wikipathways / reactome2gpml-converter

Converts Reactome pathways to GPML format

Reactome2GPML Converter

This converter converts Reactome pathways to GPML 2013a format.

The class ReactometoGPML2013.java does the actual conversion and class CLIConverter provides command line access to the converter.

Eclipse is used as Java IDE, and the project is built using ant.

After setting up the project in Eclipse or as a local .jar file, testing can be performed on a local Reactome MySQL database; download at https://reactome.org/download-data.

The conversion is performed by the convertPathway method in class org.reactome.sgml.ReactomeToGPML2013Converter. You will have to provide your correct database connection information for class MySQLAdaptor().

For more information, read the publication on this project titled: "Reactome from a WikiPathways Perspective" by Anwesha Bohler , Guanming Wu, Martina Kutmon, Leontius Adhika Pradhana, Susan L. Coort, Kristina Hanspers, Robin Haw, Alexander R. Pico, Chris T. Evelo. https://doi.org/10.1371/journal.pcbi.1004941

Recent conversions for WikiPathways and news can be found at: https://classic.wikipathways.org/index.php/Portal:Reactome

reactome2gpml-converter's Issues

Overview of pathways doesn't link them

For example, if I compare http://www.wikipathways.org/index.php/Pathway:WP1849 with the original [0] then the arrows seem to be missing.

0.http://www.reactome.org/PathwayBrowser/#DB=gk_current&FOCUS_SPECIES_ID=48887&FOCUS_PATHWAY_ID=556833

Fix log file

• Lists of Rendered and "Not Rendered" after every PW should be removed from output command line.
• Add summary of log in the end

Megapathways not converted

Problems with converting megapathways from Version 53

hardcode some types in the convertor?

Anwesha,

there are a number of ChEBI identifiers used in Reactome pathways which do not get annotated as @type="Metabolite" in the GPML.

What do you think of making a short list of known metabolites, so that when those IDs are identified the are marked as metabolite in the GPML?

That would solve a number of these errors (which are known upstream and haven't been fixed yet, it seems):

https://jenkins.bigcat.unimaas.nl/job/WikiPathways%20Curation%20(Reactome)/lastCompletedBuild/testReport/nl.unimaas.bigcat.wikipathways.curator/Metabolites/ChEBIIDsNotMarkedAsMetabolite/

Error in WP2972

one pathway with serious issues: http://www.wikipathways.org/index.php/Pathway:WP2972

create a working build.xml (Ant) or pom.xml (Maven) to compile the software

Which is needed for CI on our Jenkins server.

Missing child PW for 166520

Reactome PW 166520 (Signaling by NTRKs) was in version 63, but became a meta PW for version 64.
This PW has 3 child PWs; one was already part of pathway collections, 1 is a new pathway, and 3rd (9034015) is not a GPML (nor are the children of this PW). Check if this 3rd pathway becomes part of the 65/66 version, and if not, where this is coming from.

Fix GroupRefs

We noticed one more thing about the new Reactome pathways, the GroupRefs are no longer alphanumeric, but instead look like this:

GroupRef="1416470.4872874213801427"

They used to look like this:

GroupRef="f1da6"

Resize Labels and datanodes for complex components

Longer labels are not always multi-line, resulting in cutoff text. This is not a problem for all longer labels, some are written as multi-line labels as expected, so I don’t know if maybe this has to do with how the information in stored by Reactome. I’m attaching an example. For this particular example, the corresponding Reactome pathway does have a multi-line label.

Remove .svn/, trunk/, branches/, and autogenerated folders

.svn/ folders should not be there, but please also remove folder with autogenerated content, like bin/, and the SVN folders like trunk/ and branches/.

Unable to trace where WP1898 went in version 64

The Reactome PW number for (https://www.wikipathways.org/index.php/Pathway:WP1898)[WP1898] doesn't exist anymore, nor is the title reflected anywhere. Have to check if part of the content is moved to another PW (after uploading to version 65 and 66).

DataNodes do not show up in the PVjs viewer on test2

pathway does not display (Pathway:WP2655)

The pathway http://test2.wikipathways.org/index.php/Pathway:WP2655 does not display in my browser.

Remove unnecessary references

Identical references on interactions and anchor points seem repetitive. For cases with references on one or more interactions leading to an anchor point, the anchor point also lists the same references, which makes the pathways look busier than necessary and doesn’t really add any information. Is it OK to skip the references at the level of the anchor and keep them just for the incoming interactions? See the attached for an example. Note that I dragged the horizontal interaction (with the anchors) up a bit to show the references clearer.

Link to Reactome in Plant pathways is wrong

Output the ChEBI identifiers in their preferred format

Instead of:

<Xref Database="ChEBI" ID="15379" />

Please output the identifiers as:

<Xref Database="ChEBI" ID="CHEBI:15379" />

Please add author, copyright and license information

Problem with pathway WP1880

http://wikipathways.org/index.php/Pathway:WP1880

The protein "SLC26A1,2" has a Reactome ID and seems to be a complex but the complex components, proteins SLC26A1 and SLC26A2 are not listed at the bottom of the pathway where complexes are listed.

