This plugin aims to ease the map exploration by combining all data sources that were used to construct the AIR. It allows the user to search for elements, find regulators and targets, predict key regulators of phenotypes, and analyze network topology features. This way, new hypotheses can be generated data-independently by using the AIR as an information resource.
The plugin is divided into three individual parts: Exploration of interactions, prediction of key regulators, and visualization of network centrality features.
To get information on a specific element the user can either write its name case-independent into (1), select an element on the map, or click a data point on the plugin graphs (e.g. in (4)). After the selection, the element's type will be displayed below together with a link, if available, that, if clicked, opens up an interactive 3D visualization of the elements molecular structure in a popup window.
Four different tabs contain all information on the selected element's properties and interactions (2).
Regulations: List of all molecules that directly regulate the selected element, their molecule type, the interaction type, and PubMed references. Furthermore, the user can filter the regulators by their type, e.g. to show only microRNAs.
Targets: List of all molecules that are directly regulated by the selected element, their molecule type, the interaction type, and PubMed references.
Phenotypes: List of all phenotypes that are, directly or indirectly, regulated by the selected element, the type of regulation, and the shortest path distance between them.
Sequence: If the selected element is a protein with a valid UniProt ID, this tab will display an interactive panel (by ProtVista) to visualize properties of the protein's sequence, such as amino acid modifications or genetic variants.
2. Regulator Prediction
This function allows the user to identify key regulators of specific phenotypes. By selecting a pseudo FoldChange value for each phenotype (3), the ability of every molecule in the map to influence those phenotypes will be analyzed.
Predicted targets will be ranked by their sensitivity (ability to regulate the selected phenotypes in the selected way) and specificity (inability to regulate not selected phenotypes).
The graph (4) displays the results in a scatter plot and colors the regulators based on their type of regulation: positive, if the element induces the selected phenotypic changes, and negative if it induces the exact opposite changes. Elements marked as 'external' are not included on the maps and will link to their entry on public databases.
To analyze phenotype-specific centralities, a subnetwork was created for each phenotype, that contains the manually curated information from the phenotype-containing submaps merged with regulatory layers and feedback loops. Calculated centrality values are:
Betweenness: The ratio of shortest paths between one or multiple pairs of elements that go through the element of interest. Hub nodes, such as transcription factors, have a high betweenness centrality.
Closeness: Describes the distance of the element to one or multiple other elements (in this case the phenotypes). Here, the closeness is defined as the reciprocal shortest path distance between the element and the phenotype.
In-Degree: Number of regulators the element has.
Out-Degree: Number of targets that are regulated by the element.
Degree: Sum of In- and Out-Degree
The first part of the Centrality panel is a table (6) that shows the centrality values of all molecules for the phenotype selected in (5).
Furthermore, the user can visualize different centrality values for different phenotypes, e.g. to identify overlapping key regulators. In (7) the values and phenotypes will be selected for the x- and y-axis independently, which are then displayed in the graph (8).