Fix problem of call propagation

[fbastian]

In GitLab by @fbastian on Sep 23, 2015, 15:49

Currently, to propagate no-expression calls, we have a system to identify valid conditions for propagation: we retrieve annotated conditions, and all conditions related to their parent anatomical structures and/or parent developmental stages.

Problem, this still generates invalid conditions: If we have a valid condition "brain - adult", this will produce a valid condition "embryo - adult" (because there is a part_of relation from brain to embryo, at some point, the brain is part of an embryo).

This problem also applies to expression calls, because we propagate all of them to parent anatomical structures and/or parent developmental stages, with no filtering on valid conditions.

• First, we should verify that such invalid conditions are indeed present in the download files
• Short term solution is to discover valid conditions only by using dev. stage propagation; we will loose lots of anatomical structures...
• Long term solution is to correctly have the start stages/end stages in Uberon for each anatomical structures.

[fbastian]

In GitLab by @fbastian on Sep 24, 2015, 15:15

• Such invalid conditions are indeed present in the files
• It was decided to implement the short-term solution for the next release of the Data (Bgee 13.2)
• It was agreed that we should invest time on the start/end stages of Uberon, but that would be for late 2016...

[fbastian]

In GitLab by @fbastian on Oct 9, 2015, 14:29

Maybe we also need a way to tag a condition as "to not propagate", because we annotate gestational structures to the mother's age (see https://github.com/BgeeDB/expression-annotations/issues/42), so we have annotations such as placenta - adult, or decidua - adult, that will lead for sure to incorrect propagations.

But this problem is restrains to embryonic/gestational structures, so if we have correct start/end stages for embryo, maybe we can rely only on that.

[fbastian]

In GitLab by @fbastian on Oct 14, 2015, 22:26

See discussion re-launched a week ago about start/end stages: https://github.com/obophenotype/uberon/issues/423#issuecomment-146146305