Application

Multi-Edged Connections

As Medusa was initially implemented as a

front-end to support STRING and STITCH databases, it currently supports multi-edged connections. Two nodes can be linked with up to 10 different types of connections. The image shows two nodes that are linked with two different types of connections. This is a very important feature that makes Medusa application really competitive because it can be used in many case studies. Thus two bioentities may be coexpreesed or have evolutionary relationships or co-occur in literature...etc

File Menu

1) Open: Browse for file to load the network

2) Add/Subtract: You can load a file containig a graph and add the new information or subtract the information from the loaded view

3) Import: You can load a simple tabbed file containing new connections and update the current view

4) Save: You can save the file and reload it

5) Export: You directly expost images in jpg or png format or postscript in ps and eps formats.

Users can also export in Pajek , Cytosape, Arena3D, BioLayoutExpress and GraphViz format.

Users can also exports the HTML parameters to later load the network as an applet

6) Quit: End the application

Clustering Menu

1) k_means: Perform k-means clustering

2) Affiinity Propagation: Perform Affinity Propagation clustering

3) Markov Clustering: Perform MCL clustering (to be added)

4) Predefined: Perform clustering as given buy the user

5) Spectral: Perform Spectral clustering

6) Clear clustering: Return to the original view

More information about the clustering algorithms and their references can be found in Clustering Algorithms

The results of any clustering method is automatically exported in a text file format.

Manipulation Menu

1) Clear graph: Deleted the current graph

2) Change node shape: Selected node shapes can be changed. Noted can be diamonds, rectangles, triangles or circles.

3) Delete Nodes: Selected nodes can be deleted

4) Crop graph: Edges below a certain threshold are removed

5) Random graph: Generation of a graph with random size and random connections.

6) Rotate: Rotate the panel +90 or -90 degrees

7) Flip: Flip the graph vertically or horizontally.

8) Align Nodes: Select two nodes and align all of the rest according to relevant positions

9) Collapse nodes: If two or more nodes are selected they can get collapsed to one node

10) Expand all nodes: Expand the collapsed nodes

11) Edit graph: A popup window containing the information of the graph appears. the graph can be then manually edited

Directed Connections

Medusa currently supports directed connections. This is an important feature for networks that require such a functionality. Characteristic example are the networks that visualize pathways. Eve though Medusa doe not come with specific layouts to visualize pathways, for small sized of pathways it can give very intuitive visualizations

Layout Menu

1) Random: Random coordinated

2) Circular: circular coordinates

3) Grid: Grid coordinated

4) Relax: Interactively shrink/expand network using force directed layouts

5) Fuchterman:Force-Directed Layouts

6) Hierarchical: Show hierarchies such as gene ontologies

7) Distance Geometry: Bringing nodes closer if they are "more" related. Havel et al.

8) Parallel coordinates: Uising a multi-layered scheme. Lines can be hidden or shown in levels 1 or 2.

More information about the layouts can be found in Layout Algorithms.

 Select Menu

1) All: Select all of the nodes

2) Node: Deselect every selected node

3) By label: Users can find nodes by label. It supports regular expressions

4) By Annotations: Users can find nodes by annotations. It supports regular expressions

5) From file: Users can give a load a file containing annotations and labels and find the nodes instead of finding them one by one

6) Invert: Invert selection and deselect the selected nodes and select the ones that were node selected.

7) Grow selection: More nodes can be added as selected without having to deselect the previous ones.

Display Menu

1) Interactions: We use curves to show multi-edged interactions. If this option is not enabled, the interactions are shown as black straight lines.

2) Labels: Show or hide the annotations

3) Names: Show or hide the names

4) Directed: Show directed edges

5) Confidence: Adjust the color line according to the confidence score

6) Alpha: Adjust the alpha parameter and show lines as continuous dots.

7) Hide Edges on move: Displaying all of the edges together at once makes the network complicated. By enabling this options when a node is dragged, only the edges of this node appear.

8) Color: Medusa supports 3 predefined color schemes. Assuming that a network changes over time. The nodes can be differently highlighted to see which ones change across different experiments

9) Clustering Color: Each cluster gets signed a different color. When nodes are distributed one the plane one can se whether they belong to the same cluster or not.

 Image Menu

Load images to set as background.

Similarly clear the image from the panel

12) Recalculate: This will force recalculation of edge orientation over the whole data set. This is useful when adding edges while relying on the automatic assignment of orientation.