Annotation Conf. Call 2016-05-10: Difference between revisions

From GO Wiki
Jump to navigation Jump to search
Line 142: Line 142:
*David OS - would it make sense to extend what is in LEGO to annotation extensions so that formally a UniProtKB accession could refer to a gene or a protein, an MGI ID could refer to a gene or a protein
*David OS - would it make sense to extend what is in LEGO to annotation extensions so that formally a UniProtKB accession could refer to a gene or a protein, an MGI ID could refer to a gene or a protein
*Chris - Jenkins checks do not check for gene vs protein, but whether the overall type of entity is correct  
*Chris - Jenkins checks do not check for gene vs protein, but whether the overall type of entity is correct  
'''AI:'''David OS - can say in the OWL, for example, that the domain is a protein and then in a separate mapping file indicate what IDs can be used to indicate a protein
*Ruth - for regulation of transcription, curators could then use a UniProtKB ID to indicate the gene?  This what not be meant to indicate at what level the regulation occurs.






[[Category: Annotation Working Group]]
[[Category: Annotation Working Group]]

Revision as of 11:41, 10 May 2016

Join us using BlueJeans

https://bluejeans.com/993661940

Agenda

1. Review of all GOC Calls and GOC Calendar

  • There are a number of consortium-wide calls that happen during the week and we would like everyone to be aware of them.
  • Each of these calls is listed on the GO Calendar hosted on Google.
    • You may request access to the GO Google Calendar by emailing Kimberly, Paola, Melanie or Moni with your Google email address.
  • In the coming weeks we plan to move away from sending reminder emails about meetings, relying instead on the GO shared calendar to keep all interested parties up-to-date.
  • If you have a meeting you would like to add to the Calendar, please do so.

Current Meetings

Group Day & Time How to join Minutes Repository
LEGO Every Monday, 8:30am US Pacific Time Public BlueJeans Conference Call LEGO Calls GitHub GO-Noctua
Annotation Advocacy 2nd and 4th Tuesdays, 8:00am US Pacific Time Public BlueJeans Conference Call Annotation Conference Calls GitHub GO-Annotation
GO Managers 1st and 3rd Wednesdays, 8:00am US Pacific Time Private. Use the GO passcode at the Jackson Conference Line. GO Managers Calls GitHub GOC
Data Capture Working Group 2nd and 4th Wednesdays, 8:00am US Pacific Time Public Zoom Conference Call Data Capture Calls GitHub Test Repo
GO Directors Every Wednesday, 9:00am US Pacific Time Private. Use the GO passcode at the Jackson Conference Line. - GitHub GOC
GAF/GPAD output for LEGO Every Wednesday at 10:30am US Pacific Time Informal Skype meeting with DavidH, Heiko, and Kimberly. - Please email them to request access. See LEGO. See LEGO.
GO Ontology Editors Every Thursday at 10:30am US Pacific Time Public GoToMeeting Conference Call Ontology Calls GitHub GO-Ontology
PAINT Convenes as needed. Emails sent to GO-discuss - GitHub GO-PAINT
Software Team Convenes as needed. Emails sent to GO-software - GitHub GOC

2. Accepted Identifiers for Gene and Gene Product Entities

  • This issue is about what identifiers are acceptable for use in Column 2 and in Column 16 of the gaf, annotation extensions.

Please see:

Minutes

Berkeley: Chris, Moni, Suzi

dictyBase: Petra

EBI: Melanie

FlyBase: Giulia

MGI: David H., Li

PomBase: Antonia, Midori

Reactome: Peter

RGD: Stan

TAIR: Tanya

UCL: Rachael, Ruth

WB: Kimberly

Zfin: Sabrina

Meetings

  • Table above lists meetings, times, GO calendars - we would like to move towards using the Google calendar as a reminder for meetings
  • Some people do not have the same gmail email address as their work email - can these be synced?
  • Suzi - yes, you can link these
  • People should subscribe to the calendar
  • Guest lists can be made to everyone who wants to be reminded of calls
  • Ruth - reminder email could be to the GO list used to remind subscribers of the calls
  • AI: we will try to set up the GO calendar to send these emails

Identifiers

  • Discussion of what identifiers can be used in Column 2 and Column 16
  • We need to have a consistent use of identifiers
  • Melanie - different threads on annotation and managers call
    • Would be good to have a combined call where we walk through examples - Paul was organizing this?
  • David H. - Propose to do this on an all-hands call?
    • For Column 2, for example, MGI and WB map all IDs to their MOD gene identifiers before submitting to GOC
  • Melanie - one of the issues for Col. 16 was domain and range and enforcing what is used there
  • Suzi - in favor of cleaning up what is used in Col. 16; because of the mixed use here it is more difficult to validate the IDs
  • Ruth - not sure we need to have many examples; no controversy about what is in Col. 2, but there were issues about domain and range wrt Col. 16 - David OS raised some issues with this
  • Rachael - this had to do with consistency with OWL versions
  • Melanie - the target was the issue - its biologically different to be regulating a gene, transcript, or protein
  • David H. - in the ontology, there are domains and ranges, and in the GO view of the world processes regulate processes, not things (entities)
  • David H. - the relation, has_regulation_target, mixes processes and entities with its targets
  • Chris - sometimes gene IDs are used to represent proteins
  • Chris - because the issue of identity of the object is linked to what type of database it is, the IDs can represent either the gene or the protein and can be used in either context; this is consistent with how we already use the IDs
  • David OS - are any databases ambiguous wrt gene, transcript, protein IDs?
  • Suzi - don't know of any
  • David OS - some annotation extension relations have defined domains and ranges and 'inconsistent' use could be problematic
  • Chris - curators did want to use the more precise identifiers from their own database
  • Suzi - this is clearly an issue; stop gap solutions may be expedient, but not correct. The primary issue relates to the regulates question and information was getting lost wrt a process regulating a process or a process regulating an entity. We need to explicitly say, somehow, which of the two we're referring to. Is there another approach we could use?
  • David H. - the has_regulation_target is only one example; denoting substrates using has_input can use gene or protein identifiers
  • Chris - Noctua has formally defined the entity class as a union of gene or gene product
  • David OS - would it make sense to extend what is in LEGO to annotation extensions so that formally a UniProtKB accession could refer to a gene or a protein, an MGI ID could refer to a gene or a protein
  • Chris - Jenkins checks do not check for gene vs protein, but whether the overall type of entity is correct

AI:David OS - can say in the OWL, for example, that the domain is a protein and then in a separate mapping file indicate what IDs can be used to indicate a protein

  • Ruth - for regulation of transcription, curators could then use a UniProtKB ID to indicate the gene? This what not be meant to indicate at what level the regulation occurs.