The BioPAX Object Model, API and utilities implemented in Java
Home Page: http://biopax.github.io/Paxtools/
License: MIT License
Hi there, what is the most efficient way if I have a list of around ~50 genes to output a data frame in r that contains only the controls-state-of interaction type but for all the ~50 genes? When I query using this code it just takes such a long time and is outputting way more than I need.
t1 <- graphPc(source = 50_genes, kind = "NEIGHBORHOOD", format = "SIF", verbose = TRUE, limit = 1)
Is this a problem. i.e., should it be logged as warning or just info/debug? -
Attn: @ozgunbabur
When the JSON-LD (json-converter module) converted is used multiple times by some long living server app, such as e.g. cPath2, it can make and keep lots of large temporary BioPAX files.
Add useful metadata to SBGN nodes/edges (glyphs, arcs, ports) either as nested xml elements ('bp:' or better define own schema if needed) inside <extensions/> and <notes/>, or simply using <![CDATA[...]]> and encoded JSON object in it.
Include, e.g.:
<label/> contains display name already);One way to implement this feature in sbgn-converter, in a very flexible/extensible manner, would be using key:value entries from the Paxtools BioPAX object.annotations (bpe.getAnnotations()) map (and the annotations can be created in many different ways, e.g., within cPath2 or Sifgraph methods).
Currently, it builds only with JDK8 (and uses Java 1.6 as its source/target compatibility level).
We really need to upgrade/migrate Paxtools project configuration (POM) to make it possible at least to build with JDK9-11, a.s.a.p. And we should also increase the compatibility level to Java 1.8.
PS: I've successfully done this for other projects already, e.g., cpath2 and validator.
Looks like, UniProt (incl. Trembl) and Uniprot Isoform ID patterns somewhat overlap, which makes some IDs match both patterns, and then this code replaces 'uniprot' or 'uniprot knowledge' db name with 'uniprot isoform' (which also results in uniprot.isoform URI)
This was originally reported by Dylan (@d2fong) via email:
"
...
For compartment ids, it seems the main inconsistency between them is the capitalization of the terms. I think this seems to be an easy fix in the java sbgn-converter where we could just transform the string to lowercase before populating the compartment id.I have also seen roughly 2 abnormal characters in ids: '-' and ','.
...
"
E.g., as the result, there are both "Cytosol" and "cytosol".
There are errors, such as missing description for method arguments, use of
, etc., that makes 'site' goal fail (perhaps we'd use an older version of the site or javadoc plugins, etc., but it definitely worth polishing the javadoc).
SBGN converter does not work for control-control and modulation cases - it only generates two participants glyphs but no interactions, no edges.
Fixing this issue will help PathwayCommons/factoid#401.
(I will generated test data models and tests to fix shortly.)
Both in the released v4.3.1 and current 5.0.0-SNAPSHOT, the description of the commands is wrong;
in fact, 'toSif' can optionally have additional (mediator) columns, and 'toSifnx' - cannot.
..to generalize, make less specific to Pathway Commons BioPAX, "whitelist" (keep always not blacklisted) some molecules, etc...
The issue seems to occur when converting the large PC2v8 biopax model to GSEA or SIF.
It still needs to be confirmed where exactly, but looks like it's either in gsea-converter or, highly possible, in biopax pattern module (which uses transitive property paths a lot, e.g., "PhysicalEntity/controllerOf/controlled*:Interaction", etc.)...
