Menu
▾
▴
#75 AmiGO 1.5 and 1.8 rendering subsets as ontology classes
We are getting a weird rendering by AmiGO 1.5 of the subsets, such as "plant cells", Bryophytes, Plant Functional Traits, Musa etc
http://www.plantontology.org/amigo/go.cgi?view=details&search_constraint=terms&depth=0&query=CL&session_id=2476b1381869788. I checked all of our subsets and they are all showing up the same way.
Looks like it occurs in the GO as well, here is an example: http://amigo.geneontology.org/cgi-bin/amigo/term_details?term=goslim_yeast&session_id=9720amigo1381871966
Looks like it is occurring in the AmiGO 1.8 as well.
This causes strange results in searches, and I doubt this is intentional.
Thanks
If I'm reading your bug report correctly, we are talking about a software problem and not a recent data problem. I've got a (probably) unsatisfying answer, but I think the heart of it also involves where you're going from AmiGO 1.5 and in what time frame. Justin may be better able to answer that?
The short of it is that the options come down to:
1) Live with or otherwise work around the problem (since there is likely no easy fix without restarting development).
2) Polish the rough edges of AmiGO 1.9 and hope for the best. This would mean essentially removing all the leads to AmiGO 2.0 and "release" it. Given the amount of work for something that we wouldn't put into production ourselves, it would not get a lot of juice at this end. However, it's something that I think might be doable for a third-party familiar with the code and I could give some help there. Assuming that it does indeed fix the problem that you're having.
3) Make a move towards AmiGO 2.0. Naturally, I think this is the way to go. All development is there and it is ready to go. If there are use cases that it does not cover, we can see about covering those. Also, since it is designed to be more general than the AmiGO 1.x series, it should be easier to specialize to what you want. I'm always available to talk about the details and help out in any test installations that you might want to try. In fact, if you are loading from GAF and OWL files, I could pretty easily setup a test installation over here for you to look at.
To justify those answers, there is a little history, but you can probably safely skip it. To give a quick overview of recent AmiGO development (and ignoring what was going on with the database, etc.), looking back at the wiki, AmiGO 1.5 seems to have come out in the early part of 2008. The release notes indicate that it was mostly an evolutionary advancement from the previous versions (nothing earlier on the wiki). AmiGO 1.6 was the addition of some new page types and a light step into a new framework (DBIx::Class and CGI::Application). AmiGO 1.7 was more commitment to the new framework, a new way doing some data handling, and the first steps with Lucene/Solr. AmiGO 1.8 had much more advanced caching, more page/worker division, more experiments with Lucene/Solr, and more use of JavaScript elements. The never released AmiGO 1.9 (currently trunk) contains many parallel pages that use a Solr backend instead of SQL and more exploration of JavaScript, etc. Early in 1.9, we decided to commit to Solr as the backend and a lot more UI JavaScript on the frontend, and jumped to AmiGO 2.0 (which we are now setting up in production).
Okay, back to the problem at hand, apparently this issue with slim terms was noticed and fixed, but only relatively recently. Looking at the 1.9 release (sorry, no public server with it right now), the slim/subset pages have terms lists, GP counts, and seem to function just fine. Looking at the differences between the code in 1.8 and 1.9, the differences are very large--there is no simple way to backport the fixes from 1.9 to 1.8 because there was a lot of additional development in how data was handled and in the backend. This is even more true for trying to fix 1.5, which has not seen any activity in a very long time; to fix 1.5 would mean essentially restarting development on it.
Thanks for the detailed response- I agree that there is no point in fixing these older versions and I look forward to the new and improved 2.0 for both GO and PO. I have been exploring the Amigo 2.0 beta site and it is looking pretty good. Is there a page to view subset information?
Given some pressure, we could probably reinstate something, but after some internal discussion a while back, we decided that most all of the use cases were covered by the faceted filtering in the search. For example, if you go to the Ontology Search page, you can see the "Subset" facet along the left in the filters. So no, currently no page dedicated exclusively to the subset information...
We should talk if you find this problematic or have views on this.