View source for Guidance for updating deprecated Annotation Extension Relations

=General Guidance=

This documentation constitutes general recommendations only for updating annotations that have used a deprecated relation in the annotation extension. 
When updating annotations, it should be kept in mind that the statement in the annotation extension is '''extending the primary GO term annotated''' (in column 4 of the GAF) and also that any information in the extension should be '''physiologically relevant'''.

  The combination of GO term plus Anotation Extension needs to make sense as a class on its own - ignoring the gene product. 
  The result should be a subclass (i.e. effectively a child term) of the primary GO term used. 

Some of the actions to be taken are clear-cut and the following rules will apply;

  IF primary GO ID = GO:CC
  AND extension relation is one of dependent_on, in_presence_of, in_absence_of
  AND extension entity is ChEBI ID
  then delete extension

  IF primary GO ID = GO:CC
  AND extension relation is dependent_on
  AND extension entity is GO:BP
  then change extension relation to exists_during


The remainder of corrections will most likely require the curator to go back to the paper to determine how the entity in the extension is related to the primary GO term annotated (this demonstrates that the meaning of the existing annotation is not clear and should be modified).

If a curator comes across an example that is not covered by this guidance, and it is not clear how the annotation should be updated, the example should be brought to an annotation call to be discussed and resolved.

Additional recommendations have been made for specific annotations on previous annotation calls;

http://wiki.geneontology.org/index.php/Annotation_Conf._Call,_June_23,_2015

http://wiki.geneontology.org/index.php/Annotation_Conf._Call,_July_28,_2015

See the [[Media:Deprecated_AE_Annotations.xlsx|Excel spreadsheet containing details of annotations using deprecated relations, with assigned_by information]]

=Dependent_on=

==ChEBI identifiers==

When dependent_on is used with a ChEBI ID, consider the following;

* Is the chemical in the annotation extension is already mentioned in the GO term definition? If it is, there is no reason to add the chemical in the extension
* If the definition does not mention the chemical, should the definition be updated to include the chemical, i.e. is it always required? 
* If the primary GO term used is a Cellular Component (i.e. the curator was trying to express that a gene product is at a particular location when a specific chemical is present), this is out-of-scope for GO and the extension should be deleted.
* Is the chemical physiologically relevant? For example using a chemical to induce a physiological response: if an author uses hydrogen peroxide to induce oxidative stress then consider making an annotation to a GO term concerned with oxidative stress
* If the chemical only represents an assay condition, delete the extension
* If none of the above applies, consider using a different relation e.g. “activated_by”

'''Example 1'''
  Gene product: UniProtKB:P25632
  GO ID:		GO:0015616 DNA translocase activity
  Extension:	dependent_on(CHEBI:15422 ATP)

'''''Action:''''' ATP is in the definition of the GO term, therefore no need to put ATP in the extension (Catalysis of the reaction: ATP + H2O = ADP + phosphate, to drive movement along a single- or double-stranded DNA molecule.).

'''Example 2'''
  Gene product: UniProtKB:A0A060VXQ0
  GO ID:	GO:0090632 N-glycolylneuraminic acid (Neu5Gc) cytidylyltransferase activity 
  Extension:	dependent_on (CHEBI:18420 magnesium(2+))

'''''Action:''''' In this case the Mg2+ is not in the definition of the GO term. It appears that Mg2+ is essential for this activity, therefore consider re-writing the GO term definition, or alternatively using the “activated_by” relation instead.

==GO identifiers==

When dependent_on is used with a GO ID, consider the following;

* If the GO ID in the extension is a Biological Process or Molecular Function, consider changing the relation to part_of, if you consider the primary GO term used is part of the GO term used in the extension (see example below)
* If the GO ID in the extension is a Biological Process or Molecular Function, consider reversing the annotation, i.e. make the primary annotation to the GO term in the extension and using the other GO term in the extension with either “part_of” or “causally_upstream_of” (see examples below)
* If the primary GO term used is a Cellular Component and the GO ID in the extension is a Biological Process, consider changing the relation to exists_during. (see example below)

'''Example 1'''
  Gene product: UniProtKB: O43474
  GO ID:	GO:0045944 positive regulation of transcription from RNA polymerase II promoter
  Extension:	dependent_on(GO:0000165 MAPK cascade)

'''''Action:''''' Consider changing relation to part_of

'''Example 2'''
  Gene product: UniProtKB: Q60855
  GO ID:	GO:0097527 necroptotic signaling pathway 
  Extension:	dependent_on(GO:0004672 protein kinase activity)

'''''Action:''''' Consider reversing the GO IDs and changing the relation to part_of;

