PathwayConversion
From PathVisio Wiki
To test the current gpml, claim your pathways (~3) here:
Contents |
Kristina
- Hs_Krebs-TCA_Cycle
- Hs_Fatty_Acid_Synthesis_BiGCaT
- Hs_Wnt_Signaling
Problems
General
- I consistently get an error (StackOverflowError) on creation of the second group. Doesn't matter which group of genes I select, the first group always works, and the second produces the error, causing PathVisio to become unresponsive. After a few seconds, I get regular print statements again (from Alex's code: "TRUE INTERSECTS!"), suggesting that it is now responding. If I try to click on something, the same error appears again.
Alex
- Hs_G_Protein_Signaling
- Hs_Statin_Pathway_PharmGKB
- Hs_Notch_signaling_KEGG
Thomas
- Hs_Glycolysis_and_Gluconeogenesis (sticky edges done)
- Hs_Complement_Activation_Classical
- Hs_Cytokines_and_Inflammatory_Response_Biocarta
Problems
Hs_Glycolysis_and_Gluconeogenesis
- A metabolic reaction typically consists of two labels, an arrow in between them and one or more gene-boxes on the arrow. We currently link the arrow ends to the labels, but we should also group the gene-boxes with the arrow. Should we make a seperate group type (e.g. reactionParticipants) for this?
- We should support labels with carriage returns. The workaround (create two labels) gives problems with sticky edges (to which label should be linked to?)
- This pathway contains a couple of lines with arrows on both ends. The current solution is to create two lines, but this is confusing and hard to convert to a single edge. We should support arrows on both start and end
Poly lines and branched lines
The current approach for poly and branched lines is not ideal. For every line segment we use a seperate <line> element and the end points are linked together in the view when they have excactly the same coördinates. The advantage of this is that many pathways that are converted from GenMAPP automatically have the polylines (because the endpoints of seperate lines have the same coördinates). However, polylines represent a single edge, so it makes more sense to put them in a single <line> element with >2 points. This makes programming and conversion to e.g. Cytoscape easier too. However, branched lines will give a problem with this approach (and also with translating to edges). I'm not sure if we should suppport them.
For some reason I also got this error while messing with polylines:
Error: While writing xml to C:\thomas\pathvisio-data\pathways\Hs\Hs_Contributed_20060824\metabolic_process-GenMAPP\Hs_Glycolysis_and_Gluconeogenesis_plus.gpml org.pathvisio.model.ConverterException: class org.jdom.JDOMException: Exception in endElement: cvc-id.1: There is no ID/IDREF binding for IDREF 'f8c'.: class org.jdom.JDOMException: Exception in endElement: cvc-id.1: There is no ID/IDREF binding for IDREF 'f8c'. org.pathvisio.model.ConverterException: class org.jdom.JDOMException: Exception in endElement: cvc-id.1: There is no ID/IDREF binding for IDREF 'f8c'.
