I’ve been working with Christian Rödelsperger on some new methods for asking specific question about networks, and using the *C. elegans* nervous system data as a test case. I just wanted to share an early version of the kind of network graph that will be resulting from the analysis. Pretty cool! The center node is what the analysis focuses on, above it is all of the information flowing into it, and below it all of the information flowing out of it. The graph was made using Graphviz and then modified using Adobe Illustrator. For a larger version, look HERE.

### Like this:

Like Loading...

*Related*

I just watched your talk at CSHL, and rather enjoyed it. It’s great to see more neuroscientists approaching their subject from a mathematical (and specifically graph theoretic) perspective. I’d be interested to learn more about your methods, in particular the partitioned network flow analysis, and the comparative analysis. I’m curious exactly how you defined the centrality measure in the partitioned case. I also saw a paper recently (http://www.nature.com/msb/journal/v8/n1/full/msb201199.html) interested me in learning more about rigorous techniques for comparative network analysis.

I’ve explored the latter topic a bit, but in structural analysis rather than (putative) functional analysis. Years ago, I did a graph theoretic analysis of the (then nearly complete) nervous system of C. elegans. I worked out how to count instances of network motifs efficiently, as well as the statistics (including how to approximate count distributions over meaningful random ensembles) to detect enrichment or depletion for different network motifs. We walked away with some nice and interesting structural results, but only vague (and perhaps half-hearted) functional interpretations. These days, I’m more likely to be working with cellular regulatory networks, but the beautiful thing about an abstract mathematical approach is how well it carries over from one setting to another.

I also liked your visualization method, and think it would be particularly effective if it were made interactive. If you decide to make it interactive (for browsing your data yourself, or if you want to share some of your work online for others to see, for example), I recommend checking out the web (javascript) based interactive visualization library called d3 (http://mbostock.github.com/d3/). I’ve been looking forward to using it and have read through most of the code, though I haven’t had the chance to actually use it yet. Still, I think that should be a great library to consider and would work well with your visualization style. The main website has all sorts of interesting examples for dynamic graph / tree layout, expanding and collapsing nodes, virtually arbitrary input methods, etc. And it looks like it would be easy to build on.

I hope to hear from you; I assume you can see the email addresses bound to comments (otherwise, you can just respond to this comment and we’ll work out how to shift to a different communication band). I looked around for any publications, preprints, manuscripts, etc that you might have written on this material, but didn’t find anything directly related to your talk.

Separately, if you could use another set of eyes on any of the graph theoretic analysis you’re working on, I’d be interested to work on that area again.

Best,

Eric