Annotation Conf. Call 2016-01-12

From GO Wiki
Jump to: navigation, search

Agenda

Bluejeans

Annotation Consistency Exercises

  • Will continue monthly exercises for 2016
  • This year, we will add discussion of LEGO representation to the calls, and compare conventional GO annotations to LEGO representations.
  • Recent News on LEGO Jamboree

AI: Confirm rota schedule.

Ontology Updates

AI:

Annotation from Abstracts Only

  • From ontology meeting minutes: We thought that annotations from abstracts only (where the full text of the paper is in non-English language) were discouraged. Looks like this may not be the case for some MODs. But then, how can people verify, a posteriori, correctness of annotations?
  • We would like to review what MODs do wrt annotating abstracts to make sure we are all on the same page with this.
  • Are there any circumstances where we think this is warranted?

AI:

Deprecated Annotation Extensions

  • Rachael and Ruth have put together documentation on the proposed deprecated extensions along with recommendations for revision.

AIs:

Please document here examples of where you have questions about re-annotation of annotations with deprecated extensions.

dependent_on

Example 1

  • Co-transfection experiment that used dependent_on relation to capture requirement of both proteins to effect a GPCR signaling pathway
  • Recommendation: make direct annotations to each gene product to avoid 'reverse' annotations and use contributes_to qualifier
    • Two chemoreceptors mediate developmental effects of dauer pheromone in C. elegans.
    • To investigate whether expression of SRBC-64/66 was sufficient to confer pheromone responses in a heterologous context, we expressed one or both receptors in HEK293 cells and monitored intracellular second messenger levels upon the addition of ascarosides. Co-expression of SRBC-64 and -66, but not each receptor alone, enhanced forskolin-mediated increases in cAMP levels (Fig. 4A, B), and this increase was significantly inhibited upon addition of C6 and C9 at different concentrations (Fig. 4A-C).
    • Old annotation (one of four annotations made):
 Gene product: SRBC-64 (UniProtKB:G5ECV4)
 GO ID:	G-protein coupled receptor activity (GO:0004930) 
 GO evidence:  IDA
 Extension: dependent_on SRBC-66 (UniProtKB:O61938) 
  • Possible new annotation:
 Gene product: SRBC-64 (UniProtKB:G5ECV4)
 GO qualifier: contributes_to
 GO ID: G-protein coupled receptor activity (GO:0004930)
 GO evidence:  IDA

In this case, I will lose the information that the activity is only seen in the presence of SRBC-66, and I don't have evidence that the two receptors act as a protein complex.

  • Possible new annotation:
 Gene product: SRBC-64 (UniProtKB:G5ECV4) 
 GO qualifier: contributes_to
 GO ID: G-protein coupled receptor activity (GO:0004930)
 GO evidence:  IGI
 With: SRBC-66 (UniProtKB:O61938)

Should this instead be an IGI annotation? Ruth comment: based on the experiment both proteins appear to contribute to this activity and therefore both proteins should be annotated, thus in addition to the annotation suggested above also add:

 Gene product: SRBC-66 (UniProtKB:O61938)
 GO qualifier: contributes_to
 GO ID: G-protein coupled receptor activity (GO:0004930)
 GO evidence:  IGI
 With: SRBC-64 (UniProtKB:G5ECV4)

Kimberly comment: Yes, absolutely. I was only showing one of the annotations to illustrate.

[Rachael] If we agree that this is OK as an IGI annotation, we can use this example to update/clarify the IGI documentation. Currently the documentation says "Includes any combination of alterations in the sequence (mutation) or expression of more than one gene/gene product." This could be interpreted as over-expression (i.e. which would be IDA ordinarily, but for two genes could be IGI).