  Gene product: UniProtKB: Q60855
  GO ID:	GO:0004672 protein kinase activity
  Extension:	part_of(GO:0097527 necroptotic signaling pathway)


'''Example 3'''
  Gene product: UniProtKB: Q15485
  GO ID:	GO:0001867 complement activation, lectin pathway
  Extension:	dependent_on(GO: 0003823 antigen binding)

'''''Action:''''' Consider reversing the GO IDs and changing the relation to causally_upstream_of;

  Gene product: UniProtKB: Q15485
  GO ID:	GO:0003823 antigen binding
  Extension:	causally_upstream_of(GO:0001867 complement activation, lectin pathway)

'''Example 4'''
  Gene product: UniProtKB: P07737
  GO ID:	GO:0005938 cell cortex
  Extension:	dependent_on(GO:0006939 smooth muscle contraction)

'''''Action:''''' Consider changing the relation to exists_during.

==Protein identifiers==

When dependent_on is used with a protein ID, consider the following;

* If the protein stated in the extension is required for the Biological Process or Molecular Function described by the primary GO term, the protein in the extension should instead be annotated directly to that BP or MF
* If the protein in the extension acts together with the protein in the primary annotation, consider annotating the Complex Portal ID for the complex directly with the BP or MF.
* If the primary GO term that is annotated is a Cellular Component then consider annotating the protein ID in the extension with “protein localization to [component]”

'''Example 1'''
  Gene product: UniProtKB: Q9C0C7
  GO ID:	GO:0043552 positive regulation of phosphatidylinositol 3-kinase activity
  Extension:	dependent_on(UniProtKB:O60260 Parkin)

'''''Action:''''' In this case, consider whether Parkin should be annotated to GO:0043552 directly. Don’t attempt to put all proteins involved in a pathway into the extension.

'''Example 2'''
  Gene product: UniProtKB: Q9ESD1
  GO ID:	GO:0005615 extracellular space
  Extension:	dependent_on(UniProtKB:O70362 Phospholipase D)

'''''Action:''''' Here it appears that phospholipase D is responsible for localization of Q9ESD1, in which case the annotation should instead be;

  Gene product: UniProtKB: O70362
  GO ID:	GO:0071692 protein localization to extracellular region
  Extension:	transports_or_maintains_localization_of(UniProtKB:Q9ESD1 Prostasin)


==Pfam identifiers==

When dependent_on is used with a Pfam ID, consider the following;

* Is the Pfam domain just indicating the region of the protein being bound? If so, this could be captured using a more specific protein binding term

'''Example 1'''
  Gene product: UniProtKB: O60132 (Tea4)
  GO ID:	GO:0005515 protein binding
  With/from:	PomBase: SPBC776.02c (Dis2)
  Extension:	dependent_on(Pfam:PF00018 SH3 domain)

'''''Action:''''' In this case the paper (PMID:24554432) describes that Tea4 requires it’s SH3 domain in order to bind to Dis2. This information should not be captured in the annotation for Tea4 but instead the annotation for Dis2 should use the GO term GO:0017124 SH3 domain binding with the Tea4 identifier in the with/from field.

==Sequence Ontology identifiers==

When dependent_on is used with a Sequence Ontology ID, consider the following;

* If the SO ID is referring to a region that is being bound the gene product being annotated, consider using the occurs_at relation instead.

'''Example 1'''
  Gene product: UniProtKB: Q9UIF9
  GO ID:	GO:0042393 histone binding
  With/from:	UniProtKB: P62805 (Histone H4)
  Extension:	dependent_on(SO:0001729 H4K16 acylation site)

'''''Action:''''' The extension appears to be describing the region of Histone H4 that is being bound, so consider using occurs_at instead.

=In_presence_of=

==ChEBI identifiers==

When in_presence_of is used with a ChEBI ID, consider the following;

* Whether the chemical in the annotation extension is already mentioned in the GO term definition. If it is, there is no reason to add the chemical in the extension
* If the definition does not mention the chemical, should the definition be updated to include the chemical, i.e. is it always required? 
* If the primary GO term used is a Cellular Component (i.e. the curator was trying to express that a gene product is at a particular location when a specific chemical is present), this is out-of-scope for GO and the extension should be deleted.
* Is the chemical physiologically relevant? For example using a chemical to induce a physiological response: if an author uses hydrogen peroxide to induce oxidative stress then consider making an annotation to a GO term concerned with oxidative stress
* If the chemical only represents an assay condition, delete the extension
* If none of the above applies, consider using a different relation e.g. “activated_by”

'''Example 1'''
  Gene product: UniProtKB: Q12510 (Cmr1)
  GO ID:	GO:0034399 nuclear periphery
  Extension:	in_presence_of(CHEBI:25255 methyl methanesulfonate)

'''''Action:''''' In this case the paper (PMID:25817432) states that Cmr1 localizes to the nuclear periphery in response to methyl methanesulfonate, a DNA damaging agent. This type of information is out of scope for GO, however there is a term GO:0072702 response to methyl methanesulfonate that should be considered for a direct annotation to Cmr1 instead.

