gocentral
commented
16 years ago - assigned_to: nobody --> jenclark
Original comment by: mah11
Closed gocentral closed 9 years ago
gocentral
commented
16 years ago Original comment by: mah11
gocentral
commented
16 years ago Ruth,
I have long argued for limits on the use of "response to" terms. I prefer to reserve "response to" terms for the gene products involved in the actual detection, signal transduction, gene transcription, or (occasionally) the effector mechanisms that take place in response to a particular stimulus, and not simply for any mRNA or gene product that varies in its abundance following a stimulus. This limits my "response to" annotations to those gene products that actively interpret a stimulus and provide a clearer interpretation of what "response to" annotations mean.
The alternative approach of annotating any gene product that varies in its expression level following a stimulus (as measured, for instance, by microarray) would allow huge numbers of genes to be annotated in many high-throughput experiments without providing any clear understanding of what those genes contribute in the process of responding. The ultimate result of annotating changes in expression or protein abundance alone to "response to" terms is a mass of annotations that provide little of value to the end user, and destroy the utility of using the annotated GO terms for finding the gene products that genuinely are involved in interpreting a stimulus, rather than those that are generically involved in processes such as cell proliferation or division.
Thanks,
Alex
Original comment by: addiehl
gocentral
commented
15 years ago Not sure if this is the right item for what David was talking about "response to" issue. But I agree with Alex. Only gene products that are involved in that process should be annotated to the "response to" terms.
Just because a gene's transcription level is increased/decreased when exposed to something does not show involvement. The change in expression level just indicate that the gene RESPONDS to the stimulus, not that the subsequent gene product is involved.
Original comment by: eurie
gocentral
commented
15 years ago Hi
I completely agree with the comments below, which relate to NOT including microarray data under the response_to ... GO terms. Which is why I would really like to see the IEP GO evidence code documentation changed.
IEP: Inferred from Expression Pattern
In Reference Genome discussions I have certainly been given the impression that at least one database is in theory happy to associate response_to terms to microarray data. This is not something I do.
Ruth
Original comment by: RLovering
gocentral
commented
15 years ago I am not sure if it is easier to add comments here or in the wiki (http://wiki.geneontology.org/index.php/Annotation\_consistency:\_%27Response\_to%27\_terms\#1.\_Issue)
I can see that the additional work for GO editors to create lots of child terms below the response_to parents may be unnecessary. However, I am concerned that some curators have other focuses to their work and for them requesting GO terms is not a high priority. Consequently, if there was even just a small set of child terms which provide at least a statement about the start and finish to the process term then there is a good chance these terms would be used eg:
I disagree with Alex that the response to terms should be limited to the signal transduction aspect of a response. I think that the proteins involved in mediating the actual changes to the cell phenotype are relevant, although I appreciate that there will be a doubling up of many annotations, with genes involved in apoptosis, secretion of cytokines, cell motility etc will be annotated to these GO terms as well as to 'change in cell state in response to x'. But perhaps there are some genes that are involved in apoptosis (for example) which are not involved in the apoptosis triggered by some stimuli? If this is the case then the above structure would enable the different pathways to be identified.
Original comment by: RLovering
gocentral
commented
15 years ago Ruth,
If you read my earlier comment, you will see that I do not limit my annotations to simply the signal transduction aspects of a response, but include "gene transcription, or (occasionally) the effector mechanisms that take place in response to a particular stimulus." This certainly could include genes involved in the induction of apoptosis in response to a particular stimuli, if those genes are part of a specific pathway rather than those that are involved in general mechanisms of apoptosis. Obviously, this requires a bit of curator judgment.
Thanks,
Alex
Original comment by: addiehl
gocentral
commented
15 years ago Sorry Alex, too much to read
Great to hear we are all pretty much agreed on several points relating to this issue. It would be good to hear what the other editors to decide think about creating new child terms for the response_to terms.
Ruth
Original comment by: RLovering
gocentral
commented
15 years ago In general the definitions of the response to x GO terms is very consistent ;) and follow the same definition style eg: GO:0009268 response to pH has the definition: 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 a pH stimulus.
From this definition I have to slightly disagree with Alex's comment that: [I do find the "response to cytokine" and its children to be a bit redundant, since most of what I would annotate to such terms I would prefer to annotate to to the signaling pathway terms].
