and then again goes back to resource A . This process allows us to define a technique for the calculation of the weight matrix W. HeatS uses only the drug target bipartite data matrix and NBI uses similarity matrices of drug chemical similarity matrix and protein similarity matrix. The random walk with restart(RWR) algorithm uses all the three different matrices to compute the recommendations. Netcombo computes both NBI and RWR and then averages the scores. The prediction results tab shows the computed results using the javaScript library DataTables\cite{bib28}. The data table provides columns filters and search options. The network plot tab represent the network using the visNetwork R package \cite{bib29} The Network visualization is made using vis.js javascript library. Javascript libraries can be integrated using a binding between R and javascript data visualization libraries Fig 3.
Search Drugbank tab
The drugbank tab Fig 4 helps to search predicted interactions computed using NBI method using the drugbank database \cite{bib32}. One can search for targets given a specific drugbank ID and search for drugs given a specific hugo gene name \cite{hugo}.In fig3 the data table shows the drug target significant scores whether it is a true or predicted interaction, Mesh categories of drugs, ATC Codes and groups (approved, illicit,withdrawn, investigational, experimental). Currently the drugbank search tab only supports data computed using Netowrk based inference . The computed results and the associated meta-data are stored in a sqllite database\cite{sql} for access through shiny data tables interface.