Ideas for GOC community curation tool
From GO Public
- currently looking at developing the PomBase community curation tool into a version suitable to be promoted on and by the GOC
see here for version of the PomBase tool: http://sloth.sysbiol.cam.ac.uk/uniprot/
Suggestions on Tool Functionality
N.B. Suggestions below are made mostly in context of the specific PomBase pages.
However, the threshold for inviting annotation suggestions from users should be low, therefore if a community annotators would like to write a quick free-text entry on the annotations that they would like made to their gene of interest - this facility should be available along with some short suggestions on supplying meaningful identifiers and a reference. However it should also be made clear that annotation suggested in this free-text manner will take longer to progress to the public GO database.
- There needs to be an annotation bulk load option - if a CA has 50 annotations to a set of proteins to apply the same GO term, then it needs to be easy for the CA to enter these, if they prefer this method.
1. Find a publication to curate using a PubMed ID
- could we provide simultaneous option for community to feedback on existing annotation(s)? (see also point 4)
- could Community Annotators (CAs) be allowed to enter alternative reference info, such as an ISBN, DOI or Agricola Ref, as well as a PMID? However, it would seem best to restrict submissions to absolutely requiring a public reference, if we want to be able to easily check the veracity of CA annotations.
2. Curator details
- It would seem preferable not to accept anonymous entries - as might lead to less high-quality annotations, if the CA enters their name/email then perhaps they would be more conservative as to what they suggest.
- However, if we do require contact info, should we include information on our contact and email sharing policy here?
3. 'Create gene list for...'
- should we change the text to indicate users can suggest gene/protein or functional RNA identifiers?
- we chould not require UniProtKB accessions, but encourage users to try to find them (as useful to us, and helps the CA as well in checking what they are intending to add and allowing the tool to bring out the existing annotation set) by listing the UniProt database (and others?).
- If accession not entered, but instead a gene name, then species name should be supplied.
This confirmation step, would be helpful for annotation QA, another reason to encourage CAs should supply gene ids.
4. Where the existing GO annotations are displayed - could we have an option to allow curators to feedback on ones over which they have concerns about correctness? Perhaps there could be a short-cut to providing feedback on existing annotations from this tool's front page?
5. On the last page, at submission we need to make clear that these annotations are to be made public, how we'll acknowledge the annotations and when they should expect to see them in files/GO browsers. Also a bit on how they can get in touch with the GOC about the annotation set (include a specific annotation id for a submitted set?) Perhaps also include an option to mention their expertise and willingness to be consulted about specific ontology development projects?
Suggestions on Tool Display
- This tool should be branded with the GOC logo and colours.
- Perhaps if the tool had an icon, which included text such as 'GO annotation feedback', or similar, then could we circulate the icon, along with the tool's link to friendly GO browsers or websites that display our annotations, so that we can get feedback from users while they are looking at the existing annotation set?
- the tool would therefore be more visible than if we just advertised the annotation tool from the GOC web site.
Suggestions on Resulting Annotation Triaging.
- We currently receive some annotations, but if the tool was available and highly visible, this would lower the barrier - and we should hope for more submissions.
- Perhaps a summary of the annotations created could be regularly emailed to the GOHElp list - or an alert sent to the GOHelp list about additions to a database?
- If a database or log file of annotations was kept, then it would need to be kept private and completely seprate from the public GO data, as has not been checked and would contain submitter contact details. The GOHelp person would then be responsible for triaging the annotations between GOC annotation groups based on taxon id.
- Annotation groups would need to feed back to the CA once a decision has been made about their submitted annotation suggestions.