Add annotations to 'derives from' and 'derives into'

[GoogleCodeExporter]

These are fuzzy terms. OBI and ENVO still use them, but I think other 
biological ontologies prefer 'develops from'.

This is what the RO2005 paper says: "it holds between distinct material 
continuants when one succeeds the other across a temporal divide in such a way 
that at least a biologically significant portion of the matter of the earlier 
continuant is inherited by the later."

RO_0001000 derives from
- textual definition: a relation between two distinct material entities, the 
new entity and the old entity, in which the new entity begins to exist when the 
old entity ceases to exist, and the new entity inherits the significant portion 
of the matter of the old entity
- add example of usage: this sand derives from this rock
- add example of usage: this blood specimen derives from this blood
- add editor note: This is a very general relation. More specific relations are 
usually preferred.
RO_0001001 derives into
- textual definition: a relation between two distinct material entities, the 
old entity and the new entity, in which the new entity begins to exist when the 
old entity ceases to exist, and the new entity inherits the significant portion 
of the matter of the old entity
- move inverse from ro-edit.owl to core.owl: 'derives from'
- add example of usage: this rock derives into this sand
- add example of usage: this blood derives from this blood specimen
- add editor note: This is a very general relation. More specific relations are 
usually preferred. To avoid making statements about a future that may not come 
to pass, it is often better to use the backward-looking 'derives from' rather 
than the forward-looking 'derives into'.

Better examples would be good. The sand example is temporally vague (when does 
the rock cease to exist?). The blood example requires that you accept OBI 
jargon.

Original issue reported on code.google.com by ja...@overton.ca on 2 Feb 2015 at 3:20

[GoogleCodeExporter]

 "derives from" requires that "at least a biologically significant portion of the matter of the earlier continuant is inherited by the later."  'develops from' is generally used as a transitive relation - best defined in terms of continuity of cell identity/ lineage.   This does not require inheritance of matter and so is not a subProperty of derives_from.  A non transitive directly_develops_from (where there is little or no temporal gap between the two continutants) would be though.

The physics examples seem a bit odd given that the definition talks about 
"biologically significant" portions of matter.  Some biological examples:

The relationship between a cell and the parent cell from which it was derived 
via cell division.
The relatinoship between a nucleus and the parent nucleus from which it was 
derived via nuclear division.
The relationship between an exosome and the parent cell from which it buds.
The relationship between a zygote and the egg from which it derived via 
fertilization.
The relationship between a zygote and the spermatozoan that fertilized the egg 
from which it derived via fertilization.

Original comment by dosu...@gmail.com on 2 Feb 2015 at 3:57

[GoogleCodeExporter]

Thanks for the examples. In the editor note we could say that 'develops from' 
as often a better term (despite the differences that you note).

While the RO2005 paper says 'biologically significant', the RO OWL never had 
textual definitions, so OBI and ENVO have used these relations for 
non-biological cases. I'm suggesting that we make the more general usage 
official.

Would RO developers prefer to obsolete these terms?

Original comment by ja...@overton.ca on 2 Feb 2015 at 4:12

[GoogleCodeExporter]

Are you suggesting to obsolete “derives_from”? We have been using those in 
our resources (with the RO2005 definition), with the understanding that there 
was a biologically significant part retained.
For example, when a ribosomal subunit is methylated and which prevents 
antibiotic binding and the organism becomes resistant as a consequence. I'm not 
sure the textual definition proposed, in which the initial continuant ceases to 
exist works for this case either. 

 I don't think "develops from" works for that case - it seems as David mentions that this was intended more in a lineage sense?

Original comment by mcour...@gmail.com on 2 Feb 2015 at 10:19

[GoogleCodeExporter]

I'd like to keep these terms, but if they're too broken we can try to make 
better terms and obsolete these.

I'm proposing that we generalize the RO2005 definition to allow for 
non-biological cases that are already being used. Instead of "at least a 
biologically significant portion" I want to say "the significant portion".