If anything the definition itself makes no mention of signal transduction and seems to be focusing more on what happens to a cell 3 hours (to 3 days) after signal transduction. Personally I think this is the only term available to link the change in morphology following stimulation of a cell with the stimulus. For example stimulation of a cell with insulin leads to the uptake of glucose, I can annotated genes involved in glucose uptake (GO:0046323 glucose import) however I want to be able to say that this set of genes is involved in glucose import following insulin stimulation to distinguish this set of genes from the genes which lead to glucose import due to blood glucose levels per se or those responsible for maintaining a basal rate of glucose uptake. I agree that often there will be considerable overlaps between the genes associated with changing a cell from a resting cell to an activated cell in response to IL1 compared with those genes involved in creating an activated cell in response to IL6 but there will also be a proportion of genes not involved in both processes.
While I agree with Alex that there is no point including genes within these 'response to' GO terms which are involved in 'housekeeping-type' activities such as DNA replication, RNA synthesis, respiration (although there maybe reasons why in some cases it is appropriate to consider these genes in the 'response to' GO terms) I also think that it should not be a problem to end up with a large number of genes within a GO term group, ie for response to insulin to include all genes associated with insulin signaling, insulin regulated transcription, insulin regulated of protein synthesis, transport of insulin regulated proteins. As long as there are also more specific terms available to group these proteins according to more specific functions. After all it might be that a dataset could show that the action of a new drug is through the mimicking IL1, leading to upreg of the 'response to IL1' GO gene group, or that an individual person is deficient in IL6 response, through the downregulation of 'response to IL6' gene group. Potentially such a change could be downstream of the signal transduction pathway and would not be identified by the more specific x-mediated signaling pathway GO term.
Of course as Alex points out the complication is that some ligands/stimuli act through multiple receptors and some receptors have multiple ligands/stimuli. Which is probably why this whole ontology domain is taking a long time to resolve. I think it may be easier to work with this if each 'response to' term is considered on an individual basis. Although I think it would be helpful if the aim for each of the 'response to x' GO was to consider whether the following child terms are appropriate.
Parent: response to x (ligand/stimulus) With modified definition: A change in state or activity of a cell or an organism (in terms of movement, secretion, enzyme production, gene expression, signal transduction etc.) as a result of an x stimulus.
Definition: A series of molecular signals generated as a consequence of x binding/stimulating a cell surface receptor.
Definition: Any process that modulates the movement, secretion, enzyme production, gene expression, intracellular transport, metabolic process of a cell as a result of an x stimulus.
So that we would have: GO:0032868 response to insulin stimulus > is_a Child: GO:0032869 cellular response to insulin stimulus >> is_a NEW Child GO:0008286 insulin receptor signaling pathway, note that the definition here is: The series of molecular signals generated as a consequence of the insulin receptor binding to insulin. >> is_a NEW child GO term: change in cell state in response to insulin stimulus Definition: Any process that modulates the movement, secretion, enzyme production, gene expression, intracellular transport, metabolic process, activity of a cell as a result of an insulin stimulus.
This new GO term could then be used for the genes which are involved in the translocation of GLUT4 to the cell surface, thus enabling an increase in glucose import, it would also be used for the gene products which are involved in the regulation of the translocation of GLUT4 to the cell surface, and the regulation of lipid synthesis and glycogen metabolism.
Whereas with GO:0032496 response to lipopolysaccharide
we would have: GO:0032496 response to lipopolysaccharide >NEW, parent for term GO:0031663 lipopolysaccharide-mediated signaling pathway (definition The series of molecular signals generated in response to detection of lipopolysaccharide.) >>NEW, parent for term GO:0032497 detection of lipopolysaccharide, currently a child of response to lipopolysaccharide (no proteins annotated to this) >I think we should also include the NEW Child:change in cell state in response to lipopolysaccharide, so that genes involved in more downstream pathways have this more specific GO term available.
So TLR4 would be annotated to detection of lipopolysaccharide and GO:0034142 toll-like receptor 4 signaling pathway.
However GO:0019933 cAMP-mediated signaling Definition: A series of molecular signals in which a cell uses cyclic AMP to convert an extracellular signal into a response. Would not be included as a child of GO:0051591 response to cAMP.
As a final example if you activate endothelial cells with interleukin-1 beta these cells increase the level of E-selectin (SELE) on their cell surface which leads to an increase in the number of leukocytes adhering to the endothelial cells and the migration of leukocytes into a site of infection. SELE is therefore involved in the phenotype 'cell adhesion' associated with IL1 activated cells. To capture this I would like to use the GO term response to interleukin-1 stimulus.
