interferon-gamma-mediated signaling pathway: clearly define start and end

[cmungall]

• id: GO:0060333
• name: interferon-gamma-mediated signaling pathway
• def: "A series of molecular signals initiated by the binding of interferon-gamma to a receptor on the surface of a cell, and ending with regulation of a downstream cellular process, e.g. transcription. Interferon gamma is the only member of the type II interferon found so far."
• is_a: GO:0019221 ! cytokine-mediated signaling pathway
• part_of GO:0071346 ! cellular response to interferon-gamma

The definition does not clearly state when this process starts. All signaling terms should follow a standard template indicating the GO activity that is the first step.

Currently IFNG is annotated to this term (IDA) http://amigo.geneontology.org/amigo/gene_product/UniProtKB:P01579

I don't think it should be, but the mis-annotation is understandable given the start/end isn't clear.

Note that due to the part_of relationship, if we interpret the IDA annotation as involved_in, then IFNG is involved in response to interferon-gamma. In general we do not expect proteins to be involved in responses to their own activities, unless there is an explicit feedback loop.

The annotation relationship here should be an upstream one.

[cmungall]

Note: I encountered this when looking at this paper https://www.nature.com/articles/s41598-018-23395-2 Interpretation of biological experiments changes with evolution of the Gene Ontology and its annotations

where the authors looked at the p-value of response-to-ifng in gene set enrichment over time. I'm not sure it was a good idea to track this term. It would be better to track a term that represents a distinct mechanism, like interferon-gamma-mediated signaling pathway. it's not clear to me why we have both response and the pathway in GO. Clearly not everything involved in the response is part of the pathway but why have the response term in GO at all, except for doing IEPs which lead to circularities...

It looks like there was a massive influx of annotations to RtInG in 2017 from this paper: http://amigo.geneontology.org/amigo/reference/PMID:19144319

All of these are dual annotations to phagosome and response. This seems really unsatisfactory to me. Why were these not conjoined at least with extensions, or even GO-CAMs?

The evidence code (IDA) seems dubious. It seems a bioinformatics analysis based on MS data. In fact the bioinformatics analysis even used GO! But thankfully we didn't cycle that back in..

The annotations seem to have come from table 1, "Changes in Abundance and Phosphorylation of Selected Phagosome Proteins upon IFN-γ Stimulation". This is an IEP at best!

I think it's important for annotators to know that when a term is flooded with dubious annotations, you are sinking that term for enrichment purposes (the p value will be increased for gene sets that do not closely match the genes annotated). Perhaps this is OK if we only do this for response-to terms that don't correspond to mechanisms. But we should really be flagging these response terms to users!

[addiehl]

I absolutely agree that these are inappropriate annotations for a "response to" term, at best IEPs, and should never have been made.

Discussions about the proper use of response to terms go back over a decade, and I'm sorry to see that these terms are still used by some annotators to annotate gene expression patterns or proteins produced following a stimulation, rather than the gene products involved in the signaling pathway or proximal effector activity triggered by the stimulation that occurs prior to de novo gene expression.

[pgaudet]

As far as I know IEP is supporting the vast majority of 'response to' annotation, even when these are made with an IDA (some people consider changes in expression a direct assay for response to, since the definition is "Any process that results in a change in state or activity of a cell or an organism (in terms of movement, secretion, enzyme production, gene expression, etc.) as a result of "

We had planned to make a proposal for this for the Fall 2018 meeting (trying to focus on the action points from the Cambridge meeting right now).

Thanks, Pascale

[RLovering]

Hi signaling pathways include the ligand, and therefore the ligand will be annotated to the response to ligand term. This was agreed in a signaling meeting years and years ago. It is weird but the alternative is to have the ligand regulating the signaling pathway. see General ligand-receptor pathway diagram in http://geneontology.org/page/go-annotation-conventions#general

Also I think the term definition is clear that the pathway starts with the binding of the ligand. ie the activity of the ligand is included in the term definition.

Ruth

[pgaudet]

Hi @cmungall

Looks to me like the start is clear of the process is clear according to the definition: A series of molecular signals initiated by the binding of interferon-gamma to a receptor (and so is the end, according to the standard definitions for pathways).

And as Ruth pointed out, the ligand should be annotated to the pathway according to rules established a long time ago, see http://wiki.geneontology.org/index.php/Signaling_Curation_Manual#What_is_included_in_a_signaling_pathway

What do you think should be done here ? (If we fix the definition, then I suspect that we'll need to edit the definitions of many of these terms - @ukemi can you point the location of the standard definition of x signaling pathway?

Thanks, Pascale

Thanks, Pascale

[ukemi]

http://wiki.geneontology.org/index.php/Signaling