Annotation Conf. Call, July 28, 2015
- 1 Agenda
- 2 Overview by ontology editors
- 2.1 Example papers
- 2.2 1. Obsoleted relationship:localization_dependent_on
- 2.3 2. Obsoleted relationship:requires_substance
- 2.4 3. Obsoleted relationship:independent_of
- 2.5 4. Obsoleted relationship:independent_of
- 2.6 5. Several examples here for Obsoleted relationship:dependent_on
- 2.7 6. in_absence_of
- 2.8 2 examples for in_presence_of
Overview by ontology editors
- Nothing has been obsoleted, even if QuickGO webservice tests and their usage by Protein2GO have given that impression. The details of these tests and how they are used by protein2GO and GOA have [only just been documented](https://github.com/geneontology/annotation_extensions/wiki#checking-pipelines). We need to work with Tony S to provide a mechanism for suspending usage of potentially problematic relations pending review.
- Annotation extension relationships are meant to extend the meaning of a GO term to create a new, more restricted class. Where the relation in question is used in GO, or is related logically to one that is, OWL translation of annotation extensions can be used to group annotations more accurately and potentially in useful novel ways (for example, grouping annotations to process term by the cell-type in which they occur). A small number of relations appear instead to qualify when the relationship between a GP and a GO term applies, (e.g. dependent_on, independent_of, in_presence_of, in_absence_of). OWL translation/interpretation of the resulting annotation extensions is problematic.
From reviewing usage of these relations, some are used with extremely variable semantics it seems there may be other more exotic uses (see review of [dependent_on usage](https://github.com/geneontology/annotation_extensions/issues/17).)
Given these issues, it is important to review usage of these relations to discuss what we are trying to achieve with respect to whether/how the annotations can be integrated into tools that support our users needs.
1. Obsoleted relationship:localization_dependent_on
DSF2 relocalizes from budneck to cytoplasm upon DNA replication stress
We would capture this as: cytoplasm, localization_dependent_on: GO:0006974 (cellular response to DNA damage stimulus) or
cytoplasm, in_presence_of: CHEBI: MMS/Hydroxyurea
(there are few papers in yeast where they have show localization changes in response to certain stimuli)
[Response from Rachael] Here I would say cytoplasm exists_during cellular response to DNA damage stimulus
2. Obsoleted relationship:requires_substance
ISC1 has inositol phosphosphingolipid phospholipase activity which is dependent on phosphatidylserine. We captured it with 'requires_substace' (in_presence_of would worked too). Both these relations are slated for obsoletion.
[Response from Rachael] Phosphatidylserine is the cofactor here, should we be capturing cofactors? If so, how? Would activated_by be the correct relation (Identifies a chemical substance that increases the activity of the gene product)?
3. Obsoleted relationship:independent_of
Typically mitophagy is dependent on PARK2 / parkin (Uniprot O60260). However, in PMID:25215947, authors show that AMBRA1 (Q9C0C7) (in presence of chemical inducers of mitochondrial depolarisation such as FCCP) initiates mitophagy independent of PARK2 / parkin. What to do now that independent_of relation is obsoleted? Simplest to create new term instead? e.g. “parkin-independent mitophagy in response to mitochondrial depolarization”.
NB Examples of parkin-independent mitophagy:
PMID: 25349190 "Parkin-independent mitophagy requires Drp1 and maintains the integrity of mammalian heart and brain."
PMID: 24176932 "Loss of iron triggers PINK1/Parkin-independent mitophagy."
Paul Denny, UCL Functional Annotation Group.
[UCL discussion]In order to not create 'reverse annotations' (as agreed in Bar Harbor), UCL agreed to request the GO term parkin-independent mitophagy. This would not be a suitable approach for all cases, e.g. not just to capture a specific chemical. The new terms would only be created if it has been demonstrated that a major pathway exists that is different from the typical/common pathway, e.g. equivalent to canonical and non-canonical wnt signaling. We should not be trying to capture assay conditions, curators should try to understand the role of the chemical/protein in the physiological process. If the entity is a gene product, curators should annotate this in the forwards direction, i.e. use the gene product as the primary entity being annotated to the process. See example 4 below for when we would suggest it is appropriate to capture chemical dependency within a GO term.
DOS: I'm happy to create a class for Parkin independent mitophagy
4. Obsoleted relationship:independent_of
ISW1a complex binds to nucleosome independent of ATP. We annotated to the subunits of ISW1a to contirbutes_to nucleosome binding and included the ATP independent detail in col-16.
[Response from Rachael] Could we have a new term ATP-independent nucleosome binding? We have several ATP-dependent terms, suggesting there are ATP-independent equivalents?
5. Several examples here for Obsoleted relationship:dependent_on
i) PMID: 12482963
Both Isw1a and Isw1b complexes show nucleosome-dependent ATPase activity. How do we capture the nucleosome dependency part. Both dependent_on and in_presence_of are slated for obsoletion.
