Visualizing networks and browsing genomes

17 Sep

The Genome Informatics conference has been taking place on campus this week (Sept 15 – 19, 2010), and staff here are lucky enough to be able to get in to see the posters and attend some of the sessions.

I went along to part of Session 4: Database, Data Mining, Visualization and Curation, because there were a couple of talks I wanted to catch.

The main one was Martin Krzywinski’s talk about a suggested approach to network visualization. The second talk I was interested in was Marc Fiume’s presentation of the Savant genome browser.

My hasty notes and some useful links are below…

Linnet – linear layout for visualization of networks

Martin’s talk discussed the problem of visualizing and interpreting complex networks, and offered a possible solution.

Linnet screenshots

He started off by pointing out that a woolly mammoth hairball could tell us something about the species, but the resulting “visualization” might not be as good as seeing a life-size reconstruction (depending on what interests you). Mitigating the problem of visualizing complex networks is the aim of his work.

There is already work been done on visualizing networks on a 2D surface, and connecting the nodes involved.
Functional aspects of nodes and links are not always taken into account, however.

Throught his talk, Martin referenced Yan et al’s comparison of the E. coli genome and the structure of the Linux operating system [1]. You can read more about that study in Wired, and in more detail at the Discover magazine blog.

The original visualization of these networks contained a lot of overlaps, and Martin wanted to try and find away to tease out these “fibres”, so that the relatioships between certain genes (or functions, in Linux) can be more easily visualised.

He does this by displaying them radially, on three axes, following the original model of showing “master regulator” genes, “middle manager” genes, and “workhorse” genes (again, or functions, in Linux).

Once the network relationships between one gene and another are drawn between axes, the leaf-like images you see in the screenshot above are produced. Beautiful and informative.

He also showed examples of how this approach could be used in the study of synteny, and in the visual assessment of alignments.

A couple of commentators at the talk pointed out that the axes would benefit from being labelled, since it isn’t that easy to remember what is what, as you begin to study it. Martin pointed out that the visualization technique is still being developed, but this is absolutely something he and his colleagues would look at adding in.

Software and documentation will be made available at http://mkweb.bcgsc.ca/linnet/

[1] “Comparing genomes to computer operating systems in terms of the topology and evolution of their regulatory control networks.” By Koon-Kiu Yan, Gang Fang, Nitin Bhardwaj, Roger Alexander, Mark Gerstein. Proceedings of the National Academy of Sciences, Vol. 107 No. 18, May 4, 2010.

Savant genome browser

I also caught Marc Fiume’s talk on the Savant genome browser, being developed at the University of Toronto. Marc wisely opted for a pre-recorded demo of it’s functionality! No one needs “technical problems” in the middle of their talk, first thing in the morning!

Given the kind of work I am usually involved with, genome browsing is not something I do a great deal of, but I was keen to see the interface that Marc and his colleagues have developed.

Savant is a Java-based system, and looks as though it uses some sort of JavaScript framework (like ExtJS) to produce the web-based application. The team involved have studied a number of other genome browsers (though not Ensembl, apparently), and have picked out some of the best functionality of each.

Extra bells and whistles, not often seen in other genome browsers included:
Save sessions, bookmarking (annotations of interesting regions; exportable), lock track as overview track, data selection from a track; split-view (table view + normal view); variety of visualisation modes

There is dynamic generation of output as the user adjusts parameters

It is still being developed, and based on audience questions, there are some features not yet implented that people would really like to see (connection to remote DBs; connection to DAS servers, etc), but I was impressed by the interface, and the thought that had gone into some of the features.

Find out more at http://compbio.cs.toronto.edu/savant/

About these ads

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

%d bloggers like this: