'part of' o develops_from' -> develops_from ?

[alanruttenberg]

I'm wondering if this chain is correct. Chatted with @mellybelly about this and was looking for impact
by querying (develops_from some ectoderm) and (develops_from some mesoderm)

The chain is responsible for 6 subclasses

perilymph
otolymph
endolymph
"notochordal fluid"
"dorsal meso-duodenum"
"chorionic ectoderm"

I'm not sure about whether these are correct or not - don't know enough anatomy.

However it should not be right in the following scenario.

A' develops from A
B' develops from B
at some point A',B' fuse to make C, with parts retaining their identity.

At that point, C develops from A' and C develops from B'

A' part of C
C develops from A',B'
but A' doesn't develop from B'

[cmungall]

At that point, C develops from A' and C develops from B'

We wouldn't make this inference with the existing OWL axioms.

With the following structure:

we would get an inference that C has-developmental-contribution-from {A,B}, but this is intentional as this is a weaker axiom.

[alanruttenberg]

So there's two different questions:

1. Is the inference valid
2. Is a valid inference made or not by a tool

I'm not sure you are addressing the issue with your response. In my example it is given that develop relations hold, and I'm concerned about the consequences. In your response, if I understand what you are saying, you are denying the givens rather than addressing the consequence. C develops from A',B' is an assertion, and justified in any case since I described a fusion, which seems to satisfy the conditions for develops_from.

In any case, as I look at my example, I think it may be wrong, actually, in that if C develops from A,B by fusion shouldn't it be the case that A,B don't exist anymore under the semantics of the relations paper? I'm trying to construct a better example, as I'm still concerned that the chain isn't valid.

BTW, I also noticed that directly-develops-from isn't a subproperty of develops from, which seemed to be at odds with the definitions given. "x develops from y if and only if either (a) x directly develops from y or..."

[cmungall]

Ah, I thought I was denying an entailment rather than a given. If it is a given, then your entailments do indeed hold.

since I described a fusion, which seems to satisfy the conditions for develops_from

This is an interesting one, the intent is that develops_from would not necessarily hold for fusions or general "comings together". Bits of the ectoderm and the mesoderm develop into things like limb skin and limb muscle/skeleton that "come together" to make a limb but we don't want to infer from this that the limb develops from ectoderm (though it has a contribution from ectoderm)

Regarding whether fusions destroy the precursor elements, it's underspecified at the moment (we lack developmental axioms for fusion in general). The 2005 paper dealt with derives_from which is not necessarily a generalization of develops from. I think we would have to allow for non-destroying fusions.

[alanruttenberg]

Wouldn't hold for some fusions, or wouldn't hold for any fusion?

[tgbugs]

The example that I have in mind which I think matches the original example from @alanruttenberg is in the brain where there are interneuron populations that develop from the medial ganglionic eminence (MGE) and principle cell populations that develop from the subventricular zone (SVZ). Note that the interneuron population migrates, so this is probably not a realistic example, but if we imagine an interneuron layer that develops from the MGE and another pyramidal layer that develops from the SVZ both of which 'fuse' to form some cortical region of which they are both part. pyramidal layer part of some cortical region and interneuron layer part of some cortical region. My interpretation is that the concern is that if one were to add a develops from relationship in addition to the has part relationship, then interneuron layer incorrectly would be inferred to have developed from subventricular zone. Note that as written my interpretation is different that what is written above. In my example having develops from and has part between the two same regions should probably cause a reasoning error, since I think that develops from implies that the prior structure no longer exists, or that if it does, then it is technically incorrect to say that C in your example or some cortical region develops from it. If some cortical region comes into being after pyramidal layer and interneuron layer then I think the only additional axiom that would be needed is existence begins during.

[alanruttenberg]

BTW, I read the Relations paper to imply that develops_from is a generalization of derives_from, the union of derives_from and transformation_of, but maybe that's a misread?

[alanruttenberg]

How does 'fusion is not development' go along with the fact that many structures in Uberon that develop from both ectoderm and mesoderm. If there was no fusion at any point, what happened in the those cases?

[cmungall]

Wouldn't hold for some fusions, or wouldn't hold for any fusion?

Wouldn't hold for some. I'd propose this at a more general level than development, that we have 3 kinds of fusions, from A1, A2, ..., An -> C

• destroying, where all An cease to exist when C comes into existence
• preserving, where all An continue to exist
• partial, where some are "absorbed" and cease to exist

[alanruttenberg]

OK, then my example was aimed at one where it does hold.

[cmungall]

I didn't intend to imply fusion is not development. But regarding the ectoderm/mesoderm cases you highlighted, this is probably unintentional and there is probably an axiom that needs weakened, will investigate.

But there could be fusion (or "intermingling") cases where your example holds. A good example may be the neural crest

[alanruttenberg]

I'm looking at the definitions

x directly_develops from y if and only if there exists some developmental process p such that x and y both participates in p, and x is the output of p and y is the input of p, and a substantial portion of the matter of y comes from x, and the start of x is coincident with or after the end of y

x develops from y if and only if either (a) x directly develops from y or (b) there exists some z such
that x directly develops from z and z develops from y

I'm not sure how to read input and output other than by looking at the definitions of has_input and has_output.

P has output c iff c is a participant in p, c is present at the end of p, and c is not present at the beginning of p.

P has input c iff: p is a process, c is a material entity, c is a participant in p, c is present at the start of p, and the state of c is modified during p.

--
So to develop-from(x,y) there has to be at least one direct-develops-from step. In the development process that is that step, the input stops being "present"(existing) and the output starts. I haven't proved this yet, but doesn't that entail that if x develops from y, x and y don't coexist unless immediately-develops-from(x,y), and if then only during the developmental process for that step?

Unless there's just one developmental process per organism, which is the whole life?
Looking at the subclasses that doesn't seem to be the case. Plus many parts either don't exist at the start or at the end of life (and so can't be inputs or output, resp. of the one process).

[cmungall]

I think there is a typo in the definition of the direct form, here: "and the start of x is coincident with or after the end of y".

The intention is to block intermediates, not to destroy the precursor

E..g in https://github.com/oborel/obo-relations/wiki/ROGuide#direct-and-indirect-forms-of-relations

"Here the 'direct' prefix means that the two structures or classes should be in direct succession (no intermediates)."