[Response from Rachael] I'm not sure I would capture this. The complexes interact with nucleosomes to restructure the chromatin, so I think there are other terms you can use to describe this. Just skimming the paper, there are several annotations you could make (note: I haven't read in detail, so these are just thoughts depending on the evidence);
a. nucleosome binding part_of ATP-dependent chromatin remodeling (or part_of nucleosome positioning)
b. ATPase activity part_of nucleosome positioning
c. ATPase activity part_of nucleosome mobilization
ii) examples on the wiki how to use the dependent_on relationship. These examples were created to guide curators, if the relationship is being obsoleted it is important to give guidance on how this information can be captured, or state that it is out of the scope of GO.
IAP2-dependent regulation of NF-kB activity:
Statement from paper:
To test whether RIP3 and RIP4 have to be ubiquitinated by cIAP1/2 in order to mediate NF-kB activation, we compared RIP-mediated NF-kB luciferase reporter activity when ectopically expressed in HEK293T cells in the presence or absence of the IAP inhibitor BV6, a treatment that induces rapid auto-ubiquitination and degradation of endogenous cIAP1/2. As shown in Figure 4A, BV6 treatment greatly impaired TNF and RIP1 RIP4-induced NF-kB activation but had no impact on TAK1-mediated NF-kB induction (Figure 4A). Those results, which indicate that cIAP1/2 act upstream of TAK1, are consistent with a role for cIAP1/2 as E3 ligases regulating RIP14-mediated activation of NF-kB.
annotation: nucleus in_absence_of cAMP
Title of paper: Nucleocytoplasmic Shuttling of a GATA Transcription Factor Functions as a Development Timer. “We report here that GtaC exhibits rapid nucleocytoplasmic shuttling in response to periodic cAMP waves.” p1329
COMMENT: This was annotated in conjunction with a current ‘cytosol in_presence _of cAMP’ annotation. No other annotations would describe the shuttling as well as these two annotations together and in further conjunction with the annotation: regulation of aggregation involved in sorocarp development dependent_on cellular protein localization
NOTE: we already deleted this annotation as it was flagged in P2GO, but I really disliked deleting that one.
[Response from Rachael] I would capture this annotation as nucleus exists_during 'cellular response to cAMP' (or even 'cAMP-mediated signaling'?) and a second annotation to cytosol exists_during 'cellular response to cAMP'. For your second annotation, I believe what this is interpreted as is: gtaC is involved in regulation of aggregation involved in sorocarp development and this is dependent on the process of cellular protein localization - I don't think this is what you intended? I think what you are trying to say is the regulation of aggregation... is dependent on where gtaC is localized? If so, I would capture this as gtaC is involved in regulation of aggregation... occurs_in cytosol or nucleus (whichever is applicable).
2 examples for in_presence_of
7. in_presence_of example Q550R2 ctxB
annotation: positive regulation of gene expression has_regulation_target carA, acaA, rasG,rasC, in_presence_of cAMP
FIGURE 6: Inhibition of cAMP-activation of Ras in ctxA−/B− cells. (A) cAMP activation of RasC and G.
FIGURE 5: Cortexillin-null cells have delayed and diminished expression of cAR1 and ACA. Suspensions of WT and cortexillin-null cells in starvation buffer were pulsed with cAMP…. ctxA− and ctxB− cells had delayed and reduced expression of ACA and cAR1, and ctxA−/B− cells had almost no expression of either ACA or cAR1.
COMMENT: Even though in Fig. 6 authors mention ‘cAMP activation, I like in_presence_of’ as it implies the correct thing; dependent_on implies that the regulation of the gene expression is dependent of cAMP, but it’s really dependent on the mutated gene(s) shown in this study. The addition of cAMP is mimicking natural starvation induced behaviour and in the assay they used cAMP. A more interpretive annotation would be:
positive regulation of gene expression has_regulation_target carA, acaA, rasG,rasC, happens_during response to starvation. Here the user must know what happens during starvation and it is more my interpretation than annotating what actually happens where users can interpret.
[Response from Rachael] If the authors are studying starvation but using cAMP to mimic this you should really add the starvation term to the extension, as this is the physiological process. In this case I would annotate to positive regulation of gene expression has_regulation_target x,y,z (,) part_of cellular response to starvation. This is saying (I think!) that as a result of starvation, ctxB regulates the expression of x,y,z. I'm not sure if there is an easy way to determine when to use part_of and when happens_during, but I would expect some regulation of gene expression to occur as a result of starvation, therefore I would say reg. of gene expression is part of starvation, rather than it occurs at the same time as starvation (in which case it would be happens_during).
8. in_presence_of example Q54BD4 dstC
annotation: response to cation stress happens_during sorocarp development in_presence_of potassium chloride
There are four such annotations made for all 4 tested stress factors that resulted in developmental defects.
p.793: we investigated the development under stress of the dstC− (STATc) mutant. The dstC− mutant showed severe defects with fewer and/or smaller fruiting bodies under all of the stressing conditions (KCl, NaCl, MgCl2, LiCl) FIGURE 6. TacA is involved in the stress response during development.
COMMENT: Here it feels not appropriate to replace in_presence_of with ‘dependent_on’ and I would have lots of gripes against using ‘has_input’. Not clear what ‘has_agent’ means, never used that and wouldn’t like it either in this context.
Petra Fey, dictyBase
[Response from Rachael] Here it sounds like you would want a new term 'response to potassium cation stress', in which case I think you should use has_input so that the OWL interpretation of a new term would be correct - but an editor should confirm this.