search

geneontology / amigo

AmiGO is the public interface for the Gene Ontology.
http://amigo.geneontology.org
BSD 3-Clause "New" or "Revised" License
29 stars 17 forks source link

Reimplement slimmer functionality #144

Open kltm opened 10 years ago

kltm commented 10 years ago

There is still interest for online slimming functionality; the goal here is to first implement slimmer functionality with the improvements to the perl backend from #69, then move on to more interesting functionality (like defining your own cutsets using a tree browser).

kltm commented 10 years ago

Tagged for 2.2 for now, if #69 ends up being nasty, may be kicked to the indefinite future or binned.

kltm commented 10 years ago

@cmungall may have additional commentary/requests (that may end up being separate issues).

cmungall commented 10 years ago

I think the create-your-own-slim functionality may be best broken into its own ticket - or tagged onto a ticket for a tree browser with gene counts. The basic idea is

  1. Starting with root(s), recursively auto-open children until nodes get too specific, where this could be user defined (e.g. must have at least 1% of genes in the genome of interest)
  2. Allow user to open and close using standard tree browser functionality
  3. User clicks "save as slim". All leaf nodes are used as the slim
kltm commented 10 years ago

I agree, but let's keep it here for now and break it out once we start work here after #69--we'll have a better idea of performance and consequences at that point.

kltm commented 9 years ago

From Val:

Will the web based GO slim tool be returning at any point?
see comments in this ticket
http://jira.geneontology.org/browse/GO-1027

PomBase currently point users to
http://go.princeton.edu/cgi-bin/GOTermMapper
rather than QuickGO because it matches our output
http://www.pombase.org/browse-curation/fission-yeast-go-slim-terms

We don't want to point our users to QuickGO because the GO annotations are differently in GOA than in GO
1. We have filtered all of the annotations generated by erroneous mappings submitted to this tracker
https://github.com/geneontology/go-annotation/issues
AND

2. GOA/QuickGO filter annotations for syntax errors which are currently accepted by the
GO loading script.
geneontology/go-site#123
suzialeksander commented 4 years ago

Bumping as there was recent interest in this from Helpdesk, spawned from the "Paris" virtual Users Meeting

kltm commented 4 years ago

@suzialeksander I think this is probably not going to happen in AmiGO. Once the functionality was originally removed, we briefly had it in Galaxy (remember that?) and that never really gained traction. The API covers some of this and this is possibly better tackled via web (Jupyter) notebook or the like.

ValWood commented 4 years ago

But slimmer functionality is not so far removed from the MAtrix tool so why not provide it in the familiar AmiGO environment?

kltm commented 4 years ago

@ValWood At this point, and with many things that we'd like to do, limited resources and priorities. It may be possible to restart some of this when we start into AmiGO development again, but as it has been for the last few years (last note on ticket five years ago), we are unlikely to meet that need with our current roadmap with the current AmiGO. That said, if there is a huge push for this, versus things like QC efforts or Noctua, then resources could be freed.

ValWood commented 4 years ago

OK. It does feel a little like the public facing and major products of GO (ontology and web tools) are a little deprioritised and starved of resources. But I guess it isn't possible to do everything

lpalbou commented 4 years ago

That's certainly something to discuss more. If we are to do more User Meetings, we'll have to improve our APIs and docs indeed.

As mentioned in the other ticket, creating a slim is a different task than mapping annotations to a set of higher level terms. The former, I don't see that happening any time soon but the later has been available for some time in the API and is used by the ribbon to create the association tables.

ValWood commented 4 years ago

but the later has been available for some time in the API and is used by the ribbon to create the association tables.

This isn't something the average biologist could use to slim their own lists though. They need a web interface .We cater far too much for informaticians and not enough to lab scientists.

The slim functionality at PomBase is really . popular. We don't have an API, and we've never been asked to provide one (although we should), but the point is most of the GO (potential) user base would not know how to use the API, they will go elsewhere.

lpalbou commented 4 years ago

Agreed. I will try to reopen the discussion. Do you have a link to the pombase tool ? Thanks

ValWood commented 4 years ago

The slimmer functionality is embedded into our query tool.

https://www.pombase.org/query You can upload any pombe gene list and slim (now using process, function, component, or disease slim)

To try it you can select a list from the commonly used queries (all disease associated genes for example).

Click on the list in the results table. One of the "actions" at the top right of the results table is to slim your list

very quick, very easy.

ValWood commented 4 years ago

The 'full' slim sets are available from the front page https://www.pombase.org/

I don't have stats for the query tool, only for the static pages. The new disease slim, relaunched last month using MONDO has been accessed 101 times by 63 unique users since May 1st. The BP slim has been accessed 88 times by 55 users. I imagine many many more people use the slim feature with their own queries.

Slims are really popular for people to orientate themselves with their data. For bench scientists, probably more so than enrichment tools if they are made easily avaiable, and the slim has good coverage. They can see immediately what they are dealing with and plan where to go next.