Meeting Minutes 2nd October 2006
Meeting Minutes 2nd October 2006
Present: Poul Nielsen, Shane Blackett, Andrew Miller
Poul: Nicholas Linnaeus from the EBI has sent a letter to groups which maintain a model database, suggesting that the databases be linked, so that additions / changes to models in one repository will result in those model changes automatically propagating to other databases (in the cases where the changes fit within the specificity of each database). Poul has responded to say that he supports this idea.
Andrew noted that the problem with this approach from a CellML perspective is that we need to have automated conversions to / from CellML from the other representations. This is not straightforward, because the exact information represented by CellML does not match what is in e.g. SBML. If there is continual data flow, we can't be losing data.
Shane suggested that one approach would be to merge changes from an earlier version (so CellML => SBML : SBML changed => CellML could recognise the model at SBML => CellML stage, and so perform the conversion by merging with the earlier CellML version).
Poul suggested that we need to put all information which couldn't be represented in core CellML into the metadata, so that there was no loss.
Andrew suggested that we needed to integrate this with the ontologies work, so that changes to SBML models would correspond to ontology changes as well as CellML model changes.
Poul thinks that merging ontologies (if we have common terms e.g. between BioPAX and SBML) is the way to deal with this.
Poul noted that Tommy has applied for the Canadian Police certificate so he can get a visa to continue work here, but has discovered that it will take 150 days. To avoid this, it will be necessary for him to get a shorter visa, or for us to find someone to temporarily fill the position.
Shane has got a Win32 build of mozCmgui, although he has decided to defer work on the code to automatically make the Mozilla side process Win32 events while waiting for CORBA responses, because it will take more than a few days.
Poul: Mike Dunstan is going to help with the work Matt is doing on ontologies. He will probably stay for 4 months, but will be working less than full time.
Poul, Matt, and Andrew interviewed the sole applicant for the model curation job (with the view of putting him into the tools development job, since his background was unsuitable for the curation job). However, the candidate did not appear to be suitable for the tools development job either.
There has been another applicant for the model curation job with a biological sciences background, Poul will follow up.
Poul will also look into getting advertisements for the tools development job up where computer science students might see them (has to be before the exams).
Andrew noted that Andre has been trying to use entitites to define relative URLs. Andrew has suggested that he just put relative URLs directly into the xlink:href. This case needs to be fixed for the CellML DOM API (PCEnv alread supports it, because it uses Mozilla to load the documents).
Andrew has made a dated build of PCEnv that works on Linux. Poul is keen for this to be released, even if we don't have the other platforms working yet. Win32 is being held up by some crashes have to be debugged, and OS X hasn't been tested because an application bundle needs to be produced before the XULRunner instance can be interacted with (Andrew has almost completed the process of producing one).
Poul: Nicholas Linnaeus from the EBI has sent a letter to groups which maintain a model database, suggesting that the databases be linked, so that additions / changes to models in one repository will result in those model changes automatically propagating to other databases (in the cases where the changes fit within the specificity of each database). Poul has responded to say that he supports this idea.
Andrew noted that the problem with this approach from a CellML perspective is that we need to have automated conversions to / from CellML from the other representations. This is not straightforward, because the exact information represented by CellML does not match what is in e.g. SBML. If there is continual data flow, we can't be losing data.
Shane suggested that one approach would be to merge changes from an earlier version (so CellML => SBML : SBML changed => CellML could recognise the model at SBML => CellML stage, and so perform the conversion by merging with the earlier CellML version).
Poul suggested that we need to put all information which couldn't be represented in core CellML into the metadata, so that there was no loss.
Andrew suggested that we needed to integrate this with the ontologies work, so that changes to SBML models would correspond to ontology changes as well as CellML model changes.
Poul thinks that merging ontologies (if we have common terms e.g. between BioPAX and SBML) is the way to deal with this.
Poul noted that Tommy has applied for the Canadian Police certificate so he can get a visa to continue work here, but has discovered that it will take 150 days. To avoid this, it will be necessary for him to get a shorter visa, or for us to find someone to temporarily fill the position.
Shane has got a Win32 build of mozCmgui, although he has decided to defer work on the code to automatically make the Mozilla side process Win32 events while waiting for CORBA responses, because it will take more than a few days.
Poul: Mike Dunstan is going to help with the work Matt is doing on ontologies. He will probably stay for 4 months, but will be working less than full time.
Poul, Matt, and Andrew interviewed the sole applicant for the model curation job (with the view of putting him into the tools development job, since his background was unsuitable for the curation job). However, the candidate did not appear to be suitable for the tools development job either.
There has been another applicant for the model curation job with a biological sciences background, Poul will follow up.
Poul will also look into getting advertisements for the tools development job up where computer science students might see them (has to be before the exams).
Andrew noted that Andre has been trying to use entitites to define relative URLs. Andrew has suggested that he just put relative URLs directly into the xlink:href. This case needs to be fixed for the CellML DOM API (PCEnv alread supports it, because it uses Mozilla to load the documents).
Andrew has made a dated build of PCEnv that works on Linux. Poul is keen for this to be released, even if we don't have the other platforms working yet. Win32 is being held up by some crashes have to be debugged, and OS X hasn't been tested because an application bundle needs to be produced before the XULRunner instance can be interacted with (Andrew has almost completed the process of producing one).