Question

RCytoscape performance

0

Entering edit mode

Chad Burrus ▴ 20

@chad-burrus-4348

Last seen 10.2 years ago

Does anyone have any experience with the RCytoscape package and highly connected graphs?? I've got a graph being built dynamically with R and then pushed over to Cytoscape via the RCytoscape plugin.? However, the communication between R and Cytoscape is taking an incredibly long time--3-4 days for a graph with 362 nodes and 65703 edges, to be precise.? I'm still working on figuring out exactly which of the commands is taking so long (or if it's a combination of all of them, as I'm coming to suspect), but I've reproduced the code I'm using to interact with RCytoscape below. It's basically just the tutorial code, though (from http://db.systemsbiology.net:8080/cytoscape/RCytoscape/vignette/RCytos cape.html). Some context: I'm running R 2.11.1, Cytoscape 2.7.0, and CytoscapeRPC 1.2 on a Linux box with a manual install of the RCytoscape and XMLRPC packages as described at http://db.systemsbiology.net:8080/cytoscape/RCytoscape/installation/in dex.html. I can get more information if you need it. Thanks in advance for any help. ------------------------------------------ Chad Burrus MS Bioinformatics, Candidate School of Informatics and Computing Indiana University, Bloomington Phone: (812) 783-9127 ********************************** R Code ********************************** # graph is a graphAM object that's fully constructed except for the attributes added below graph = initNodeAttribute(graph, "sourceDataset", "char", "X") nodeData(graph, xNames, "sourceDataset") = "X" nodeData(graph, yNames, "sourceDataset") = "Y" graph = initEdgeAttribute(graph, "dot", "numeric", 0) # add the edgeData at the last minute to improve performance edgeData(graph, from, to, "dot") = dotData # from here on, things are really slow cw = CytoscapeWindow(paste(prefix, "Network"), graph) displayGraph(cw) redraw(cw) ********************************** End R Code

graph RCytoscape graph RCytoscape • 1.4k views

ADD COMMENT • link updated 14.0 years ago by Paul Shannon ★ 1.1k • written 14.0 years ago by Chad Burrus ▴ 20

score 0 · Answer 1 · 2010-11-12

0

Entering edit mode

Paul Shannon ★ 1.1k

@paul-shannon-578

Last seen 10.2 years ago

Hi Chad, This should be vastly improved after changes I checked into the release branch yesterday. My code for sending edges to Cytoscape was extraordinarily inefficient. I can send you RCytoscape 1.0.1 offline if that would be any help. - Paul > Does anyone have any experience with the RCytoscape package and highly > connected graphs?? I've got a graph being built dynamically with R and > then pushed over to Cytoscape via the RCytoscape plugin.? However, the > communication between R and Cytoscape is taking an incredibly long > time--3-4 days for a graph with 362 nodes and 65703 edges, to be > precise.? I'm still working on figuring out exactly which of the > commands is taking so long (or if it's a combination of all of them, > as I'm coming to suspect), but I've reproduced the code I'm using to > interact with RCytoscape below. It's basically just the tutorial > code, though (from > http://db.systemsbiology.net:8080/cytoscape/RCytoscape/vignette/RCyt oscape.html). > > Some context: I'm running R 2.11.1, Cytoscape 2.7.0, and CytoscapeRPC > 1.2 on a Linux box with a manual install of the RCytoscape and XMLRPC > packages as described at > http://db.systemsbiology.net:8080/cytoscape/RCytoscape/installation/ index.html. > I can get more information if you need it. > > Thanks in advance for any help. > > ------------------------------------------ > Chad Burrus > MS Bioinformatics, Candidate > School of Informatics and Computing > Indiana University, Bloomington > Phone: (812) 783-9127 > > > *

ADD COMMENT • link 14.0 years ago Paul Shannon ★ 1.1k

0

Entering edit mode

That sounds great. I've poked around with the source a bit, so I noticed a couple of those efficiencies, but I wasn't having much luck making the changes myself. You can send the updated version to chburrus@indiana.edu or just give me a link to download. Thanks for the help. ------------------------------------------ Chad Burrus MS Bioinformatics, Candidate School of Informatics and Computing Indiana University, Bloomington Phone: (812) 783-9127 On Fri, Nov 12, 2010 at 10:24 AM, Paul Shannon <pshannon@systemsbiology.org>wrote: > Hi Chad, > > This should be vastly improved after changes I checked into the release > branch yesterday. My code for sending edges to Cytoscape was > extraordinarily inefficient. > > I can send you RCytoscape 1.0.1 offline if that would be any help. > > - Paul > > > > Does anyone have any experience with the RCytoscape package and highly > > connected graphs?? I've got a graph being built dynamically with R and > > then pushed over to Cytoscape via the RCytoscape plugin.? However, the > > communication between R and Cytoscape is taking an incredibly long > > time--3-4 days for a graph with 362 nodes and 65703 edges, to be > > precise.? I'm still working on figuring out exactly which of the > > commands is taking so long (or if it's a combination of all of them, > > as I'm coming to suspect), but I've reproduced the code I'm using to > > interact with RCytoscape below. It's basically just the tutorial > > code, though (from > > > http://db.systemsbiology.net:8080/cytoscape/RCytoscape/vignette/RCyt oscape.html > ). > > > > Some context: I'm running R 2.11.1, Cytoscape 2.7.0, and CytoscapeRPC > > 1.2 on a Linux box with a manual install of the RCytoscape and XMLRPC > > packages as described at > > > http://db.systemsbiology.net:8080/cytoscape/RCytoscape/installation/ index.html > . > > I can get more information if you need it. > > > > Thanks in advance for any help. > > > > ------------------------------------------ > > Chad Burrus > > MS Bioinformatics, Candidate > > School of Informatics and Computing > > Indiana University, Bloomington > > Phone: (812) 783-9127 > > > > > > * > [[alternative HTML version deleted]]

ADD REPLY • link 14.0 years ago Chad Burrus ▴ 20