Exception in thread "pool-2-thread-3" java.lang.StackOverflowError
at java.lang.AbstractStringBuilder.expandCapacity(AbstractStringBuilder.java:130)
at java.lang.AbstractStringBuilder.ensureCapacityInternal(AbstractStringBuilder.java:114)
at java.lang.AbstractStringBuilder.append(AbstractStringBuilder.java:415)
at java.lang.StringBuilder.append(StringBuilder.java:132)
at java.lang.Class.toString(Class.java:151)
at java.lang.String.valueOf(String.java:2854)
at java.lang.StringBuilder.append(StringBuilder.java:128)
at org.biopax.paxtools.impl.BioPAXElementImpl.hashCode(BioPAXElementImpl.java:224)
at java.util.HashMap.hash(HashMap.java:366)
at java.util.HashMap.put(HashMap.java:496)
at java.util.HashSet.add(HashSet.java:217)
at org.biopax.paxtools.controller.TransitivePropertyAccessor.transitiveGet(TransitivePropertyAccessor.java:40)
at org.biopax.paxtools.controller.TransitivePropertyAccessor.transitiveGet(TransitivePropertyAccessor.java:42)
at org.biopax.paxtools.controller.TransitivePropertyAccessor.transitiveGet(TransitivePropertyAccessor.java:42)
at org.biopax.paxtools.controller.TransitivePropertyAccessor.transitiveGet(TransitivePropertyAccessor.java:42)
at org.biopax.paxtools.controller.TransitivePropertyAccessor.transitiveGet(TransitivePropertyAccessor.java:42)
...
It's in the recursive method private void transitiveGet(D bean, Set values)
I think, the biopax model may contain circles/loops, such as a complex contains itself (perhaps those do not have much sense and have to be fixed by the validator), but transitiveGet method must survive and handle this anyway.
As GNU Trove4j is deprecated perhaps the paxtools-library should be removed?
Currently, Paxtools API allows one set an incompatible type EntityReference using setEntityReference(EntityReference er) method without getting any exceptions or even log messages (exception is thrown only if data are read from a file/stream, but not when one does e.g., protein.setEntityReference(smallMoleculeRef).
In some cases, the SBGN-ML retrieved from the PC2 web service contains several nodes that are not connected to anything else.
Pathway: 'Signaling by BMP' -- Reactome
Notes: Viewing the associated BioPAX record in ChiBE reveals ‘Member of’ edge connecting these nodes.
Minimum code to reproduce
BioPAXFactory factory = BioPAXLevel.L3.getDefaultFactory();
Model model = factory.createModel();
model.addNew(EntityReference.class, "3d8d8d63-997e-4967-9469-0da3f691f659");
Error
For version 5.2.1-SNAPSHOT:
18:21:48.865 [main] INFO org.biopax.paxtools.util.BPCollections - System property: paxtools.CollectionProvider=null
18:21:48.867 [main] INFO org.biopax.paxtools.util.BPCollections - Using the default CollectionProvider (creates HashMap, HashSet).
18:21:48.871 [main] ERROR org.biopax.paxtools.model.BioPAXFactory - Failed creating BioPAX object: interface org.biopax.paxtools.model.level3.EntityReference, URI: 3d8d8d63-997e-4967-9469-0da3f691f659
java.lang.InstantiationException: null
at sun.reflect.InstantiationExceptionConstructorAccessorImpl.newInstance(InstantiationExceptionConstructorAccessorImpl.java:48)
at java.lang.reflect.Constructor.newInstance(Constructor.java:423)
at org.biopax.paxtools.model.BioPAXFactory.create(BioPAXFactory.java:76)
at org.biopax.paxtools.impl.ModelImpl.addNew(ModelImpl.java:120)
at factoid.model.TemplatesModel.main(TemplatesModel.java:254)
Exception in thread "main" java.lang.NullPointerException
at org.biopax.paxtools.impl.ModelImpl.add(ModelImpl.java:158)
at org.biopax.paxtools.impl.ModelImpl.addNew(ModelImpl.java:121)
at factoid.model.TemplatesModel.main(TemplatesModel.java:254)
For version 5.1.0:
SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".
SLF4J: Defaulting to no-operation (NOP) logger implementation
SLF4J: See http://www.slf4j.org/codes.html#StaticLoggerBinder for further details.
