Open ValWood opened 6 years ago
The code for summarising the annotation currently doesn't consider the ancestry of the extension ranges. It only looks for exact matches. Your example would have been filtered if the bottom annotation had the extension "part of protein localization to chromatin"
See also #863
closing https://github.com/pombase/website/issues/863
This is redundant, although slightly different example.
A ancestor term without an extension should be filtered if there is a more specific term with an extension.
This is related to pombase/website#277 except that this issue is about the extension range values and #277 is about the extension range relations. It's best to handle them together.
Another example for checking
RNA polymerase II CTD heptapeptide repeat kinase activity | phosphorylates rpb1
RNA polymerase II CTD heptapeptide repeat kinase activity phosphorylates rpb1
is redundant with the statements below
and another
I think we have examples of most of the types. It might be good to move forwards with this because it would reduce the complexity of the GO annotation for the users quite a bit.
It might be good to move forwards with this because it would reduce the complexity of the GO annotation for the users quite a bit.
I haven't worked out how to do it yet. :-(
Does the way you handled the filters for sending the targets to the query builder help (it seems very related)?
The thing I'm struggling with is that we need a way to efficiently compare all annotations (for a given term) against each other (all vs all) while also taking into account the ancestry of all the terms in all the extensions. There are some cases where all vs all comparison of the annotations and ancestry could get slow because some terms have lots of annotations with extensions: https://www.pombase.org/term/MOD:00696
Currently because we ignore the extension term ancestry there is a trick that means we can shortcut the all vs all comparison. I've been trying to work out a trick to use that includes ancestry.
Yes it does sound very head melty.
binds hrk1 part_of "protein localization to chromatin"
seems to be fully redundant with the bottom one?