OBO round trip

From WandoraWiki
(Difference between revisions)
Jump to: navigation, search
(OBO-Wandora-OBO round trip results)
 
(12 intermediate revisions by one user not shown)
Line 1: Line 1:
OBO round trip is a special import-export tool found in '''File > Export > OBO round trip...'''. Tool is used to test how OBO flat file ontologies convert to topic maps and back again to OBO ontologies. Tool can round trip several ontologies at once. Round tripped OBO files are stored to '''wandora_round_trip''' folder beside original OBO files. Conversion cycle should not loose or add any information to ontologies. Conversion changes can be tracked for example with the '''obodiff''' application shipped with the [http://www.oboedit.org OBOEdit].
+
OBO round trip is a special import-export tool found in '''File > Export > OBO round trip...'''. Tool is used to test how OBO flat file ontologies convert to topic maps and back again to OBO ontologies. Tool can round trip several ontologies at once. Round tripped OBO files are stored to '''wandora_round_trip''' folder beside original OBO files. Conversion cycle should not loose or add any information to ontologies. Conversion changes can be tracked with the '''obodiff''' application shipped with the [http://www.oboedit.org OBOEdit] for example.
  
 
OBO round trip uses tool's own internal topic map for OBO-topic map-OBO conversion and leaves the current topic map untouched in Wandora.
 
OBO round trip uses tool's own internal topic map for OBO-topic map-OBO conversion and leaves the current topic map untouched in Wandora.
  
Round tripping current (2008-01-22) OBO ontologies available at [http://www.obofoundry.org OBO Foundry], and comparing round tripped ontologies to original ontologies gives a [http://www.wandora.org/wandora/download/other/round_trip_compare_results.txt result file]. Some notes about the results:
+
== OBO-Wandora-OBO round trip results ==
  
Synonym differences on '''envo.obo''' are due to a design decision where OBO export tries to collapse all identical synonyms into a single synonym with several origin definitions. '''envo.obo''' contains identical synonyms such as:
+
Round tripping current (2008-01-22) OBO ontologies available at [http://www.obofoundry.org OBO Foundry], and comparing round tripped ontologies to original ontologies with '''obodiff''' gives a [http://www.wandora.org/wandora/download/other/round_trip_compare_results.txt result file]. Some observations follow.
 +
 
 +
Synonym differences on '''envo.obo''' are due to a design decision where Wandora's OBO export tries to collapse all identical synonyms into a single synonym with several origin definitions. The original '''envo.obo''' contains identical synonyms such as:
  
 
   id: ENVO:00000004
 
   id: ENVO:00000004
Line 15: Line 17:
 
   id: ENVO:00000004
 
   id: ENVO:00000004
 
   synonym: "administrative division" RELATED [ADL:FTT, Getty:TGN]
 
   synonym: "administrative division" RELATED [ADL:FTT, Getty:TGN]
 
It looks like Wandora doesn't handle all escaped characters right. Look at '''FIX:0001087''' in '''fix.obo''' for example. Backslash in '''{''' and '''}''' has duplicated. Have to fix this!!!
 
  
 
In '''po_temporal.obo''' empty definition origin descriptions do not survive round trip. Original OBO ontology contains for example:
 
In '''po_temporal.obo''' empty definition origin descriptions do not survive round trip. Original OBO ontology contains for example:
Line 28: Line 28:
 
   def: "OBSOLETE. The meristematic zone is formed" [GR:ap]
 
   def: "OBSOLETE. The meristematic zone is formed" [GR:ap]
  