Exception in thread "main" java.lang.NullPointerException
at org.biopax.paxtools.impl.ModelImpl.add(ModelImpl.java:158)
at org.biopax.paxtools.impl.ModelImpl.addNew(ModelImpl.java:121)
at BiopaxExp.BiopaxExp.Main.main(Main.java:14)
Debugging observations so far
As it is show in the screenshot above constructorAccessor property is set to instantiationexceptionconstructoraccessorimpl while it is set to delegetingconstructoraccessorimpl for the other element types that I tried so far and was able to add with success (e.g. Interaction).
When PC is queried to return the result in SBGN, it used to be the case that complex members were part of their parent/owner complexes (in SBGNML, member glyphs were simply listed inside their owner complex glyphs) but that no longer seems to be the case. From the layout of the returned pathway, it seems like members are properly contained / nested in their parent complexes but that is not the case!
Also see iVis-at-Bilkent/newt#176
As we explained, there listed (unification and relationship xrefs') ID's were not designed for ID-mapping (e.g., from HGNC Symbol to NCBI Gene ID).
We cannot completely solve this due to the xrefs come from original data and sometimes multiple BioPAX entity reference types, but could probably decrease the level of confusion if we'd skip outputing for relationship Xrefs, or at least skip here - https://github.com/BioPAX/Paxtools/blob/master/pattern/src/main/java/org/biopax/paxtools/pattern/miner/ExtendedSIFWriter.java#L193 - for RelationshipXref having relationshipType other than "identity"...
With some BioPAX data (quite large but normal, not too complex), it takes too long to apply the default layout (currently, it's done by default), despite converting to SBGN-ML takes just a few seconds.
E.g., I run
java -Xmx16g -jar paxtools.jar toSBGN beta-pc9-transfac-pathway-CAGGTG_V.owl CAGGTG_V.sbgn.xml
, and after a few seconds, when converting from BioPAX to SBGN is done, it prints -
SBGN-PD Layout is running...
success ratio: 0.6532992762877821
enhanced ratio: 0.838058748403576
Total execution time: 1336800 miliseconds.
The layout is implemented in the chilay "library".
Attached is an example of the use of blank spaces in IDs (e.g. "extracellular matrix") when querying PC for SBGN output. This creates problems in libSBGN library (in handling SBGNML files).
BRCA1_BRCA2_COMMONSTREAM.sbgnml.txt
Hi,
I'm trying to update MICommunity/psimi which requires org.biopax.paxtools.io.jena.JenaIOHandler. However commit e63cdcc removed this library. What is the suggested method to migrate?
@emekdemir @ozgunbabur
This method
https://github.com/BioPAX/Paxtools/blob/master/paxtools-core/src/main/java/org/biopax/paxtools/controller/ModelUtils.java#L1149
does not replace merged interactions in all the biopax object properties, such as, e.g., not in stepConversion, etc.
This method should call ModelUtils.replace instead of updating just a couple of properties...
Currently, e.g., SimpleIOHandler can sets " name " value instead of (trimmed) "name" from <bp3:displayName> name </bp3:displayName> ...
Confirmed using 'All' BioPAX archive from Pathway Commons v8.
This issue here causes PathwayCommons/cpath2#258.
@emekdemir , @ozgunbabur shall we urgently fix the Control interface, or at least ControlImpl (which allows for multiple values, and where checkControlled(controlled) always returns 'true')?
@ozgunbabur , I see lots of warnings in PC logs (cpath2, PC9, which uses Paxtools) like:
WARN o.b.p.q.w.GraphL3Undirected - Invalid BioPAX object to wrap as node. Ignoring: http://identifiers.org/wikipathways/WP143/dd235
Why is this a warning, is this a problem, bug?.. If not, why to log as warning instead of debug?..
I suggest we make it at least an option for the Completer to skip sub-pathways. This would make large and nested BioPAX sub-model that can result from a PC2 web service query ('get' or 'graph') more usable and reasonable for converting to SBGN-ML, SIF and for vizualisation.
Looks, this cannot be simply done via @autocomplete annotation on properties 'pathwayComponent' and 'controlled', because the properties' range is of either Pathway or Interaction type...
But we could e.g. skip deep traversing here if the value is of type Pathway.
