Annotation Conf. Call 2017-02-28

From GO Wiki
Jump to: navigation, search

Bluejeans URL

https://bluejeans.com/993661940

Agenda

GO Meeting Reminder

Noctua Modeling Discussion

Sabrina Toro, Zfin

Midori Harris, PomBase

  • myo2_rlc1_PMID19570908
  • Regulation of fission yeast myosin-II function and contractile ring dynamics by regulatory light-chain and heavy-chain phosphorylation.
  • In vitro and in vivo assays using nonphosphorylatable or phosphomimetic Rlc1 mutations show that Rlc1 somehow positively regulates the actin filament-based motor activity of Myo2 as part of actomyosin contractile ring contraction, in turn part of mitotic cytokinesis.
    • We were very tempted to say that Rlc1's unknown MF regulator activity directly positively regulates Myo2's microfilament motor activity because:
      • Rlc1 has previously been shown to bind Myo2;
      • The in vitro assays (Fig.1, Table 3) show an effect on Myo2 motor activity with only F-actin, Myo2, Cdc4. and Rlc1 (heavy chain, essential light chain, and regulatory light chains respectively) present.
    • The reason we didn't, and instead included the actin filament-based movement "link", is that the is_a hierarchy has motor activity as a subtype of NTPase activity:
     GO:0017111 ! nucleoside-triphosphatase activity
     -- is_a GO:0003774 ! motor activity
     ---- is_a GO:0000146 ! microfilament motor activity
 
  • From this, reasoning would infer this regulation hierarchy (whether regulation terms are instantiated or not):
     regulation of nucleoside-triphosphatase activity
     -- is_a regulation of motor activity
     ---- is_a regulation of microfilament motor activity
 
  • ... and erroneously conclude that Rlc1 does regulate Myo2's ATPase activity. The Rlc1 mutant phenotypes here say it doesn't (Fig. 76, Table 3).
  • Questions:
    • Should the MF ontology change to motor activity has_part nucleoside-triphosphatase activity?
    • Does LEGO have any way to capture when a molecular mechanism isn't known, but one possibility has been ruled out?

April Discussion

Minutes

  • On call: Alice, Barbara, Chris G, David H, David OS, Edith, George, Giulia, GOA, Harold, Helen, Karen, Kimberly, Li, Midori, Pascale, Paul, Ruth, Sabrina, Stacia, Stan, Tanya, Terry, Petra

Noctua Modeling

Sabrina (Zfin) - role of TFs in neutrophil maturation

    • Two TFs, C-myb and Cebp1
    • Represent TFs with MF term for transcription factor activity, sequence-specific DNA binding
    • Indicate gene promoter bound with has_input (lyz)
    • occurs_in:
      • chromatin - based on ChIp assay
    • part_of:
      • transcription - based on direct assays and mutant phenotypes
    • transcription factor complex
      • based on co-IP of C-myb and Cebp1
    • There is no information about the precise role of lyz in neutrophil maturation
    • Transcription part_of cell maturation
      • Based on mutant phenotype
    • Neutrophil maturation didn't exist in GO
      • Created term by linking cell maturation 'results in maturation of' neutrophil
      • Evidence was direct assay based on what the authors show about this process
    • Questions:
      • Harold - transcription factor complex - why is there a separate individual for the complex as well as both each TFs separately?
        • Sabrina - this is what the data suggested
      • Karen - can use the more specific RNA polII transcription factor activity?
        • Sabrina - Yes, could do this.
      • Kimberly - Might be a nice check to have to prompt curators to use the more specific MF term when the more specific BP term is also used.
      • David H - What would the reasoner give if we run it?
      • Paul T. - The coming design templates should include transcription and will give curators a standard way to annotate these things. Curators will then just need to fill in the specific entities and evidence.
      • Tried running the reasoner
      • David OS - discussion about use of 'directly activates'
        • In this case, 'directly activates' feels right, but we need guidelines
      • Karen - there are guidelines for this as part of the TF annotation guidelines
      • Paul T. - 'direct' means a direct physical interaction between two entities that execute the functions
      • David OS - need guidance about when to use 'directly positively regulates' between an MF and a BP
      • Karen - does this suggest that the current transcription curation guidelines are too narrow?
      • Paul T. - this representation of transcription was arrived at after looking at several different possibilities and was thought to be the most intuitive representation. We need to make sure we have a standard representation, though. Even if this is ultimately not the way we want to go, we need to have a standard view.
      • David OS - we also need to make sure we align the representation of transcription in Noctua models is the same as in the ontology

AI: Make a ticket wrt consistent representation of transcription in the ontology and Noctua. - David making now.

Midori - Regulation of Myosin Activity

  • Examining the role of the myosin regulatory light chain - what activities can be associated with it?
  • Role of myosin in actomyosin contractile ring contraction is well studied, accepted
  • Paper looks at rlc1 mutants and the subsequent effects on myo2 activity
  • Some unknown regulatory activity of rlc1 regulates the motor activity of which myo2 is a part
  • But we thought about representing this as rlc1 directly regulates myo2
    • Why? - It is known that rlc1 binds myo2 and rlc1 can affect myo2 activity in in vitro assays when the only participants are the myosin complex and actin filaments
  • The ontology classifies motor activity as a type of nucleoside triphosphatase activity
    • If we said that rlc1 directly regulates myo2, then we would expect the reasoner to also say that rlc1 regulates the nucleoside triphosphate activity
    • However, this would be wrong, because the paper shows that rlc1 doesn't change the myo2 ATPase activity
  • Questions:
    • Is the representation of motor activity in the ontology correct?
      • No, it is a complex molecular function that needs to be represented as such in the ontology.
    • Can we use a NOT annotation to indicate that the data doesn't support regulation of the ATPase activity?
    • Karen - can you use PRO representation of modified forms to indicate that role of the specifically phosphorylated rlc1?
    • Does Noctua allow NOT annotations?
      • No, but there is a ticket for this.
    • Paul T - how deep into the mechanism do we want to go?
      • Could say that the regulation is direct, use whatever evidence is available to support that.
      • Then, if we know what subfunction is actually regulated, we could capture that in GO.
    • Midori - concern with saying that the regulation is direct is that, at present, we'd wind up with an incorrect inference due to the current placement of motor activity in the ontology
    • Paul T - would make the annotation, don't put the model into production, and then put in a ticket to correct the ontology
    • Midori - PomBase would probably wait on this right now
    • Paul T - biological processes are programs made up of molecular activities
    • David OS - can fix the ontology for this
      • We need a generic fix for functions where ATP is used as an energy source
      • How would we describe the effector activity of the motor?
    • David H - this issue of incorrect inferences from compound molecular functions has been around for a while, and the MF refactor should help address it

AI: Add this example to the github repository for MF refactor.

    • Tanya - Bigger picture: do we want to capture everything we know about a set of gene products in a model? The direct binding of rlc1 and myo2? Where does that go, if anywhere? Or do we want to keep models ‘clean’ and showing only one process?
    • Stacia - what is sustainable for Noctua curation? If you can import everything from a Noctua model, that might make sense, otherwise it becomes difficult to view and understand.
    • Tanya - Where would the known binding between myo2 and rlc1 be included in this model?
    • Kimberly - Use binding information as supportive evidence for the direct regulatory relationship?
    • Midori - Could represent the activity as being enabled by the complex and articulate the activities of each member of the complex?

Next Meetings

  • March 14th - regular annotation group call
  • April 11th - regular annotation group call
  • We will not have a call on March 28th, as it overlaps with ISB meeting at Stanford.