"Response to" terms
The following is some key points raised in a discussion between Alan Bridge, Pascale Gaudet, Rachael Huntley and Rama Balakrishnan during 2012.
Currently, the annotations to response to terms are really problematic, and much overused. In many cases, curators appears to make annotations to "response to X" when in fact they are really saying "responds to X". See also the Use_of_Response_To_Terms_in_Annotation page by Alex Diehl, which highlights the main problems with using these terms.
Alex's thoughts boil down to how curators want to use these terms; either 1) to annotate any gene product that changes in its expression level in the presence of X to "response to X.", thus yielding thousands of gene products annotated to response to terms that provide little added value and masking the gene products that are truly involved in a response or 2) to annotate gene products that are involved in the mechanism of a response, e.g. by using more informative terms, such as "response to dietary excess", which is defined in terms of the response mechanism.
In the second case, we would ideally like to annotate; "gene product X is required for (cellular, organismal) response Y to stimulus Z" and also be able to say what X, Y and Z are.
These annotations could be used in situations where you have definitive experimental evidence for a requirement for a given gene product X in a specific cellular/organismal response Y to a stimulus or perturbation Z (RNAi), but when you don't know the mechanism underlying this requirement (maybe synthetic lethal data tells you it's not in a known pathway). If you did know the mechanism you could (for instance) annotate with the correct biological process term.
In some cases, these terms seem to be used with another (complementary) term that is more specific, and which provides more biological meaning. For example, A2A259 is annotated to:
GO:0001581 - detection of chemical stimulus involved in sensory perception of sour taste - IDA PubMed 16891422
GO:0071468 - cellular response to acidity - IDA PubMed 16891422
The first annotation is interesting - this protein is detecting chemicals that tell us food is sour. The chemical used to test this is citric acid - which is sour of course, and whose acidic nature is I guess is the source for the second annotation. This annotation tells you nothing about the response though, it's really there to tell us sour things are acidic.
If the "response to" terms are to be restricted for use with gene products that are actively contributing to the mechanism of the response, i.e. the effect does not occur when the gene product is absent, then curators should first consider what the response or cellular response to the particular stimulus is likely to be. Is the gene product you are annotating shown experimentally to be required for that response to occur, i.e. does it mediate the effect that the stimulus has?
Annotation of gene product 'X' to 'cellular response to nitric oxide'. The cellular response in this experiment is production of gene product 'Y' 'X' is required for this response because in the absence of 'X', no 'Y' is produced. Therefore it is acceptable to say that 'X' has a role in the cellular response to nitric oxide.
If we follow this rule, then we need;
1. To have in vivo experimental data for using a cellular response to term
2. To have a biological readout corresponding to the process being annotated to (cell differentiation, DNA damage, etc).
One possibility to avoid over-use of these terms is to create more descriptive "response to" term names and definitions.
GO:0072432 response to mitotic cell cycle G1/S transition DNA damage checkpoint signal - "A process that acts directly to delay or stop progression through the cell cycle in response to signals generated as a result of mitotic cell cycle G1/S transition DNA damage checkpoint signaling; contributes to a mitotic cell cycle G1/S transition DNA damage checkpoint."
For this example we could introduce more specific terms such as, 'DNA repair in response to mitotic cell cycle G1/S transition DNA damage checkpoint' and 'cell cycle arrest in response to mitotic cell cycle G1/S transition DNA damage checkpoint'. This would help curators to make sensible use of these terms.