GO:0034097 response to cytokine stimulus >is_a child term: response to interleukin-1 stimulus >> is_a child GO:0070498 interleukin-1-mediated signaling pathway
New Parent: response to interleukin-1 stimulus > is_a NEW child GO term: change in cell state in response to interleukin-1 stimulus Definition: Any process that modulates the movement, secretion, enzyme production, gene expression, intracellular transport, metabolic process, activity of a cell as a result of an interleukin-1 stimulus. To which I could associate SELE.
Ruth
Original comment by: RLovering
gocentral
commented
15 years ago Archive of Alex email 10 Mar 2009 Ruth,
I stand by my statement:
I prefer to reserve "response to" terms for the gene products involved in the actual detection, signal transduction, gene transcription, or (occasionally) the effector mechanisms that take place in response to a particular stimulus, and not simply for any mRNA or gene product that varies in its abundance following a stimulus.
Reading the comment in the SF item 2665559, I can see the argument for why "response to X" might be the parent of "X signaling pathway," where X is a ligand, and I basically agree with it. We have two classes of "X signaling pathway" terms in the GO, however; one class has X as a ligand, such as "LPS signaling pathway" and the other class where X refers to a receptor, such as "TLR4 signaling pathway." Clearly, "LPS signaling pathway" should have "response to LPS" as a parent by this logic, whereas, "TLR4 signaling pathway" should not, since TLR4 has other ligands besides LPS.
On the other hand, I do find the "response to cytokine" and its children to be a bit redundant, since most of what I would annotate to such terms I would prefer to annotate to to the signaling pathway terms. And I fear, as always, that certain folks want to employ these terms to annotate everything that varies in its expression in the presence of a cytokine, rather than the proteins that are closely involved in the response.
But because of the way "response to" terms are defined, how they get used essentially comes down to the judgment and training of particular annotators, who I would hope understand that by being more careful in the application of these terms, the value of the annotations is much greater. Annotating all six hundred or more gene products that vary in their expression level in the presence of interferon-gamma to "response to interferon-gamma" does not say much if 575 of those gene products are also annotated to "response to IL-4" and "response to IL-10" and are really just involved in basic processes such as "DNA replication" or "translation" that are downstream consequences.
Thanks,
Alex
Original comment by: RLovering
gocentral
commented
15 years ago Archive Midori email 11 Mar 2009
Hi Alex,
Thanks for your comments. The third paragraph states my main outstanding question rather better than I did myself: where X is a ligand, is it ever not redundant to have both "response to X" and "X signaling pathway"? My foray into "gene products outside the signaling pathway" was an attempt to explore what, if anything, could reasonably be annotated to "response to X" but not to "X signaling pathway"; I certainly don't favor extending annotation to everything that changes expression following a stimulus, and I don't know any of these signaling phenomena well enough to know whether any gene products other than those in a given signaling pathway meet a set of sensible criteria for annotation to "response to" the ligand in question. In short, when, if ever, do we need both "respnose to [ligand]" and "[ligand]-mediated signaling pathway?
I'm also quite willing to entertain the notion of rephrasing the "response to X" definitions for any X, ligand, receptor, or otherwise. But I'm not qualified (in that I lack the requisite biological expertise) to draft the revised version, though I'm happy to help implement whatever is agreed upon.
midori
Original comment by: RLovering
gocentral
commented
15 years ago Should we make this an agenda item at the GOC meeting?
Original comment by: pgaudet
gocentral
commented
15 years ago I think it would be good to discuss at GOC if editorial team happy to do this?
Ruth
Original comment by: RLovering
gocentral
commented
15 years ago now on the agenda
Ruth, can you prepare a bit of introduction?
m
Original comment by: mah11
gocentral
commented
15 years ago I would welcome discussion of this at GOC assuming the teleconferencing system actually works well enough to allow discussion. Clearly there are both ontology structure issues and annotation usage issues to work out. I am preparing a long comment to be posted here as well, probably tomorrow.
Thanks,
Alex
Original comment by: addiehl
gocentral
commented
15 years ago Here is my long posting, at last:
General Principles
In dealing with the “response to X” terms, I have wanted to avoid changing the definition to avoid the problem of possibly invalidating existing annotations. I rather prefer to influence the usage of the terms in annotation. But I would argue that the existing definition format of the “response to” terms is incomplete and consequently misleading. Ideally the definition should have been written “The process of 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 a stimulus.” In this way, it would be more clear that the gene products annotated to this term are not simply a list of the things that change in a cell following a stimulus, but rather gene products that are actually involved in the process of change in response to a stimulus. A proper Aristotelian definition for these (and many other) GO terms from the start would have helped.