(Reported by Egon)

On some pathways, one or more nodes are missing.

It seems to coincide with nodes that have “Reference Entity” listed as “Unkown:Unknown" in Reactome. I don’t recall if this was the case with the human converted pathways also. Could we still add the node, just with an empty xref, but with the same label as Reactome?

Example: http://www.wikipathways.org/index.php/Pathway:WP3111

WP2785, WP3334, WP3574, WP3339, WP2790, WP2766 not present in version 66

This PW has been updated to version 65, but doesn't exist in 66. However, the description mentions that it will be created in the future. I'll leave the pathway in for now, have to check version 67 for presence of PW.

Build instructions are missing

How can I compile the source code, and how do I run the converter? What do I need to do to set up MySQL? Etc.

Problem in WP1825

Warning! Requested path "rectangle-double" is not available with linetype of "Double". Using linetype of "Solid" instead

Change RNA color to purple

RNAs should be coloured in purple

Layout of complex components

Optimize layout of complex components to save space

Datasource

All proteins annotated with datasource uniprot-swissprot should be annotated with uniprot-trembl

Created GPML file names

Names of GPML files should be changed to reactomeID.gpml i.s.o. pathway title. (lots of special characters)

"PubChem Compound" & "NCBI Nucleotide" unknown DataSource v66 (and before)

Another 2 exceptions, which I've seen for version 66 but also previous ones:

"java.lang.IllegalArgumentException : no DataSource known for PubChem Compound "
&
"java.lang.IllegalArgumentException : no DataSource known for NCBI Nucleotide "

Have to check out why the datasource annotation is not working in this case...

trim spaces in text labels of data nodes

In the current GPML created some labels have redundant spaces at the end of labels, e.g. in WP3579:

<DataNode TextLabel="PC " GraphId="cdb40" Type="Metabolite" GroupRef="bbf32">
  <Attribute Key="cellular_location" Value="endoplasmic reticulum membrane" />
  <Attribute Key="complex_id" Value="R-ALL-5684863" />
  <Attribute Key="copies_num" Value="1" />
  <Graphics CenterX="240.0" CenterY="1851.0" Width="100.0" Height="20.0" ZOrder="32768" FontSize="10" Valign="Middle" Color="0000ff" />
  <Xref Database="ChEBI" ID="CHEBI:16110" />
</DataNode>

Problem with SBGN interactions

The SBGN interactions in the reactome pathways are not converted properly. ANy ideas about how we should go about converting them ?

For eg in the pathway http://test2.wikipathways.org/index.php/Pathway:WP2638
We see the following problem :

Store duplicates found in complex component list

Duplicates found in complex component list should be store instead of deleted

Add datasource as proper hyperlinks

Change datasource hyperlinks in the WikiPathways description

WP2655 not getting visualized on test2

Missing Reactome Pathway

"Glycerophospholipid biosynthesis": http://www.reactome.org/PathwayBrowser/#SPECIES=48895&DIAGRAM=5953473

reported by Egon

Exception: class not found -> renderableChemicalDrug v66

During the conversion of version 66, there was an exception printed out:
java.lang.ClassNotFoundExeption : org.gk.render.RenderableChemicalDrug .

Could this class be an update to the Reactome Schema?

Will work on printing an error report specifically for these exceptions, so we can locate them in the Reactome PW + schema, and go from there.

Annotate Pathway Nodes with WikiPathways IDs

pathways have unconnected lines: caused by converter or in original Reactome version?

a clear from this recent curation event in WP4082:

the build.xml has hardcoded paths

... pointing the /home/... Please use relative paths.

Look at the not converted list of Plant pathways

Some of the plant pathways in the list seem to have a diagram. Check out what might be wrong : https://cloud.bigcat.maastrichtuniversity.nl/public.php?service=files&t=16fe7b31fb69cb5c3efcf26427b81cf4

Change Complex colour to brown

Problems with Plant Pathways

Hi Anwesha,

After reviewing about half of the pathways, here’s what I found:

1. On some pathways, one or more nodes are missing. It seems to coincide with nodes that have “Reference Entity” listed as “Unkown:Unknown" in Reactome. I don’t recall if this was the case with the human converted pathways also. Could we still add the node, just with an empty xref, but with the same label as Reactome? Example: http://www.wikipathways.org/index.php/Pathway:WP3111
2. So far, I only found one pathway with serious issues: http://www.wikipathways.org/index.php/Pathway:WP2972
3. Minor: The pathway titles should be capitalized.

Thanks,

Kristina

write documentation for command line tool

See empty https://github.com/wikipathways/reactome2gpml-converter/wiki/User-Guide-for-converting-Reactome-pathways-to-GPML-from-the-command-line

Create link between complex nodes and their components

Complex nodes and their component should be linked using a dynamic property complex_id. This can be used to connect the components in the RDF and in the ComplexViz Plugin.

