Talk:Working Group 2 - Expression visualization and clustering

Online visualisation of large datasets

Here is the method that I have previously used for visualising a dataset that was too large to import directly into Biolayout: 1. Define "clusters" of co-expressed CTSSs using whatever clsutering method is chosen. 2. Calculate average expression profile for each co-expressed cluster (CEC - please forgive the clunky nomenclature!). Each CEC then becomes a node in a much smaller network of CECs. 3. Edges are drawn between similar CECs by calculating the Pearson correlation between the average expression profiles (other methods for determining similarity could easily be used here). 4. The CEC network (ie the summary graph) is imported into biolayout the storage file (.layout) is edited so that the size of each node is proportional to the number of CTSSs in that CEC. This graph should be light and quick to load, even on relatively ordinary PCs.

I'll upload an example when I work out how. Tom has already implemented a feature in Biolayout that does a websearch when a user double-clicks on a node, so it should be easy to change this so that a new window opens containing the indivudual CTSS data when a user double-clicks on a CEC node. That way, each user only loads the data that he/she is interested in, rather than having to hold the entire network in RAM. Kenny 09:22, 28 February 2011 (UTC)