Re: [Obo-relations] a modest proposal for transformation_of

[Alan Ruttenberg]

Hi Marijke (and Barry),

I wonder if you (or anyone else) could comment on how to apply the framework (or another) to a problem I am currently wrestling with.

The issue is the representation of a molecular process. As it happens it is motivated from immunology work with Lindsay Cowell and Anna Maria Masci, but situation described occurs in many areas of biology. 

In that process, we have at the start a cell membrane in which we have protein complexes embedded. By protein complex I mean a set of non-covalently bound amino-acid-based gene products (folded, and possibly post-translationally modified). As it happens some of this conglomerate pokes from membrane into the extracellular space ("above") and some in to the cytoplasm ("below"). For the sake of argument, let's assume that there are two gene products that are part of this complex.

Another protein comes along and non-covalently binds to the complex. In the process stuff moves around and the part below the membrane surface changes shape, and a part that was previously exposed only to other parts of the complex is now exposed outward. 

Two such complexes can now come together to form a larger complex, being bound together at the point on each that was recently exposed.

Let's assume that all these processes are reversible in normal physiological conditions - there can exist some point later at which all the proteins that make up the dimer last described can float around the cell independently. Let's also assume that there are only certain ways in which they come together (or come apart - perhaps only in different ways) under physiological conditions - if we consider sets of instances of these types of proteins, only certain sets will be bound together in complexes, and there are only some of the many ways we might imagine, that they can come together or apart. 

I am reasonably confident that we can reason about this starting with 6 entities - the amino acid chains that were the result of 6 translation processes. We can say, at least, that the whole material of these entities is part of the final large complex. But after that what? Certainly we *talk* about both parts and aggregates of these, of their shapes or their part's shapes changing over time, and about their functions or their part's functions (over time?), and about processes in which they participate.

However, from the perspective of ontology, what are the entities, what are the types, what are the relations? Are there transformations? Derivations? On what information that I have provided do the answers depend, if any? If there is not one answer, what are some possibilities arrangements that should be considered, and what are some that should be ruled out?

Any hedging you see in the description is out of a desire to avoid being explicit about such things as identity criteria, since that's part of the puzzle.

-Alan

On Thu, Apr 23, 2009 at 7:05 AM, Marijke Keet <ke...@inf.unibz.it> wrote:

Dear all,

I, and so I've heard some other people, are not quite happy with the limitations that one can't really represent temporal stuff, yet some of the RO relations do allude to temporal things. The temporal-DL research is moving forward, but not rapidly in the sense that there will be implementation support for it anytime soon. There are several aspects of the temporality, however, that one can represent in an a-temporal setting. I gave that track a try taking RO's transformation_of relation as an example [1]. The additional constraints one can represent are fully backward compatible with RO's definition; i.e., they do not change the meaning, but instead seek to represent the intention of the relation in a more refined way.

Comments are welcome (if you are a BFO-groupie, then one is advised to read the 2-sentence caution at [2] before commenting).

[1] C. Maria Keet. Constraints for representing transforming entities in bio-ontologies. KRDB Research Centre Technical Report KRDB09-2, Faculty of Computer Science, Free University of Bozen-Bolzano, Italy. April 22, 2009. http://www.inf.unibz.it/krdb/pub/TR/KRDB09-2.pdf
[2] http://keet.wordpress.com/2009/04/23/refining-constraints-on-ro%e2%80%99s-transformation_of-relation/

Best regards,
Marijke

C. Maria Keet
KRDB Research Centre
Faculty of Computer Science
Free University of Bozen-Bolzano
Via della Mostra 4
39100 Bolzano
Italy
tel: +39 04710 161287
fax: +39 04710 16009
email: ke...@inf.unibz.it
web: http://www.inf.unibz.it/krdb/
home: http://www.meteck.org

------------------------------------------------------------------------------
Stay on top of everything new and different, both inside and
around Java (TM) technology - register by April 22, and save
$200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
300 plus technical and hands-on sessions. Register today.
Use priority code J9JMT32. http://p.sf.net/sfu/p
_______________________________________________
Obo-relations mailing list
Obo-re...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obo-relations

[Michel Dumontier]

Hi all,

As a starting point, I've always advocated molecular identity on the basis of molecular structure. Thus non-covalent complexes are formed from molecules that maintain their identity throughout the lifetime of the complex. Just as you and I continue to maintain our identify while bend or stretch, molecules may also change their conformation (but not their configuration!) that yields new functionality. Thus, the use of ro:transformation_of seems most appropriate to describe the temporal relation from one conformation to another.

 

Unfortunately, the preRNA -> mRNA example provided in the ro:transformation_of description [1] is not consistent wrt to molecular identity championed above. I would argue that the configurational changes that occur between a preRNA to an mRNA is best described by ro:derives_from. Furthermore, I believe that ro:transformation_of should be limited to where the changes results in a change in a sub-class membership, and never between disjoint classes. Change of class membership to one that is disjoint with the prior, might be a better description for ro:derives_from, as opposed to the current criteria "biological significant portion of matter", which is clearly subjective and does not act as a consistent guide.

[1] http://www.obofoundry.org/ro/#OBO_REL:transformation_of

-=Michel=-

--
Michel Dumontier
Assistant Professor of Bioinformatics
http://dumontierlab.com

[Nigam Shah]

This proposed way of distinguishing transformation_of and derives_from
is very appealing. What is needed is declaration of the necessary
disjoints in the OBO ontologies to make the proposal work. Otherwise
the two relationships can not be distinguished on that basis.