After all, the GO biological process ontology is an ontology of biological processes, not of states or phenotypes. Any term needs to be interpreted as a biological process regardless of the term name or definition, and any gene product that is annotated to a term needs to be interpreted as somehow involved in the carrying out of the process itself, not as a ligand that stimulates a process, nor as a product that results from the process. Thus we do not annotate a ligand to the signaling pathway for that ligand, and we do not annotate individual proteins to “translation” or “gene expression” etc., if they are the products of a translation or gene expression or for that matter, a “response to process,” in most cases.
However, the meaning of “response to X” is a little tricky, in that sometimes the product of a “response to process” carries out an effector action against the cause of a stimulus. Thus, if antibacterial peptides are produced as the result of a bacterial stimulus, the individual peptide gene products ought to be annotated to “response to bacterium,” (although perhaps “defense response to bacterium” is more appropriate). If a particular enzyme involved in a glucose catabolism is produced as the result of a glucose stimulus, then a “response to glucose” stimulus seems appropriate.
On the other hand, gene products produced in response to an insulin stimulus that do not somehow affect the level of insulin should not be annotated to “response to insulin,” as they are simply the end products of the process that do not directly feedback on the stimulus, but simply reflect the state or phenotype of the cell or organism following the stimulus. Going even further, whether gene products that indirectly affect the level of insulin should be annotated to “response to insulin” is questionable, since the cause and effect linkages are weak and dependent on additional factors.
Specific responses to Ruth’s posting of 2009-03-11 13:11.
1) It is important to stress that the “response to” definitions make no reference to timescale of the response, and thus signaling pathway components involved in the immediate response to a stimulus ought to be annotated to the matching “response to” term. Hence, “X signaling pathway” terms can be is_a children to “response to X” terms, because despite the names and definitions all signaling pathway terms refer to processes, not static entities, and the process of signaling through that pathway is a type of cellular response to a particular stimulus.
And for pragmatic reasons alone, any attempt to impose a timescale on the “response to” terms is inappropriate at this point given the number of existing annotations that would need to be checked.
2) As noted above, the GO biological process ontology is an ontology of biological processes, not states or phenotypes. Saying a protein is part of for instance “the ‘response to IL-1” GO gene group” based on an expression experiment is the reporting of the state of a cell or organism following a stimulus, not evidence of involvement in a process. The process of responding ends with the making of certain gene product, but does not include that product, unless it enables the handling of that stimulus by cell or organism. A simple expression experiment is insufficient data to answer the question of whether the products resulting from a stimulus enable handling of the stimulus.
Furthermore, people who analyze microarray data use the GO to identify biological processes that may be activated following a particular stimulus, but they also use other sophisticated software to compare the states represented by expression levels before and after a stimulus. This is not what the GO is designed for.
3) I don’t see the need for the terms “change in cell state in response to X”, as such terms are entirely redundant with the parent “response to term”.
4) It is entirely wrong to annotate E-selectin (SELE) to “response to IL-B stimulus” because E-selectin is produced as a result of a IL-1 beta stimulus but not involved in the process of responding to the IL-1 stimulus. It is a product of a process, and products of processes are not annotated to those processes. We have no GO term for “made in response to X” which seems what is intended by the “change in cell state in response to X” terms.
5) As argued by me previously and accepted at the time by group discussion, the process of detection is a type of a response, specifically the induced response of a receptor molecule in the presence of its ligand, physical or otherwise. Receptor molecules respond to the presence of their ligands (generally) by conformation change leading to activation of an enzymatic activity and transmission of a signal.
Original comment by: addiehl
gocentral
commented
14 years ago Original comment by: jenclark
gocentral
commented
14 years ago Addressed at annotation camp.
Original comment by: jl242
gocentral
commented
14 years ago Original comment by: jl242
Hi
Following a recent RefGenome conference call the question whether there is consistency in the annotation of 'response to stimulus' was raised.
See http://wiki.geneontology.org/index.php/Annotation\_consistency:\_%27Response\_to%27\_terms
I would be grateful if a GO curator could comment on whether a refined definition for these terms is needed which would define the start and finish of these processes.
Thanks
Ruth
Reported by: RLovering
Original Ticket: "geneontology/ontology-requests/5388":https://sourceforge.net/p/geneontology/ontology-requests/5388