@ozgunbabur , @emekdemir, @gbader share your thoughts please!
@emekdemir @ozgunbabur
Seems, nobody uses it; and the source code, e.g. -
https://github.com/BioPAX/Paxtools/blob/master/paxtools-core/src/main/java/org/biopax/paxtools/controller/Integrator.java#L868
is, well, I've no words for this (- waste is the most neutral perhaps).
Running graph queries on the entire SIF model vs. corresponding BioPAX could greatly improve performance. We'd implement this in paxtools-query module. But there are several problems...
For example, a SIF file/model, such as Pathway Commons' one, usually contains one ID type (HGNC Symbol or UniProt AC) for genomic entities and one (ChEBI) for chemicals, and misses quite a few interactions due to lack of those IDs (we'd use a URI based version of SIF; anyway, id-mapping is required, etc..)
Refs PathwayCommons/cpath2#202
Refs #19
This can help issue: PathwayCommons/app-ui#326
Dear BioPax developers,
On line 669 of the SimpleIOHandler the datatype print is slightly off.
Using a tool like rapper or riot to convert the output of the rdf/xml to n-triples, we see the difference (example from rhea-db)
<http://biopax.rhea-db.org/level3/57967_stoichiometry_right_3249>
<http://www.biopax.org/release/biopax-level3.owl#stoichiometricCoefficient>
"1.0"^^<xsd:float> .
was printed but we expect
<http://biopax.rhea-db.org/level3/57967_stoichiometry_right_3249>
<http://www.biopax.org/release/biopax-level3.owl#stoichiometricCoefficient>
"1.0"^^<http://www.w3.org/2001/XMLSchema#float> .
Quite a lot of tools auto-correct this difference down stream. But if one uses a tool that is spec rdf/xml compliant this gives a difference when doing SPARQL values comparison. Changing from a numeric comparison to a string comparison.
We will open a PR with this fix soon.
Regards,
Jerven
Would it be possible to switch from commons-lang 2 to commons-lang 3?
If so, I'd be willing to submit a PR.
Hello Ozgun,
I'm working with the L3ToSBGNPDConverter.java trying to find a way to use pre-determined points to position the layouts as an alternative to the Chilay layout that is currently being applied in the createSBGN() method. I would like to ask you a few questions on this and how you might go about doing it.
I would really appreciate your opinion and if you have some time to chat my email is '[email protected]'.
Thanks for the consideration,
Peter
The artifact id to import BioPax as a Maven project is actually 'paxtools-core', but the downloads page (http://search.maven.org/#artifactdetails%7Corg.biopax.paxtools%7Cpaxtools%7C4.3.1%7Cpom) indicates it as 'paxtools' and Maven is unable to recognize this.
@emekdemir @gbader @ozgunbabur
According to the comments about that BioPAX property http://www.biopax.org/owldoc/Level3/classes/Stoichiometry___10864550.html (- there can be 0 or 1 Stoichiometry per physical entity) and that Stoichiometry objects cannot be directly associated with any interactions (unfortunately by design), the physicalEntity property should be both functional and inverse functional.
Here, the BioPAX model xml:base should be matched instead of constant URI prefix which in fact can be different in different pathway data files...
PS:
Ideally, we'd probably URI-encode the BioPAX URIs instead of stripping the base and replacing special characters (though this should be probably done in a way to allow mapping from the SBGN XML IDs back to original BioPAX RDF/JSONLD URIs, at least for physical entities and pathways; this is very important for web apps).
Hi there,
I need to use Paxtools through Python. Therefore, I am running the Java code of Paxtools through Python. I am trying to load a biopax model, making some process on it and finally converting it to sbgn by using sbgn-converter of paxtools. Everything works fine except the biopax to sbgn conversion.
The problem that happens for the conversion step is that when I call writeSBGN() of L3ToSBGNPDConverter the python rocketship icon appears and then the process had not end though I have waited for a while.
