Add all positive annotations for sub-taxa of a NOT annotation

[fbastian]

See issue #4.

When creating a NOT annotation, and in order to correctly retrieve in which taxa a structure independently evolved, it is necessary to capture all positive annotations for sub-taxa of the NOT annotation.

For instance, we provide a NOT annotation for UBERON:0002470 autopod region at the vertebrata level, and a positive annotation at the tetrapoda level. We miss yet another positive annotation at the Dipnoi or Coelacanthimorpha level.

cc @ANiknejad

[fbastian]

Note that sometimes there is no sub-taxon positive annotation. For instance, a structure is originally though to originate in tetrapoda, but it then showed to originate in vertebrata: we will add a NOT annotation at the tetrapoda level.

There can be several sub-taxon annotations only in cases of independent evolution. Otherwise, the NOT annotation is only used to capture the rejection of a previous hypothesis.

edit: this example is actually incorrect, it is valid the other way around, see my next message.

[cmungall]

Hmm, this is different than what I would expect. I would have thought a NOT at the tetrapoda level would mean that it's not homologous in tetrapods.

I think it's useful to record that a structure did NOT originate within a particular taxon, but I think this should be recorded in a separate way

[fbastian]

Oh, you are actually correct, sorry (I'm still confused because we recently changed the meaning of an annotation, from '1 row = 1 common ancestor' to '1 row = evidence of homology')

In my previous example, we would not create a NOT annotation for tetrapoda, you are correct.

But consider the opposite example: a structure previously thought to originate in vertebrata, now considered to originate in tetrapoda; we would create a NOT annotation at the vertebrata level; no pair of valid sub-taxa to consider, only tetrapoda.

[cmungall]

OK, but there are some intermediate nodes between vertebrata and tetrapoda. So the interpretation of the pair annotations above would be: we believe this structure arose some time after the rise of vertebrates, either during the rise of tetrapods, or possibly inbetween?

[fbastian]

If we thought it was possibly invetween, we would provide the annotation.

We use NOT annotations mainly to capture controversies, so, most of the time along with a positive annotation. These annotations become one, positive, SUMMARY annotation, with a confidence code 'conflicting evidence lines'.

NOT annotations that do not have a corresponding positive annotation, for the same taxon, can be seen as optional information, a "you thought it appeared in this taxon, but, haha, guess what, you were wrong". We will always try to provide an annotation capturing the 'true' common ancestor for the corresponding structure. You don't need to guess using the taxonomy inbetween.

[ANiknejad]

there is an example with CL:0000216 Sertoli cells: positive and NOT annotations on Amniota, while there were only positive annotations on Vertebrata and on Chordata: I now added NOT annotations on these last both.taxons. The negative annotations are all based on this reference: PMID:17026980 "Sekido R, Lovell-Badge R, Mechanisms of gonadal morphogenesis are not conserved between chick and mouse. Dev Biol (2007)" "(…) nephrogenous mesenchyme contributes to both Sertoli cells and steroidogenic cells. This is the first demonstration via lineage analysis that steroidogenic cells originate from nephrogenous mesenchyme, but the revelation that Sertoli cells have different origins between chick and mouse suggests that, during evolution, mechanisms of gonad morphogenesis may diverge alongside those of sex determination."

Do you agree? @cmungall @fbastian

[fbastian]

Well, your example adresses a different problem: that when we have NOT annotations, we must propagate them to already annotated parent taxa.

Here, we speak about making sure that when we have a NOT annotation, we also capture all sub-taxa where the structure is homologous, in case of independent evolution.

[ANiknejad]

ok I guess, I understand. It was just an example to point out how to make positive and NOT annotations consistent through the homology file

[fbastian]

Check if current pipeline generates a warning if some NOT annotations miss positive annotations for sub-taxa.