1. Any graph-theoretic algorithm can be coded in R using the "graph"
package as a basis.

2. I don't know  whether graphviz is capable of doing the advanced
visualizations that pigale is, since some things are "unpublished".

3. "graph" itself is "generic" and "non-biological", as are many of the
supporting tools in the Bioconductor repository.

4. Even better, if you're willing to do a bit of heavy lifting, R has
the capability of calling arbitrary C/C++ code, sending and receiving
information from it, etc., and such code could be packaged to work as a
library within R! This means that pigale could be made into an R
extension package.

As you've probably guessed, I'm a shameless R bigot. I don't know the
folks who built pigale -- I went to the web site and saw the feature set
you posted. It's all way over my head; my usage of graph theory is
strictly limited to the structures that arise in computer performance
analysis -- stochastic Petri nets and their reachability graphs, for
example.

I haven't found many GPL packages for the *advanced* performance
analysis domain. PEPA is the best I've uncovered. If all you care about
is steady-state queuing models, Neil Gunther's "PDQ" is great, but it
can't do the really advanced stuff. I'm not even sure PEPA can do
transient analysis; for that, I think you have to roll your own. There
are quite a few good packages in the academic arena that are free to
academic researchers, but they all have steep price tags for commercial
users. I'm working on a GPL R package to do what these tools do, and it
will use "graph" and "graphviz" and a number of other existing R
packages. And yes, I will submit it to Portage as an ebuild. :)

David Grant wrote:

>Can graph and graphviz do all this? in a generic non-biological way?
>  
>
>
-- 
[email protected] mailing list

Reply via email to