Example 2

  • Co-transfection experiment demonstrating need for a tethering molecule to facilitate enzymatic activity
  • Recommendation: make direct annotations to avoid 'reverse' annotations and use contributes_to qualifier
    • LIN-28 and the poly(U) polymerase PUP-2 regulate let-7 microRNA processing in Caenorhabditis elegans.
    • PUP-2 was previously shown to polyuridylate an artificially tethered RNA in Xenopus oocytes but was inactive without tethering. Therefore, we tested whether LIN-28 might be able to recruit PUP-2 to mediate pre–let-7 uridylation (Fig. 3c). We incubated anti-Flag immunoprecipitates from cell extracts expressing HA-Flag–PUP-2 and/or SBP–LIN-28 with radiolabeled pre–let-7 and radiolabeled UTP. HA-Flag–PUP-2 uridylated pre–let-7 only in the presence of SBP–LIN-28 (Fig. 3c). We also confirmed LIN-28–dependent uridylation of pre–let-7 by PUP-2 in vitro (Supplementary Fig. 6b).
  • Old annotation:
 Gene product: PUP-2 (UniProtKB:Q09408)
 GO ID: RNA uridylyltransferase activity (GO:0050265)
 GO evidence: IDA
 Extension: dependent_on LIN-28 (UniProtKB:P92186), has_direct_input let-7 (WBGene00002285)
  • Possible new annotations:
 Gene product: PUP-2 (UniProtKB:Q09408)
 GO ID: RNA uridylyltransferase activity (GO:0050265)
 GO evidence: IDA
 Gene product: LIN-28 (UniProtKB:P92186)
 GO qualifier: contributes_to
 GO ID: RNA uridylyltransferase activity (GO:0050265)
 GO evidence: IDA
 Gene product: PUP-2 (UniProtKB:Q09408)
 GO ID: protein complex (GO:0043234)
 GO evidence: IDA
 Gene product: LIN-28 (UniProtKB:P92186)
 GO ID: protein complex (GO:0043234)
 GO evidence: IDA
  • OR
 Gene product: PUP-2 (UniProtKB:Q09408)
 GO ID: RNA uridylyltransferase activity (GO:0050265)
 GO evidence: IGI
 With: LIN-28 (UniProtKB:P92186) 
 Gene product: LIN-28 (UniProtKB:P92186)
 GO qualifier: contributes_to
 GO ID: RNA uridylyltransferase activity (GO:0050265)
 GO evidence: IGI
 With: PUP-2 (UniProtKB:Q09408)

Ruth Comment: Looking at the abstract, key points here are: LIN-28 directly binds let-7 pre-miRNA to prevent Dicer processing. PUP-2 and LIN-28 interact directly, and that LIN-28 stimulates uridylation of let-7 pre-miRNA by PUP-2 in vitro. I would not use contributes_to, although I agree I do wonder how we decide when something contributes to the function and when it is an 'activator' of the function.

To annotate LIN-28 I would use the following GO terms:

  • GO:0051347 positive regulation of transferase activity (or more specific new term negative regulation of RNA uridylyltransferase activity) has_direct_input PUP-2 (UniProtKB:Q09408), part_of GO:2000632 negative regulation of pre-miRNA processing
  • GO:0008047 enzyme activator activity (or more specific new term transferase activator activity or more specific new term RNA uridylyltransferase activator activity ) has_direct_input PUP-2 (UniProtKB:Q09408), part_of GO:2000632 negative regulation of pre-miRNA processing

New GO term positive regulation of pre-miRNA uridylation has_direct_input let-7 (WBGene00002285), part_of GO:2000632 negative regulation of pre-miRNA processing

  • GO:0070883 pre-miRNA binding IPI let-7 (WBGene00002285), part_of GO:2000632 negative regulation of pre-miRNA processing
  • GO:0019899 enzyme binding IPI PUP-2 (UniProtKB:Q09408), part_of GO:2000632 negative regulation of pre-miRNA processing
  • GO:2000632 negative regulation of pre-miRNA processing has_direct_input let-7 (WBGene00002285)

[Rachael] Would it also be appropriate to annotate to a term such as "miRNA binding, bridging" (which doesn't exist but would be equivalent to the protein binding, bridging term)? - in the extension of this annotation you could then have causally_upstream_of RNA uridylyltransferase activity/positive regulation of pre-miRNA uridylation/negative regulation of pre-miRNA processing.

To annotate PUP-2 I would use the following GO terms:

  • a uridylation BP term, haven't read the paper so don't know if GO:0071076 RNA 3' uridylation is too specific? Or new GO term pre-miRNA uridylation, or stick with the more general new GO term: RNA uridylation, IDA, has_direct_input let-7 (WBGene00002285), part_of GO:2000632 negative regulation of pre-miRNA processing
  • RNA uridylyltransferase activity (GO:0050265); IDA, has_direct_input let-7 (WBGene00002285), part_of GO:2000632 negative regulation of pre-miRNA processing
  • GO:2000632 negative regulation of pre-miRNA processing has_direct_input let-7 (WBGene00002285)
  • GO:0005155 protein binding IPI LIN-28 (UniProtKB:P92186), part_of GO:2000632 negative regulation of pre-miRNA processing
  • GO:0070883 pre-miRNA binding IPI let-7 (WBGene00002285), part_of GO:2000632 negative regulation of pre-miRNA processing

Also I wonder if it should be protein-RNA complex rather than protein complex?