The RO2005 definition is very clear that the new entity begins to exist when 
the old entity ceases to exist. Without that jump from old entity to new entity 
we have a transformation and not a derivation. (I can add an editor note about 
that.) Here's a longer quotation:

"Derivation as a relation between instances. The temporal relation of 
derivation is more complex. Transformation, on the instance level, is just the 
relation of identity: each adult is identical to some child existing at some 
earlier time. Derivation on the instance-level is a relation holding between 
non-identicals. More precisely, it holds between distinct material continuants 
when one succeeds the other across a temporal divide in such a way that at 
least a biologically significant portion of the matter of the earlier 
continuant is inherited by the later. Thus we will have axioms to the effect 
that from c derives_from c1 we can infer that c and c1 are not identical and 
that there is some instant of time t such that c1 exists only prior to and c 
only subsequent to t. We will also be able to infer that the spatial region 
occupied by c as it begins to exist at t overlaps with the spatial region 
occupied by c1 as it ceases to exist in the same instant."

So the key question is: is it the same ribosomal subunit after methylation? If 
so, then methylation is a transformation. If not, then methylation is a 
derivation (in the sense here).

Original comment by ja...@overton.ca on 3 Feb 2015 at 2:21

[GoogleCodeExporter]

I guess that is difficult to answer. What does "being the same" means? We know 
that a significant portion of biological material is retained, but the 
properties of the continuant have changed (e.g. from sensitive to resistant to 
antibiotic) As sensitivity is a disposition, and disposition are tied to the 
physical makeup of the continuant, does that mean that the bearer changed 
enough to not be "the same"?
What do others think?

Original comment by mcour...@gmail.com on 3 Feb 2015 at 6:34

[GoogleCodeExporter]

what happens in the case of tree grafting, where you take a portion of one tree 
t1 and add it to another tree t2, so as to unify their vasculature. do you have 
a new entity t3 where t1 and t2 cease to exist, or is t3 a transformation of t2 
and t1?

Original comment by michel.dumontier on 3 Feb 2015 at 11:43

[GoogleCodeExporter]

> So the key question is: is it the same ribosomal subunit after methylation? 
If so, then methylation is a transformation. If not, then methylation is a 
derivation (in the sense here). 

I'd say transformation - and I think GO would too.  Identity is a tricky issue, 
but it is often important to model and  can, I think, be successfully defined 
for at least some cases.   In the case of cell lineage, cell division is the 
point at which identity shifts - with division producing either two new 
individuals (fission) or one (budding - parent retains identity.)   This fits 
well with biologists intuitions about identity.

For the protein example: If we consider metabolism of some chemical X (which GO 
models) to consist of processes by which X is either synthesised or degraded, 
then we need identity criteria to define whether a process that changes 
chemicals of type X results in a change of identity or not.  I think biologists 
would agree that phosphorylation of a protein does not give it a new identity, 
but that its identity is lost if it is cut into pieces by a protease.  The 
former process is not protein metabolism, the latter is.

Original comment by dosu...@gmail.com on 5 Feb 2015 at 1:29

[GoogleCodeExporter]

Thinking about the cell budding example.  This doesn't fit with the extended 
definidion of derives_from in RO2005.  A cell begins to exist at budding, but 
the parent cell continues to exist and retains its identity.  

A derives_from relation that doesn't only apply when the entity of the left 
hand side of the relationship retains its idenity would be much more broadly 
useful.  Other examples:

The relationship between an exosome and the parent cell from which it buds.
The relationship between a cutting and plant from which it was cut.

Original comment by dosu...@gmail.com on 5 Feb 2015 at 2:23

[GoogleCodeExporter]

I agree with David on Melanie's case.

Michel raises anther tricky case. I'm sure there's literature on this, but I 
don't know it.

