Default clustering in the demo

[ugurdogrusoz]

Enhance the demo (choose an appropriate one of the demos) by taking an arbitrary flat graph and generating a hierarchically clustered version, where clusters are represented by compound nodes. This should be useful in providing a default mechanism to handle complexity of large graphs. For clustering, we could try using this algorithm or something similar in Cytoscape.js core. For loading an arbitrary graph we could use the import GraphML facility as in this demo.

[ugurdogrusoz]

• [x] 1. When I upload some graphml files, I see some disconnected nodes (corresponding to empty clusters). We should remove those during clustering.

• [x] 2. "markov" -> "Markov"

• [x] 3. Limit Markov clustering depth to be greater than 0

• [x] 4. Let's add a checkbox with title "Apply Markov clustering on load" at the end and apply Markov clustering automatically if this is check during uploading from a file.

• [x] 5. Let's keep the colors in the UI to a mininum.

[ugurdogrusoz]

1 Those disconnected nodes are gone but I see some flickering on load (when apply Markov on load is checked). Try checking this option, then load circle.graphml in the samples. Note that the graph is first rendered in the middle of the canvas and then rendered again to the lower left and moved to the center with animation.

3 I can still make the depth 0 or negative.

[ugurdogrusoz]

On load, there is still flickering as described in previous comment. Let's leave this for later.