[Rachael] I agree with these comments.

But wrt decision on IDA v IGI I am not sure what I would do, prefer IGI if this is agreed.

[Rachael] Agree

Example 3

  • Enzymatic activity requires presence of another protein that acts as an upstream activator
  • Recommendation: make direct annotations to avoid 'reverse' annotations and use contributes_to qualifier
    • Since SGK-1, AKT-1, and AKT-2 all contain a highly similar kinase domain, we tested if they are able to phosphorylate their candidate substrate DAF-16 in vitro. For this purpose, we constructed transgenic wild-type and pdk-1(sa680) mutant strains expressing gfp-tagged akt-1, akt-2, or sgk-1. The fusion proteins were immunopurified from worm extracts and used for kinase assays with purified DAF-16. Both AKT-1, AKT-2, and SGK-1 purified from a wild-type background phosphorylated DAF-16 ( Figure 4A). In contrast, all three kinases purified from a pdk-1 mutant background were not able to phosphorylate DAF-16 (Figure 4A). This demonstrates that functional PDK-1 is required to activate AKT-1, AKT-2, and SGK-1 in vivo.
    • In pull-down experiments, we found that both in vitro-translated PDK-1 and DAF-16 interact with bacterially expressed, GST-tagged SGK-1, AKT-1, and AKT-2.
  • Old annotation (one example):
 Gene product: AKT-1 (UniProtKB:Q17941)
 GO ID: protein kinase activity (GO:0004672)
 GO evidence: IDA
 Extension: has_direct_input DAF-16 (UniProtKB:O16850),dependent_on PDK-1 (UniProtKB:Q9Y1J3)
  • New annotations:
 Gene product: AKT-1 (UniProtKB:Q17941)
 GO ID: protein kinase activity (GO:0004672)
 GO evidence: IDA
 Extension: has_direct_input DAF-16 (UniProtKB:O16850)
 Gene product: PDK-1 (UniProtKB:Q9Y1J3)
 GO ID: protein kinase activator activity (GO:0030295) 
 GO evidence: IMP
 Extension: has_direct_input AKT-1 (UniProtKB:Q17941)

Ruth comment: yes I agree with the new annotations. Also could annotate PDK-1 as GO:0001934 positive regulation of protein phosphorylation has_regulation_target or more specifically regulates_o_has_input DAF-16 (UniProtKB:O16850).

[Rachael] Agree

Example 4

  • Protein interaction that only occurs in the presence of a specific DNA sequence
  • Recommendation: use occurs_at
    • The Caenorhabditis elegans JNK signaling pathway activates expression of stress response genes by derepressing the Fos/HDAC repressor complex.
    • Next we asked whether FOS-1 could interact with HDA-1. T7-FOS-1 and FLAG-HDA-1 were co-expressed in HEK293 cells. We immunoprecipitated FLAG-HDA-1 with anti-FLAG antibodies, and probed for the T7-FOS-1 on a Western blot with anti-T7 antibodies. We failed to detect an association between FOS1- and HDA-1 (Figure 7C, lane 1). However, if we transfected in the Pkreg-1::venus reporter along with T7-FOS-1 and FLAG-HDA-1, we could detect an association between FOS-1 and HDA-1 (Figure 7C, lane 3). Furthermore, removal of the TRE2 site from the Pkreg-1::venus reporter reduced this interaction (Figure 7C, lane 4). These results suggest that HDA-1 and FOS-1 can associate on the kreg-1 promoter.
  • Old annotation (one of two reciprocal annotations made):
 Gene product: FOS-1 (UniProtKB:G5ECG2)
 GO ID:	histone deacetylase binding (GO:0042826)
 GO evidence:  IPI
 With/from:	HDA-1 (UniProtKB:O17695)
 Extension:	dependent_on RNApol_II_promoter (SO:0000170),dependent_on kreg-1 gene (WB:WBGene00018725)
  • New annotation:
 Gene product: FOS-1 (UniProtKB:G5ECG2)
 GO ID:	histone deacetylase binding (GO:0042826)
 GO evidence:  IPI
 With/from:	HDA-1 (UniProtKB:O17695)
 Extension:	occurs_at RNApol_II_promoter (SO:0000170)
 Gene IDs are not currently valid for occurs_at relation, so cannot add the specific gene. 
 Only GO and SO IDs are currently valid.

