NetConfer allows comparison of multiple interaction networks using various methodologies organized as ‘analysis workflows’. The uploaded networks are automatically clustered based on their overall similarity and presented to the user for ease of selection. Each workflow can then be applied over a user selected set of networks for further analysis. Various workflows focus on comparative analysis of nodes, edges, graph properties, shortest paths, community structures and network cliques. Please refer to the documentation and tutorial section for more details.
A brief description of the individual workflows is presented below.
This workflow identifies nodes/edges which are commonly shared or exclusive to one or more of the selected set of networks. It uses a powerful visualization technique called ‘upset plot’ augmented with interactive user operations to graphically display the similarities and differences.
This workflow allows identification and comparison of key nodes between selected networks using local graph property measures. For an ordered selection, a ‘delta centrality’ calculation is made to identify discriminating nodes and displayed using parallel coordinates.
This workflow allows performing a comparative analysis of shortest paths between a given pair of nodes across a selected set of networks. The output is presented graphically for identification of critical nodes in information flow.
This workflow is designed to identify and capture changes in community structure (densely connected group of nodes) across the selection. Communities are inferred using Newman and Louvain algorithms. A ‘sankey plot’ is used to represent changes in communities between a set of selected networks.
This workflow is designed to find and compare ‘cliques’ (closely knit subset of nodes in a graph) between networks. Cliques can be useful in several biological systems like identification of co-expressed genes, finding (and comparing) motifs or analyzing microbial communities.
NetConfer provides a variety of results in both textual and graphical format. Following is a brief overview of some of the key visualizations generated by NetConfer
Global Properties Across Networks
Each job submitted to NetConfer starts with generation of a preview of how global properties of a network compares against other networks submitted in the job. This preview, which is in the form of interactive and customizabel grouped bar plots, provides preliminary cues towards similarities/ differences across submitted networks
.
Structural similarities across networks
Jobs (consisting of multiple networks in each submission) are tagged to personalized 'Workflows dashboard'. The workflows dashboard provides end-users the utility of selecting (all or a subset of) submitted networks by referring an 'edge-jaccard' based similarity tree of networks, the nodes of which are constituted by interactive checkboxes for the networks. The networks so selected can be re-directed for intended analytical workflows.
Compositional (Node and Edge) Similarities
Highly interactive and explorable network diagrams connected to Upset plots, representing the common set of edges and nodes across the selected set of networks have been provided in NetConfer.
Comparing changes in Properties and Centralities (ΔCentrality)
Changes in global properties and centrality measures of various common nodes across the networks can be easilty probed using the interactive tables and D3's parallel co-ordinates based multi-panel hive plots.
Tracing Shortest Paths between nodes
Simplistic yet highly informative Path Maps representing shortest paths between a given pair of nodes across all the networks submitted in the workflow. Such maps can be highly useful for indentifying nodes that are consistently between a given pair of nodes in various networks selected by the user. Particularly useful for gene, metabolic and microbiome networks.
Exploring Communities and Cliques across networks
D3 based sankey diagrams consisting of sorted set of communities on each axis have been creatively employed to visualize community structures and transitions across networks. The node to node connections across different neworks help track transitions in community compositions between a given pair of networks. Cliques can also be explored using interactive, searchable tables.
Browser | Tested | Version recommended |
---|---|---|
Firefox | Yes | Latest |
Chrome | Yes | Latest |
Safari | Yes | Latest |
Feel free to reach out to us