==GO identifiers==

When in_presence_of is used with a GO ID (GO:Complex only allowed), consider the following;

* Consider annotating the Complex Portal ID with the GO term instead
* Is the GO:complex present only included as an assay condition? If so, this should not be captured and the extension should be deleted.

'''Example 1'''
  Gene product: UniProtKB: P49757
  GO ID:	GO:0030335 positive regulation of cell migration
  Extension:	in_presence_of(GO:0005587 collagen type IV trimer)

'''''Action:''''' From this paper (PMID:19581412) it appears as if the type IV collagen is included only as a substrate for the cell migration and it is not clear that this is biologically relevant to the study. It would be recommended to delete the extension as it is an assay condition.

'''Example 2'''
  Gene product: UniProtKB: Q96J84 (Neph1)
  GO ID:	GO:0005886 plasma membrane
  Extension:	in_presence_of(GO:0031941 filamentous actin)

'''''Action:''''' In this paper (PMID:21402783), an intact actin cytoskeleton is required for localization of Neph1 to the plasma membrane. Unfortunately this is out-of-scope for GO annotation, therefore the extension should be deleted.

==Protein identifiers==

When in_presence_of is used with a protein ID, consider the following;

* If the protein stated in the extension is required for the Biological Process or Molecular Function described by the primary GO term, the protein in the extension should instead be annotated directly to that BP or MF
* If the protein in the extension acts together with the protein in the primary annotation, consider annotating the Complex Portal ID directly with the BP or MF.
* If the primary GO term that is annotated is a Cellular Component then consider annotating the protein ID in the extension with “protein localization to [component]”

'''Example 1'''
  Gene product: UniProtKB: P40096
  GO ID:	GO:0003713 transcription coactivator activity
  Extension:	in_presence_of(UniProtKB:Q92317)

'''''Action:''''' If the proteins form a complex to perform this activity then each of them should be annotated to the GO term. Consider also annotating the Complex Portal ID for the complex with this GO term. If one of the proteins is regulating the activity they can be annotated to the activity using the contributes_to qualifier.

=In_absence_of=
==Protein identifiers==

When in_absence_of is used with a protein ID, consider the following;

* Processes and activities occur when many proteins are absent, is the statement in the extension physiologically relevant?
* If so, can the protein ID in the extension instead be annotated directly with a negative regulation GO term (or positive regulation if the primary GO term is a negative regulation)?
* Can a new GO term be requested that covers the concept, e.g. some existing terms include "chaperone mediated protein folding independent of cofactor" and "non-canonical Wnt signaling pathway", the latter of which describes a beta-catenin-independent pathway.

'''Example 1'''
  Gene product: UniProtKB: O15455 (TLR3)
  GO ID:	GO:0097527 necroptotic signaling pathway
  Extension:	in_absence_of(UniProtKB:Q13489 cIAP1)

'''''Action:''''' In this paper (PMID:21737330), cIAP1 blocks the TLR3 effect on necroptosis, therefore cIAP1 should be annotated directly to negative regulation of necroptotic signaling pathway. It is appropriate to annotate TLR3 to necroptotic signaling pathway without any extension.

==ChEBI identifiers==

When in_absence_of is used with a ChEBI ID, consider the following;

* If the chemical only represents an assay condition, delete the extension
* Consider using the “inhibited_by” relation instead
* If the primary GO term used is a Cellular Component (i.e. the curator was trying to express that a gene product is at a particular location when a specific chemical is absent), this is out-of-scope for GO and the extension should be deleted.

'''Example 1'''
  Gene product: UniProtKB: P10080 (SBP1)
  GO ID:	GO:0010494 cytoplasmic stress granule
  Extension:	in_absence_of(CHEBI:17234 glucose)

'''''Action:''''' In this paper (PMID:23222640), the authors performed the experiments under glucose deprivation because they wanted the cells to be in a defined stressed state. Therefore this represents an assay condition and so it is recommended that the extension is deleted.

'''Example 2'''
  Gene product: UniProtKB: P10503 (PutA)
  GO ID:	GO:0043565 sequence-specific DNA binding
  Extension:	in_absence_of(CHEBI:26271 proline)

'''''Action:''''' In this paper (PMID:8483946), proline prevents binding of PutA to the ''put'' operator sites, therefore the relation "inhibited_by" should be used instead.

[[Category:Annotation extension]]