The "t3 derives from t1+t2 (an aggregate)" strategy can work. It might be the 
best fit for the zygote case, since both egg and spermatozoon contribute 
(biologically) significant portions of the matter and the zygote can't exist 
without both: "this zygote derives from this (aggregate of) egg and 
spermatozoon".

I don't think this strategy works as well for the tree graft example, since the 
tree can continue to exist without the graft. So I wouldn't use 'derives from' 
here. With just ROCore, maybe the graft becomes 'part of', but full RO might 
have a better relation.

This is a fuzzy term that raises lots of hard questions about identity and 
change. I agree with David that a more general term would be useful. But I'm 
happy with the improvements we've got to the existing term, and I think we 
should leave the bigger questions for another tracker item.

Original comment by ja...@overton.ca on 5 Feb 2015 at 2:35

[GoogleCodeExporter]

Updated proposal:

RO_0001000 derives from
- textual definition: a relation between two distinct material entities, the 
new entity and the old entity, in which the new entity begins to exist when the 
old entity ceases to exist, and the new entity inherits the significant portion 
of the matter of the old entity
- add example of usage: this cell derives from this parent cell (cell division)
- add example of usage: this nucleus derives from this parent nucleus (nuclear 
division)
- add editor note: This is a very general relation. More specific relations are 
preferred when applicable, such as 'directly develops from'.

RO_0001001 derives into
- textual definition: a relation between two distinct material entities, the 
old entity and the new entity, in which the new entity begins to exist when the 
old entity ceases to exist, and the new entity inherits the significant portion 
of the matter of the old entity
- move inverse from ro-edit.owl to core.owl: 'derives from'
- add example of usage: this parent cell derives into this cell (cell division)
- add example of usage: this parent nucleus derives into this nucleus (nuclear 
division)
- add editor note: This is a very general relation. More specific relations are 
preferred when applicable, such as 'directly develops into'. To avoid making 
statements about a future that may not come to pass, it is often better to use 
the backward-looking 'derives from' rather than the forward-looking 'derives 
into'.

Original comment by ja...@overton.ca on 5 Feb 2015 at 2:37

[GoogleCodeExporter]

Fusion is tricky.   For cells fusing to an existing syncytium, it makes more 
sense to say the syncytium retains its identity but gains a nucleus.  Obvious 
example is development of muscle fibers.  I don't hink we should try to make 
such cases fit under a broadened derives from relation.  Better to come up with 
a new abstract (core?) relation.

The apple tree graft  example is even trickier - because two different but 
perfectly reasonably identity criteria are in play: genomic identity and 
identity based on being a single contiguous organism.  I woudn't want to let 
this edge case get in the way of some resolution here though.

Original comment by dosu...@gmail.com on 5 Feb 2015 at 2:44

[GoogleCodeExporter]

This issue was closed by revision r404.

Original comment by ja...@overton.ca on 6 Feb 2015 at 5:17

• Changed state: Fixed

[GoogleCodeExporter]

Commenting on David's contribution on Feb 5:

> So the key question is: is it the same ribosomal subunit after methylation? 
If so, then methylation is a transformation. If not, then methylation is a 
derivation (in the sense here). 

The assumption in BFO circles has been that, for independent continuants, 
derivation is always either fusion or splitting or budding or capturing, that 
is: there is no derivation which involves exactly 1 thing deriving from exactly 
1 thing. Candidate examples, earlier, had been fetus to embryo and human to 
corpse. I have no objection to keeping 'transforms_from' as a relational 
expression to represent such cases. However, note that it was earlier defined 
as a substantial change (thus one entity of type 1 transforms itself over time 
so that it is no longer ot type 1 but rather of type 2). Rather, it is a matter 
of the entity in question having different qualities (and dispositions, etc.) 
in different time periods.

Original comment by ifo...@gmail.com on 9 Feb 2015 at 2:02

[GoogleCodeExporter]

> The assumption in BFO circles has been that, for independent continuants, 
derivation is always either fusion or splitting or budding or capturing, that 
is: there is no derivation which involves exactly 1 thing deriving from exactly 
1 thing. 