I figured out that the problem is gone when I tried to totally eliminate the layout call inside createSBGN() by commenting out the fallowing code segment:
final boolean layout = doLayout && n < this.maxNodes && !arcMap.isEmpty();
try {
//Must call this, although actual layout might never run;
//in some real data tests, skipping createLayout method
//led to malformed SBGN model, unfortunately...
(new SBGNLayoutManager()).createLayout(sbgn, layout);
} catch (Exception e) {
throw new RuntimeException("SBGN Layout of " + model.getXmlBase()
+ ((model.getName()==null) ? "" : model.getName()) + " failed.", e);
}
if(!layout) log.warn(String.format("No layout, for either " +
"it's disabled: %s, or ~ no. nodes > %s: %s, or - no edges: %s",
!doLayout, maxNodes, n>maxNodes, arcMap.isEmpty()));
The problem persists when I try to prevent the layout by setting the doLayout parameter as false. BTW I have tried two different libraries to run Java through Python. That libraries are jnius and jpype. The same thing has happened in both cases.
I have a few questions considering these:
doLayout parameter as false)I got a BioPAX neighborhood sub-network using Asmita's gene set; see PathwayCommons/cpath2#257 (it's 82 Mb). Converting that to SIF takes to long (and I am still waiting... and even begin suspecting endless loop...) even if I use Paxtools/Java and only one rule:
java -Xmx32g -jar paxtools.jar toSIF graph.owl.gz graph.sif.txt seqDb=hgnc chemDb=chebi include=controls_state_change_of
When using the paxtools pattern librry's SIFSearcher from pcviz (uniprot-scraper script), got several of these exceptions:
Exception in thread "pool-7071-thread-1" Exception in thread "pool-7071-thread-6" Exception in thread "pool-7071-thread-2" Exception in thread "pool-7071-thread-3" Excep
tion in thread "pool-7071-thread-11" Exception in thread "pool-7071-thread-4" Exception in thread "pool-7071-thread-8" Exception in thread "pool-7071-thread-5" java.lang
.NullPointerException
at org.biopax.paxtools.pattern.Searcher.search(Searcher.java:174)
at org.biopax.paxtools.pattern.Searcher.search(Searcher.java:158)
at org.biopax.paxtools.pattern.miner.SIFSearcher$1.run(SIFSearcher.java:163)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)
...
Completer's filter uses the default values (forward=true;backward=false) for all the object properties and thus follows e.g. nextStep property despite this annotation. I tested and confirmed that this always returns null.
I suspect the annotation is not available at runtime (probably, due to no retention value is there set)...
This also causes PathwayCommons/cpath2#267.
(I build, test and run Paxtools with Oracle JDK1.8 on Ubuntu Linux; that might matter...)
@ozgunbabur @emekdemir
Almost fixed this bug.. wait.
Example text:
<label text="dephosphorylated "receiver" RAF/KSR1"/>
The label has "receiver" which closes the value for the text attribute and causes an xml parsing error.
It's easy to see that Cloner object does not actually use the Model passed via the first parameter of the clone method.
The Cloner defines a Traverser with the Visitor being the Cloner itself, implementing 'visit' method, which does not use the "source" model at all; it only uses the internal property - 'targetModel', which is also not a good idea, because 'targetModel' gets renewed and returned every time the clone(..) method is called; so Cloner cannot be run in several threads...
Possible fix:
public synchronized Model clone(Set<BioPAXElement> toBeCloned) instead;Model targetModel = factory.createModel(); and pass it to the traverser as traverser.traverse(bpe, targetModel); (instead of passing 'source' there);@ozgunbabur , @emekdemir please review this idea (I can fix then)!
The idea is to make the sbgn-converter generate nodes/edges for "empty" and blackbox sub-pathways
(i.e., for those that have no interactions in them originally or as the result of detaching the sub-network from a larger, e.g. Pathway Commons', BioPAX model).
ModelUtils.getParentPathways (paxtools-core) and therefore SearchEngine.index (paxtools-search) gets into endless loop when the translated, normalized KEGG data is used (or any model that integrates that data).
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.