[Rachael] I recently asked David OS for the ability to add Gene IDs to coincident_with. We should ask him whether it is appropriate to use a gene ID with occurs_at.

Minutes

  • Present:
    • Berkeley - Chris, Suzi
    • dictyBase - Petra
    • EBI - Aleks, Alex, Eleanor, Melanie, Paola
    • MGI - David H., Li
    • PomBase - Antonia, Midori
    • SGD - Edith, Stacia
    • TAIR - Tanya
    • UCL - Barbara, Rachael, Ruth
    • WB - Kimberly
    • Zfin - Sabrina

Annotation Consistency Exercises for 2016

  • We will continue to do them in 2016 adding LEGO curation for each paper.
  • The people who attended the Geneva LEGO meeting in December 2015 will present first and present both the conventional approach and the LEGO approach.
  • As people become more comfortable with LEGO representation they should feel free to contact Seth and get a log in to do a LEGO model. At the beginning we expect that most people will do conventional annotation, but eventually people will try to do LEGO.
  • The LEGO documentation is available on the Noctua landing page. Scroll down to the Quick Start section for appropriate links.
  • We are still working out some bugs with respect to generating GAFs from the LEGO models. We might want to look at the GAFs on later calls.

Ontology Issues

Non-physiological Response Terms

  • Please do not request terms that represent response to non-physiological processes such as response to drugs (e.g., intercalating agents) or toxins that are used as experimental reagents.
  • These terms have been added to enable cross-referencing with other ontologies, such as the Experimental Factor Ontology (EFO).
  • In the pombe Phenotype Ontology, there are cross-references to some of these terms but it should not contradict the rule against using these to annotate genes. These x-refs are used for terms where they wouldn't want to make a direct GO annotation anyway.
  • Kimberly raised the issue that our documentation specifies that GO is used to annotate gene products. Do we want to add something to the documentation that explains why these terms are in the ontology?

AI: The ontology editors will tag the existing terms as Do Not Annotate and add a comment as to why these terms are not appropriate for annotating gene products.

AI: Tanya will obsolete response to microwave radiation.

AI: We should add to the GO documentation on the web site to explain these other uses.

Annotating from Abstracts of Peer-Reviewed Papers

  • This issue specifically refers to a paper that was not published in English, so the annotations would be from the English-language abstract.
  • A related issue extends to the use of solely using Figures for annotations if the Figure Legends are in English.
  • Agreed: We should never annotate from just an abstract.
  • Figures and Figure Legends, however, are much more than just an abstract.
  • Agreed: English language papers are the preferred source of GO annotations. The reason for this is so that other curators, ontology editors, and our users (who primarily speak English in a professional context) can go back and look at the original source of the annotation, if needed.
  • Agreed: If there is no other source for information than a non-English language paper, it is acceptable to annotate that paper as long as you have access to the data from the paper, either from Figures and Figure Legends, or you are a native speaker of the language in which the paper is written. The key is that the curator have sufficient understanding to confidently annotate the paper.

AI:Update documentation on web site to reflect this policy.

Deprecated Relations for Annotation Extensions (AEs)

  • How many people have had a chance to look at the violations?
  • Kimberly, Midori, UCL, SGD are working on them, MGI hasn't done them yet, EBI has done part of theirs, TAIR didn't have any.

dependent_on

  • Cotransfection experiments
  • Experiment required presence of two different GPCRs to see evidence of signaling
  • Originally annotated each GPCR to the appropriate MF and BP terms, using dependent_on as the AE relation and the other GPCR as the AE entity
  • Simply removing the AE and using the IDA evidence code plus a contributes_to qualifier, doesn't capture that the experiment shows both are needed
  • Alternative solution is to use the IGI evidence code for co-transfection experiments and make reciprocal annotations to each of the GPCRs
  • The evidence code for co-transfections has been under discussion before, but MGI has done it this way, Ruth has done it this way, Kimberly did it this way.
  • The key here is that the author is looking at a physiological interaction between the two genes.
  • Single transfections that assess physiological roles will still be annotated using the IDA evidence code.
  • Single transfections for which the authors acknowledge a likely effect of overexpression will use the IMP evidence code.
  • Transfection of more than one entity to assess if these genes/gene products functionally interact will use the IGI evidence code.

AI:Update the IGI evidence code documentation with specific examples to reflect this policy.