Update with BridgeDb 2.2

Pathway with no elements

WP2767 does not have any elements: http://wikipathways.org/index.php/Pathway:WP2767

reported by Anders

WP2719 has a Reactome ID for a metabolite which has an incorrect format

It has this GPML, but 111875 is not a valid Reactom ID according to identifiers.org:

  <DataNode TextLabel="Ca2+" GraphId="a415d" Type="Metabolite">
    <Attribute Key="cellular_location" Value="endoplasmic reticulum lumen" />
    <Graphics CenterX="2809.0" CenterY="582.0" Width="50.0" Height="30.0" ZOrder="32768" FontSize="10" Valign="Middle" Color="0000ff" />
    <Xref Database="Reactome" ID="111875" />
  </DataNode>

But there are more pathways with integer Reactome IDs.

Many pathways are not renderable

This is the error I get using the converter.
GKBReader.createRenderableFromType(): java.lang.ClassNotFoundException: org.gk.render.RenderableEntitySet
java.lang.ClassNotFoundException: org.gk.render.RenderableEntitySet
at java.net.URLClassLoader$1.run(URLClassLoader.java:217)
at java.security.AccessController.doPrivileged(Native Method)
at java.net.URLClassLoader.findClass(URLClassLoader.java:205)
at java.lang.ClassLoader.loadClass(ClassLoader.java:323)
at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:294)
at java.lang.ClassLoader.loadClass(ClassLoader.java:268)
at java.lang.Class.forName0(Native Method)
at java.lang.Class.forName(Class.java:191)
at org.gk.persistence.GKBReader.createRenderableFromType(GKBReader.java:648)
at org.gk.persistence.GKBReader.createRenderableFromElement(GKBReader.java:632)
at org.gk.persistence.GKBReader.openNodes(GKBReader.java:474)
at org.gk.persistence.GKBReader.openProcess(GKBReader.java:398)
at org.gk.persistence.GKBReader.openProcess(GKBReader.java:312)
at org.gk.persistence.DiagramGKBReader.openProcess(DiagramGKBReader.java:68)
at org.gk.persistence.DiagramGKBReader.openDiagram(DiagramGKBReader.java:62)
at org.reactome.convert.common.AbstractConverterFromReactome.queryPathwayDiagram(AbstractConverterFromReactome.java:199)
at org.gk.gpml.ReactometoGPML2013.convertPathway(ReactometoGPML2013.java:489)
at org.gk.gpml.CLIConverter.convertReactomeToGPML(CLIConverter.java:126)
at org.gk.gpml.CLIConverter.convertReactomeToGPMLByID(CLIConverter.java:162)
at org.gk.gpml.CLIConverter.main(CLIConverter.java:42)

The renderable entity set (equivalent to the Pathway Class on our end) seems to be missing.

List of non renderable pathways in version 50 : https://cloud.bigcat.maastrichtuniversity.nl/public.php?service=files&t=e5900286773689ce08820c088bd57f5e

Possible problem with referral to groups

Anwesha, can you please have a look at these two WikiPathways commits I made:

http://www.wikipathways.org/index.php?title=Pathway%3AWP2678&diff=78711&oldid=76870
http://www.wikipathways.org/index.php?title=Pathway%3AWP2792&diff=78712&oldid=77083

In both cases, the GPML has @GroupRef attributes that point to a Group element that does not exist.

Can you check in your code where this problem is coming from? The conversion or the Reactome data?

Nucleotide salvage PW not properly converted v.62 Reactome (8956321)

The GPML file is created, however it is empty.
There are no reported errors in the log file.
The PW still exists in Reactome; the parent PW is created (Metabolism of nucleotides, 15869), however this is a meta PW. The child pathways 'Purine salvage' and 'Pyrimidine salvage' are not created by the converter.

So, for now, I will keep the original PW at version 61 4082, and check if version 63/64/65/66 improve something in the conversion; If not, we should check the MySQL database to see what is going wrong here.

capitalize pathway titles

WP1903 has a complex with an ChEBI identifiers which is not an ChEBI identifier

The output GPML on WikiPathways is (r83141):

  <DataNode TextLabel="DAG" GraphId="e667d" Type="Complex">
    <Attribute Key="cellular_location" Value="plasma membrane" />
    <Graphics CenterX="491.5" CenterY="29.0" Width="35.0" Height="22.0" ZOrder="32768" FontSize="10" Valign="Middle" ShapeType="RoundedRectangle" Color="a52a2a" />
    <Xref Database="ChEBI" ID="R-HSA-CHEBI:17815" />
  </DataNode>

However, "R-HSA-CHEBI:17815" is not a ChEBI identifier. I am not sure how to check the Reactome source and not sure if the problem is in the convertor code or in the Reactome pathway source.