> I enjoyed following the graph drawing thread a few weeks ago. A serious 
> attempt at "better than graphviz" could be fun and worthwhile.

I was going to reply to that thread as well but didn't get to it.  There is a 
great need for something like this in bioinformatics.  Down the road a bit, we 
will want something like it for the startup business which I am in.  We are 
developing bioinformatics software and one of the things on our to do list is a 
module for graphing protein-protein interactions.

Have a look at this paper:

I painstakingly created figure 1C in Cytoscape.  I was unable to make similar 
graphs, much less the more sophisticated ones I would like to create, for 
subsequent publications because it just took too much time.  Of course I wanted 
to automate the process.  I tried scripting Cytoscape but found that to be too 
tedious and unreliable.  I looked into creating a Cytoscape plugin but it was 
too much work and I would have had to learn Java.  Most of my programming 
experience, so far, has been Perl and R.  I researched a bunch of different 
options and then turned to a GraphViz library for R but found that it was too 
limited.  Finally, I wrote some custom R code to create GraphViz code.  This 
got me the farthest but I would really like to have something nicer than 
GraphViz to work with.

Figure 1C is combining raw gene expression profiling data with Gene Set 
Enrichment Analysis (GSEA) to indicate the activity of various transcription 
factors.  The raw data and GSEA scores are then graphed in the context of the 
most significant transcription factors and their known target genes.  This can 
be a very powerful way of looking at the data under the right circumstances.  
The more common situation is looking at a combination of expression profiling 
data with mutation calls from sequencing in the context of known 
protein-protein interactions.  There are some commercial options which do some 
of this but they are not very good and not really open enough to allow for 
accessible and reproducible research.  Our model is to charge for participation 
in our network and for what will eventually become our platform software.  The 
analytical tools and community contributed analytical modules will be open 
source so you can see exactly what they are doing.  

Moving forward, something which has been an idea for a long time but not yet 
done satisfactorily, as far as I know, is to show proteins as either logic 
gates or analog circuits.  Proteins are viewed, in many cases, as forming 
signal networks.  It would be good to be able to show an animation of how the 
signal flow over the network changes with different environmental signals.  
Alternatively, hold the environmental signals constant and show what happens to 
the signal flow if you add/remove a mutation in one of the proteins.  So that's 
a step or two beyond what GraphViz does but I think it's a logical direction to 
go for further development.  


You received this message because you are subscribed to the Google Groups 
"Racket Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to racket-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to