I think that's fine.  It certainly fits with the relationships of progeny to 
parent in cell division.  But it also means, presumably, that we need different 
relation where we have 1 thing -> exactly 1 thing. 

> Candidate examples, earlier, had been fetus to embryo and human to corpse. I 
have no objection to keeping 'transforms_from' as a relational expression to 
represent such cases. However, note that it was earlier defined as a 
substantial change (thus one entity of type 1 transforms itself over time so 
that it is no longer ot type 1 but rather of type 2). Rather, it is a matter of 
the entity in question having different qualities (and dispositions, etc.) in 
different time periods.

I guess the question is - what counts as a major change? Phosophrylation of a 
protein might be minor,  but we can certainly classify phosphorylated vs 
unphosphorylated proteins in OWL and model that phosphorylation state as a 
quality.  I think it makes more sense to me to think of this in terms of 
conservation of identity - I don't think there would be any dispute amongst 
biologists that it is the same protein.

Original comment by dosu...@gmail.com on 10 Feb 2015 at 4:16

[GoogleCodeExporter]

On the matter of proteins we have discussed this in PRO and it seems the choice 
is that it becomes a new thing when it is phosphorylated. There are several 
considerations:

1) Smaller chemical structures seem to clearly change identity on constituent. 
If we say that proteins don't then we push the issue towards having a boundary 
at some size molecule between proteins and small molecules. Where will it be? 
note that addition of exactly the same parts to a small molecule will certainly 
change identity (intuitively). Yet there are still places in chemistry where 
identity of small molecules is tracked even though constituents change. Just 
less of it.

2) Like smaller molecules, the changed protein can have radically different 
properties. In some cases phosphorylation cause conformational changes that 
radically change the chemical and functional properties of the molecule, as 
shape is a large determinant of these properties.

3) Certainly we will say something about having a different kind of part - the 
modification group. If we also ascribed this to be a quality then we have to 
worry about making sure the two views are synchronized.

4) The space of modification includes smaller groups like in phosphorylation 
but also larger molecules such as ubiquitin. I expect that the larger the 
modification is the less likely would consider the entity to be the same. And 
consider the other direction - modification which subtract parts. If we cut a 
protein in half the halves neither would be considered to be the same protein 
as the original. If we follow the idea that "small modifications" are qualities 
then PRO is saddled with a difficult problem of deciding in each case of 
modification whether original entity survived or not. I think that will be very 
painful for the developers and for the users.

4) Without a working model of time we have practical issues in describing *any* 
property of a particular *changing*.

It is a bit uncomfortable saying that the old protein ceases to exist and the 
new modified one starts, as our intuitions connect ceasing to exist with 
energetics that are not present. However I believe the same trouble happens 
with small molecules, so this isn't an issue specific to proteins. I think we 
just have to accept that ceasing and beginning to exist are relatively less 
important than other factors that impinge on the issue of identity over time, 
and to remember that ceasing to exist does not mean that matter is destroyed or 
that the change increases the entropy substantially.

I think I will dispute that biologists would undisputedly say it was the same 
protein. It will depend on the type of biology. In addition there are more than 
biologists in the scientific world. Finally, we need more than a judgement of 
sameness vs differentness but also a consideration of what the consequences of 
the choice are. I think this is a very tricky area because intuitive judgements 
will differ and because we don't get particularly good guidance from more basic 
philosophical principles. 

>However, note that it was earlier defined as a substantial change ("thus one 
entity of type 1 transforms itself over time so that it is no longer of type 1 
but rather of type 2"

As stated this principle is not helpful. Any change to an entity yields another 
thing that is of type entity. We could try to modify it so that we mean 
something like the most specific type of the things, but that will be 
problematic as well as over time we tend to learn more and be able to better 
discern more specific types.

Original comment by alanruttenberg@gmail.com on 6 May 2015 at 10:01