Completeness status changes at '''so.obo''' and '''fly_anatomy.obo''' are due to duplicate relationship definition. Same relationship is defined with modified '''relationship''' and exact relationship term such as '''part_of'''. For example in original '''so.obo''':
+
Completeness status changes at '''so.obo''' and '''fly_anatomy.obo''' seem to be due to a duplicate relationship definition. Same relationship is defined with modified '''relationship''' and exact relationship term such as '''part_of'''. For example in original '''so.obo''':
  
 
   id: SO:0000034
 
   id: SO:0000034
 +
  is_a: SO:0000696 ! implied link automatically realized ! oligo
 +
  intersection_of: SO:0000696 ! oligo
 
   intersection_of: has_quality SO:0001183 ! morpholino
 
   intersection_of: has_quality SO:0001183 ! morpholino
 
   relationship: has_quality SO:0001183 ! implied link automatically realized ! morpholino
 
   relationship: has_quality SO:0001183 ! implied link automatically realized ! morpholino
Line 37: Line 39:
  
 
   id: SO:0000034
 
   id: SO:0000034
 +
  is_a: SO:0000696 ! oligo
 +
  intersection_of: SO:0000696 ! oligo
 
   intersection_of: has_quality SO:0001183 ! morpholino
 
   intersection_of: has_quality SO:0001183 ! morpholino
 
   relationship: has_quality SO:0001183 ! morpholino
 
   relationship: has_quality SO:0001183 ! morpholino
 +
 +
More detailed comparison of original and Wandora roundtripped OBO ontologies is found in MSc Theses Kivelä A.: [http://www.wandora.org/wandora/download/other/gradu_kivela.pdf OBO-ontologioiden kuvaaminen Topic Map-muotoon], 2008. (in Finnish). (English abstract is also available: [http://www.wandora.org/wandora/download/other/gradu_kivela_abstract_en.pdf Converting OBO ontologies to Topic Maps]).

Latest revision as of 21:35, 21 August 2008

OBO round trip is a special import-export tool found in File > Export > OBO round trip.... Tool is used to test how OBO flat file ontologies convert to topic maps and back again to OBO ontologies. Tool can round trip several ontologies at once. Round tripped OBO files are stored to wandora_round_trip folder beside original OBO files. Conversion cycle should not loose or add any information to ontologies. Conversion changes can be tracked with the obodiff application shipped with the OBOEdit for example.

OBO round trip uses tool's own internal topic map for OBO-topic map-OBO conversion and leaves the current topic map untouched in Wandora.

[edit] OBO-Wandora-OBO round trip results

Round tripping current (2008-01-22) OBO ontologies available at OBO Foundry, and comparing round tripped ontologies to original ontologies with obodiff gives a result file. Some observations follow.

Synonym differences on envo.obo are due to a design decision where Wandora's OBO export tries to collapse all identical synonyms into a single synonym with several origin definitions. The original envo.obo contains identical synonyms such as:

 id: ENVO:00000004
 synonym: "administrative division" RELATED [ADL:FTT]
 synonym: "administrative division" RELATED [Getty:TGN]

Wandora round tripped ontology contains these synonyms collapsed into a single synonym specification:

 id: ENVO:00000004
 synonym: "administrative division" RELATED [ADL:FTT, Getty:TGN]

In po_temporal.obo empty definition origin descriptions do not survive round trip. Original OBO ontology contains for example:

 id: PO:0001027
 def: "OBSOLETE. The meristematic zone is formed" [GR:ap ""]

This is round tripped to:

 id: PO:0001027
 def: "OBSOLETE. The meristematic zone is formed" [GR:ap]

Completeness status changes at so.obo and fly_anatomy.obo seem to be due to a duplicate relationship definition. Same relationship is defined with modified relationship and exact relationship term such as part_of. For example in original so.obo:

 id: SO:0000034
 is_a: SO:0000696 ! implied link automatically realized ! oligo
 intersection_of: SO:0000696 ! oligo
 intersection_of: has_quality SO:0001183 ! morpholino
 relationship: has_quality SO:0001183 ! implied link automatically realized ! morpholino

has round tripped to

 id: SO:0000034
 is_a: SO:0000696 ! oligo
 intersection_of: SO:0000696 ! oligo
 intersection_of: has_quality SO:0001183 ! morpholino
 relationship: has_quality SO:0001183 ! morpholino

More detailed comparison of original and Wandora roundtripped OBO ontologies is found in MSc Theses Kivelä A.: OBO-ontologioiden kuvaaminen Topic Map-muotoon, 2008. (in Finnish). (English abstract is also available: Converting OBO ontologies to Topic Maps).