WBPaper00038399 GO linking comments
The links to the following terms should be removed because the term in the text is part of a phrase that should have its own term or when linked alone does not reflect the actual meaning of the phrase it is part of in the text.
- 'cell' when referring to 'cell fate', 'cell corpses', and 'cell type'
- 'binding' when referring to 'binding sites', 'binding regions'
- 'transcription' when referring to 'transcription factor'
- 'motor' when referring to 'motor neuron' (note: motor neuron would probably be linked to a WB anatomy term)
- 'RNAi' when referring to 'RNAi library'
The links to the following terms should be removed because in the text it is referring to a specific entity that is not reflected in the term
- 'RNAi' when referring to discreet RNAi objects ex. dpl-1(RNAi)
The links to the following terms should be removed because the term does not mean the same thing as the GO term
- 'midbody' - GO term is wrong for this word in this paper
I agree with Karen's evaluation of some of the links listed above. Links that are made to text that is part of a larger phrase with a different or broader meaning than that of the GO term aren't appropriate.
I found instances of this with (at least):
- binding - 'binding sites'
- cell - 'cell fate'
- cell death - 'cell death gene'
Note, though, that not all instances of these terms should be deleted. For example, in some cases linking out to cell was appropriate.
There were only two terms where all instances of the linking should be removed:
Time considerations: it took me ~50 minutes to go through the article and check all of the uses of each term.
1) The term "dimerization" with reference to the E2F/DP heterodimer links out to the term "protein homodimerization activity"; I think linking dimerization is valid and important, but it should link to the term "protein dimerization activity" or "protein heterodimerization activity" to better reflect the context.
2) I had same comments as Karen and Kimberly with regards to "cell", "binding", "motor", and "midbody".
3) As I think was already mentioned for entities "grayed out" in the entity table, the terms "core promoter" and "transactivation" are dead links.
I have begun looking at this project. I see that this is time-intensive, especially if we want to do something of high quality and of value to users. I also agree with all of above curator comments about links made to individual words in a phrase and about context sensitivity of the terms. Will continue to look into this.