CentiScaPe is a Cytoscape app that allows, by using a very plain interface, to compute different network centrality parameters in single and in multiple networks at the same time. It allows to plot the results by using real and normalized values in different plots. Several centralities are available: Degree, Average Shortest Path, Eccentricity, Closeness, Betweenness, Centroid, Stress, Radiality, Eigenvector, Edge Betweenness, Bridging Centrality. Each one could be computed for directed and undirected networks; CentiScaPe gives the possibility to assign a weight to the edges and, currently, we added a new feature that permits to assign a weight to each node by using an attribute.
Goal
The new version permits an in-depth analysis of biological networks, from the topological point of view. The main issue with this methodology is that it requires a lot of literature mining in order to validate and verify the results that, in general, remain speculative. In order to overcome this limitation we decided to add a feature that allows to validate the results by using random networks, i.e. by adding a sort of statistical evaluation. By randomizing a network we expect to highlight all the differences between the real and the random process[1]: this step gives strength to all the findings that come from a topological analysis and adds a statistical background that could integrate the biological information from experimental data or text mining.
We will integrate CentiScaPe with some algorithms that were already developed in an existing, but no more supported, app for the Cytoscape platform, version 2.6, that is RandomNetworks [2][3]. We will develop some algorithms that are able to generate networks by following a specific model: Watt-Strogatz, Erdos-Renyi, Barabasi-Albert and some others. Another interesting feature will be the degree-preserving algorithm that is able to randomize the edges by keeping the degree of each node fixed. Finally the comparison between the centralities before and after the randomization could be analyzed by using the topological measures of CentiScaPe.
Background
CentiScaPe is a Cytoscape app that allows, by using a very plain interface, to compute different network centrality parameters in single and in multiple networks at the same time. It allows to plot the results by using real and normalized values in different plots. Several centralities are available: Degree, Average Shortest Path, Eccentricity, Closeness, Betweenness, Centroid, Stress, Radiality, Eigenvector, Edge Betweenness, Bridging Centrality. Each one could be computed for directed and undirected networks; CentiScaPe gives the possibility to assign a weight to the edges and, currently, we added a new feature that permits to assign a weight to each node by using an attribute.
Goal
The new version permits an in-depth analysis of biological networks, from the topological point of view. The main issue with this methodology is that it requires a lot of literature mining in order to validate and verify the results that, in general, remain speculative. In order to overcome this limitation we decided to add a feature that allows to validate the results by using random networks, i.e. by adding a sort of statistical evaluation. By randomizing a network we expect to highlight all the differences between the real and the random process[1]: this step gives strength to all the findings that come from a topological analysis and adds a statistical background that could integrate the biological information from experimental data or text mining.
We will integrate CentiScaPe with some algorithms that were already developed in an existing, but no more supported, app for the Cytoscape platform, version 2.6, that is RandomNetworks [2][3]. We will develop some algorithms that are able to generate networks by following a specific model: Watt-Strogatz, Erdos-Renyi, Barabasi-Albert and some others. Another interesting feature will be the degree-preserving algorithm that is able to randomize the edges by keeping the degree of each node fixed. Finally the comparison between the centralities before and after the randomization could be analyzed by using the topological measures of CentiScaPe.
[1] Network Motifs: Simple Building Blocks of Complex Networks. Milo et al, Science 25 October 2002
[2] RandomNetworks App
[3] RandomNetworks Project Website
Technology and Skills
Cytoscape, CentiScaPe, Java
Potential Mentors
Giovanni Scardoni (CBMC, University of Verona)
Contact
cytoscape-discuss@googlegroups.com, giovanni.scardoni@gmail.com