Make layout stored in GML file instead of layout.table
Closed this issue · 4 comments
Store the node coordinates in the GML graphlet files instead of in the layout.table file.
Low-priority, but want to keep this in mind as we start storing annotations, so that we don't end up with (1) an FCS file, (2) a GML file, (3) a layout.table file, (4) an annotations file ... when we could just have an FCS and GML.
I see that GML files have the x and y coordinates stored under the "graphics" attribute for vertices. Yet igraph in R seems to ignore them when reading / writing / altering those GML files. As near as I understand, the layout.table is created such that the coordinates for the vertices can be written should the GML ever need to be altered after the fact. Is this true?
I don't understand your question. Again please?
rbruggner reply@reply.github.com wrote:
I see that GML files have the x and y coordinates stored under the "graphics" attribute for vertices. Yet igraph in R seems to ignore them when reading / writing / altering those GML files. As near as I understand, the layout.table is created such that the coordinates for the vertices can be written should the GML ever need to be altered after the fact. Is this true?
Reply to this email directly or view it on GitHub:
#4 (comment)
Just noting that the node coordinates are, in fact, already stored in the GML files as "graphics" attributes. Trying to confirm that the layout.table is only necessary if you plan on modifying the GML files after initial generation because the igraph package doesn't read the complex "graphics" attribute (containing the coordinates) from the generated GML files. Just noting this because I think the desired coordinates are already there but the R igraph-gml parser is inadequate.
Of course, I have no idea if cytospade uses the layout.table file at all.
Oh. Unless this has changed since inception, no changes are made to the
GML files after creation. I assume it's possible to make igraph read
coordinates, but I've never used that package. I haven't looked into
java classes for reading/writing GMLs, but should be easy. The other
issue is that the layout needs to be applied to all GML files, not just
the currently loaded one. There was some benefit there to writing to a
layout.table file, because if the plugin crashes, the layout.table is
slightly less corruptable.
On 10/17/2011 6:07 PM, rbruggner wrote:
Just noting that the node coordinates are, in fact, already stored in the GML files as "graphics" attributes. Trying to confirm that the layout.table is only necessary if you plan on modifying the GML files after initial generation because the igraph package doesn't read the complex "graphics" attribute (containing the coordinates) from the generated GML files. Just noting this because I think the desired coordinates are already there but the R igraph-gml parser is inadequate.
Of course, I have no idea if cytospade uses the layout.table file at all.