LibSBGN Meeting notes, March 10 2011
Present: Alice, Anatoly, Augustin, Falk, Martijn, Tobias
Report from Gatersleben
- Alice, Tobias, Martijn and Falk met in Gatersleben. We worked on ER implementation, which resulted in several test cases and proposals.
- We also started writing up a manuscript.
- Anatoly asks if the library is mature enough for publication? - It is definitely mature enough for somebody who wants to add SBGN-ML support to a tool. Our main goal now is to draw new developers.
- Augustin is in favor, he remarks that validation will be important for the reviewers.
- Which journal to choose?
ACTION: Falk will send a presubmission enquiry.
Discussion of proposals
Proposals 1 and 3
- Anatoly remarks that proposal 1 and 3 have the issue that the position of the port could be outside the parent arc, this must be validated. Alice: yes, and we've seen this problem in PD with arcs pointing to process nodes, the coordinates of the arc don't have to match the coordinates of the port.
- Anatoly asks who is responsible for validation, the reader or the writer? - Both must do validation. You have to validate upon reading because external files could be corrupted. You have to validate upon writing to guard against bugs in your own code.
- Anatoly asks if SBGN-ML can be used to store a graph without coordinates? - Alice: no it was decided a long time ago that this is outside the scope of LibSBGN.
Proposals 1 and 3 are passed without further discussion.
Proposals 2 and 4
- Regarding proposal 2, the part before the implicit xor in an assignment arc doesn't really have a name, although it has its own look (i.e. no arrow head)
- For proposal 4, there are two options to group elements belonging to an interaction. 4a proposes a utility function, 4b proposes a grouping element
- Solution 4b has the advantage of being platform-independent
- 4b looks a bit like a hyperedge
- If we go for 4b, we might also want to revisit the solution for proposal 2, which has a similar hyperedge-like quality
ACTION: Alice will raise naming of assignment arcs to the SBGN editors.
ACTION: Martijn will adjust proposal 2, and put 4a and 4b up for a vote.
- Proposal 5: there isn't much opposition to this proposal anymore. Anatoly recognizes the extra effort that maintaining multiple namespaces bring.
- GraphML was brought up. Why don't we use GraphML?
- GraphML must be heavily extended for our purposes, these extensions will be proprietary (from a non-LibSBGN standpoint).
- Tobias: exchange of GraphML between tools doesn't work in practice.
- GraphML doesn't have tool support that is as good as XML proper.
ACTION: Martijn will put up proposal 5 for a vote for confirmation
- We want to build a validation layer with schematron. For each validation rule, we'll need
- an identifier
- textual description
- A schematron rule
- SBGN-ML negative test-case
- SBGN-ML positive test-case
- We'll start gathering rules on this wiki page: ValidationRules
- For Alice, this can go hand in hand with her SBGN editorial duties.
- According to Augustin, textual description can be stored in schematron xml.
- Augustin also suggests to also include "warning rules" for usage that is not wrong but goes against recommendation.
- A schematron rule and ant target to run is already checked into subversion by Augustin
ACTION: everybody can start adding rules to the ValidationRules wiki page
ACTION: Martijn and Augustin will create a SchematronHowto wiki page
ACTION: everybody: read up on schematron before Harmony, start playing with a couple of rules.
- After the release, we have to update our Roadmap. After some discussion, we came up with this:
- Milestone 2: Support for ER and AF. Addition of schematron validation rules. Release targeted around COMBINE 2011.
- Milestone 3: Detailed graphical specification, and extensibility
- Milestone 4: Annotations, linking with external models (e.g. BioPAX), and MIRIAM compliance.
- Extensibility support, i.e. opening SBGN-ML to third-party extensions, was briefly discussed.
- Advantage: allows to support more use cases, which can draw in more users
- Disadvantage: if an extension becomes too popular, this might cause the same issues as any non-standard format.
- Falk: let's put exensibility support in M3, after finalizing ER and AF support.
Bug tracker update
- The bug tracker was quickly updated to reflect the new roadmap.
- During the Harmony meeting in NY we'll try to set aside a session just for creating rules. This could be any day but Friday, as Falk has to leave early.
- We'll also set up an informal dinner on Sunday evening, for those who are there.
- There won't be another online meeting before Harmony.
- A recording of this meeting is available from Martijn upon request.