[BioC] help with rendering layers in Rgraphviz

Kasper Daniel Hansen khansen at stat.Berkeley.EDU
Wed Jun 13 22:34:48 CEST 2007


On Jun 13, 2007, at 2:12 AM, Mark W Kimpel wrote:

> Kasper,
>
> Thanks for your suggestions. Based on your experience with  
> Graphviz, do
> you think my idea of rendering the edges and nodes in separate  
> layers is
> doable and the right way to go?

I have no experience with layers in Graphviz, but I would assume they  
have some way of handling this problem, which might be pretty common  
- and layers sound like it could be it.

You have a quite large graph, so in any way you might want to look  
for Graphviz based tools in order to explore it (see the graphviz  
homepage under tools I think)  - or is the R plotting looking  
reasonable? For other reasons I am myself looking at displaying a  
20.000 node bipartite graph (or rather will be soon), so if anyone  
has any good software suggestions, I am open as well.

Having said all of that, what you are asking is a reasonable request,  
which should be incorporated into Rgraphviz. However, I certainly do  
not have time to work on that right now :(

Kasper

> Mark
>
> ---
>
> Mark W. Kimpel MD  ** Neuroinformatics ** Dept. of Psychiatry
> Indiana University School of Medicine
>
> 15032 Hunter Court, Westfield, IN  46074
>
> (317) 490-5129 Work, & Mobile & VoiceMail
> (317) 663-0513 Home (no voice mail please)
>
> ******************************************************************
>
> Kasper Daniel Hansen wrote:
>> On Jun 12, 2007, at 11:20 AM, Mark W Kimpel wrote:
>>
>>> I am using Rgraphviz in "neato" mode to layout a graph with about  
>>> 500
>>> nodes and 1500 edges. Rgraphviz first renders the nodes and then the
>>> edges, meaning that I have a lot of edges overlapping with nodes. I
>>> would like to put the edges behind the nodes to avoid this.
>>>
>>> In reviewing documentation of both Graphviz and Rgraphviz, I do not
>>> think using the "no overlap" option would work with a graph of this
>>> size. Being able to see each and every edge is also not important  
>>> for my
>>> application, so I am wondering if I can use layers to render the  
>>> edges
>>> in a layer behind the nodes.
>>>
>>> Will this work? If not, any other suggestions? If it will, could an
>>> expert supply me with some sample code to render edges and nodes in
>>> separate layers? I could not find any in the documentation I  
>>> reviewed.
>>
>> In the code for Rgraphviz we just use Graphviz to layout the  
>> graph, ie.
>> we get the (x,y) positions of the nodes and the function forms of the
>> lines. Having those (and some other parameters we also need) we  
>> render
>> (draw) the graph using standard R functions. So I doubt that Graphviz
>> has any option that will affect this from the point of view of
>> Rgraphviz. You could export the graph in dot format and then use the
>> command line version of Graphviz to actually render (draw) the graph.
>> \
>> Or you could try and hack the R code as you please. That should  
>> not be
>> terrible hard in fact. You can get the relevant code by
>> R> getMethod("plot", "Ragraph")
>>
>> (Ragraph is the name of the object after we retrieve it from Graphviz
>> layout routine).
>>
>> Kasper
>>
>>
>>> Thanks,
>>> Mark
>>> -- 
>>>
>>> ---
>>>
>>> Mark W. Kimpel MD  ** Neuroinformatics ** Dept. of Psychiatry
>>> Indiana University School of Medicine
>>>
>>> 15032 Hunter Court, Westfield, IN  46074
>>>
>>> (317) 490-5129 Work, & Mobile & VoiceMail
>>> (317) 663-0513 Home (no voice mail please)
>>>
>>> _______________________________________________
>>> Bioconductor mailing list
>>> Bioconductor at stat.math.ethz.ch
>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>> Search the archives:
>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>
>>
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/ 
> gmane.science.biology.informatics.conductor



More information about the Bioconductor mailing list