BioPAX Level 2 feedback

Please use this page to post questions, comments, and bug reports on the Level 2 release candidate.

Latest version of the owl file: http://cvs.sourceforge.net/viewcvs.py/*checkout*/biopax/biopax/release2/biopax-level2.owl

Latest version of the documentation: http://cvs.sourceforge.net/viewcvs.py/*checkout*/biopax/biopax/release2/biopax-level2-documentation.pdf

Ontology Feedback

• Joanne Luciano Request formal adoption of issues identified in BioPAXReview.pdf which is an extensive review of the Level 2 Specification document and the Protege OWL Level 2 implementation. I conducted this review at Manchester University (UK) in collaboration with Matthew Horridge two weeks prior to the Japan f2f meeting. Alan Rector, Robert Stevens and Alan Ruttenberg also contriubted to this review which Matthew presented at the JapanF2F meeting in Tokyo.

  • bader - Nov.26.2005 - I think we said we'd do this seriously for Level 3, but some of these are easy to do for Level 2. So, I would propose that we do the easy ones for Level 2 and leave the rest for Level 3, so as not to delay Level 2 further.

  • bader - Dec.9.2005 - as per last conference call, Joanne is going to make a list of proposed easy fixes and send them to the group before Dec.21.2005

    • Joanne Luciano - Dec.21.2005 - Here are the ListOfEasyFixes

• Alan Ruttenberg - Dec.30.2005 Level 2 Release Candidate Feedback/Complex components confusion

• Matthias Samwald - Feb.15.2006 - the organisation of the properties "ORGANISM" and "TISSUE" is highly counter-intuitive and problematic.ORGANISM has the domain {rna, dna, protein, complex, pathway} and the range {biosource}. The description of the property makes it clear that ORGANISM should indeed be used to refer to organisms, e.g. homo sapiens, that contain the referenced biological entity [*]. This does not match with the range of this property. The definition of biosource is: "The biological source of an entity (e.g. protein, RNA or DNA). Some entities are considered source-neutral (e.g. small molecules), and the biological source of others can be deduced from their constituentss (e.g. complex, pathway). Examples: HeLa cells, human, and mouse liver tissue." So, a biosource can also be a "mouse liver tissue", which is not the same as the organism "mouse". biosource also has a property called TISSUE that has the range openControlledVocabulary. This is quite a mess. If we want to say that Protein A originated from the heart tissue of the rat, we would probably need to express this as "<Protein A> <ORGANSIM> <RatHeartTissue>", "<RatHeartTissue> <TISSUE> <HeartTissue>". This is logically false (RatHeartTissue is not an organism but a part of an organism) and highly counter-intuitive. A more consistent approach would be to say, for example: "<Protein A> <TISSUE> <RatHeartTissue>", "<RatHeartTissue> <SPATIAL-PART-OF> <RatOrganism>".

• Andrea Splendiani - Feb.16.2006 - BioPAX namespace should not change globally across levels, but rather only new components (classes and properties) should be labeled with a new namespace. More at /BioPAX namespace issue.

• bader - Feb.20.2006 - PARTICIPANT property in sequenceParticipant should be restricted to be only sequences (DNA, RNA, protein)

• Emek Demir- Feb.21.2006 - Why availability is multiple?

• Emek Demir- Feb.21.2006 - There are multiple organisms for Complex. Even when we want to represent multiple specie complexes (i.e. in vitro constructs) I think it is better to represent it with the organism of the component.

• Emek Demir- Feb.21.2006 - There are multiple sequences for DNA? Possibly a bug. Also multiple organisms?

• Emek Demir- Feb.21.2006 - String policies: Do we have restrictions on allowing empty strings? Don't know cases can potentially be represented with empty strings, which is a very bad way to do it, so I think we should not allow empty strings in none of those fields.

  • Alan Ruttenberg We can add this to the ontology. NoEmptyStringExample.

• Emek Demir- Feb.21.2006 - Some naming problems can be found in /Naming Suggestions

• Emek Demir- Feb.22.2006 - PARTICIPANTS -- Non OWL restrictions: Some of the following rules ( that I believe to be valid) are not possible(?) to specify in owl in a general manner ( although one can hack and close each instance manually by cardinality restrictions).

  • Control can not have generic participants.

  • Complex formation must be represented from left to right (?) (Assuming that you mean something like the right has to be a complex with the all the left and their components as it's components. If so can't be done in OWL (triangle) Alan Ruttenberg)

  • Transport must at least two PEPs belonging to the same physical entity, one on LEFT and one on RIGHT, but with different compartments. (Can't be expressed in OWL (triangle) Alan Ruttenberg)

  • Alan Ruttenberg comments above about not possible in OWL have the following qualifications: May be possible in OWL 1.1, May be possible to express intended meaning by modeling in slightly different way.

• Emek Demir- Feb.24.2006 As Alan suggested, we can define for modulation controller=1, controlled=1, participants<=2 restriction to partially close the world. Problems still exist with control and catalysis though.

• Please add any new requests here.

Documentation Feedback

• Matthias Samwald - Feb.15.2006 - Clarify meaning of ORGANISM - I guess that the intention of ORGANISM is to refer to the organism in which the biological entity is found naturally and not under experimental conditions. So, if we would transfect Xenopus oocytes with a human protein, the ORGANISM of the expressed proteins is still 'human', not 'Xenopus'.

  • bader - Feb.15.2006 - Yes, this is correct and we should update the documentation.

• Emek Demir- Feb.21.2006 - Pathway Step should not be used across pathways, although there is a general documentation related to reuse of utility classes, I think it is better if we reinforce it in the Pathway Step documentation as well.

  • Alan Ruttenberg What's the reason for this restriction?. Assuming we want this restriction (and that on utility classes in general), we can add a constraint to the ontology to say so: NoReuseOntologyExample

• Please add any new requests here.