Nigam.

[Marijke Keet]

Hi All,

I concur with Michel and Nigam regarding the molecular identity. Moreover, the requirement that Michel adds on "changes results in a change in a sub-class membership": that is precisely what came out as constraints I proposed for refining (more precisely: describing and constraining) the transformation_of relation; the argumentation for it is taken from OntoClean basics and temporal conceptual modelling. Currently, such entities can be put willy nilly somewhere in the ontology (mentioned as option 3 in the example in section 4 of the TR, and there are such occurences in extant bio-ontologies), but they shouldn't, because if they were, one cannot guarantee the objects keep their identity during the transformation.

For other parts of the puzzle Alan mentions, I'd have to look into it more, but my first directions are the distinctions between the parts & wholes & aggregates and  instance & individual that has unity (i.e., on what it is to be a whole, see e.g. [1][2]). Regarding the latter, it gives some direction to answer if the protein, the protein complex, and/or dimer are 'loose parts' or individuals in their own right. In addition, the RO has only trasnformation_of, derives_from and a discussion on fusion and fission, but for temporal conceptual modelling, more types of evolution constraints (in their terminology) have been identified (see [3] second part of the paper for an overview); clearly, ontology development and conceptual modelling are different, but it might be useful for broadening one's horizon on variations of relations between changing objects over time.

[1] Peter Simons. 1987. Parts: a study in ontology. Clarendon press, Oxford
[2] Guarino, N., Welty, C.: Identity, unity, and individuality: towards a formal toolkit for ontological analysis. In: Proc. of ECAI-2000, IOS Press, Amsterdam (2000)
[3] Artale, A., Parent, C., Spaccapietra, S.: Evolving objects in temporal information systems. Annals of Mathematics and Artificial Intelligence 50(1-2) (2007) 5-38

Marijke

Michel Dumontier ha scritto:

[Michel Dumontier]

On Fri, Apr 24, 2009 at 4:10 AM, Colin Batchelor <Batch...@rsc.org> wrote:

Michel Dumontier writes:

> Just as you and I continue to maintain our identify while bend
> or stretch, molecules may also change their conformation (but
> not their configuration!) that yields new functionality. Thus,
> the use of ro:transformation_of seems most appropriate to describe
> the temporal relation from one conformation to another.

I concur, certainly for the protein complex case described by Alan.

> Unfortunately, the preRNA -> mRNA example provided in the
> ro:transformation_of description [1] is not consistent wrt
> to molecular identity championed above.

Hmm.  Wearing both my chemist and Sequence Ontology hats, I would say that given that there is nothing inherent in the chemical structure of mRNAs that differentiates them from other kinds of RNA, I don't think molecular identity is appropriate here. 

Or rather, the mRNA sequence may be a ro:transformation_of the preRNA sequence, but the mRNA molecule derives_from the preRNA molecule.  As far as the OBO ontologies are set up, we have a rich description of transcript sequences but a very poor description of molecules.

But RNAs are a tricky edge case between molecular identity and biological identity anyway.

As i recall, pre-RNA indicates an RNA before it is chemically processed into another form... so there is definitely a change in molecular identity (according to structure-based molecular identity). You're right that many other RNA "types" are due to function / role, and so they should appear as more specialized sub-types of RNA, under those conditions.

Cheers,

-=Michel=-

 

best wishes,
Colin.


--
Dr Colin Batchelor MChem MRSC, Team Leader, Informatics R&D
Informatics Department,
Royal Society of Chemistry, Thomas Graham House, Cambridge UK CB4 0WF
batch...@rsc.org t: +44 1223 432280







DISCLAIMER:

This communication (including any attachments) is intended for the use of the addressee only and may contain confidential, privileged or copyright material. It may not be relied upon or disclosed to any other person without the consent of the RSC. If you have received it in error, please contact us immediately. Any advice given by the RSC has been carefully formulated but is necessarily based on the information available, and the RSC cannot be held responsible for accuracy or completeness. In this respect, the RSC owes no duty of care and shall not be liable for any resulting damage or loss. The RSC acknowledges that a disclaimer cannot restrict liability at law for personal injury or death arising through a finding of negligence. The RSC does not warrant that its emails or attachments are Virus-free: Please rely on your own screening.

[Chris Mungall]

On Apr 23, 2009, at 12:14 PM, Michel Dumontier wrote:

> Hi all,
>
> As a starting point, I've always advocated molecular identity on the
> basis of molecular structure. Thus non-covalent complexes are formed
> from molecules that maintain their identity throughout the lifetime
> of the complex. Just as you and I continue to maintain our identify
> while bend or stretch, molecules may also change their conformation
> (but not their configuration!) that yields new functionality. Thus,
> the use of ro:transformation_of seems most appropriate to describe
> the temporal relation from one conformation to another.

What is your definition of molecule -- in particular does it include
large macromolecules? It seems counterintuitive for a single base pair
mutation to cause an entire DNA molecule to cease to exist and be
replaced by a new one.

> Unfortunately, the preRNA -> mRNA example provided in the
> ro:transformation_of description [1] is not consistent wrt to
> molecular identity championed above. I would argue that the
> configurational changes that occur between a preRNA to an mRNA is
> best described by ro:derives_from.

I agree this was a poor choice of example

> Furthermore, I believe that ro:transformation_of should be limited
> to where the changes results in a change in a sub-class membership,

Not sure exactly what you mean

> and never between disjoint classes.

I think the opposite - I think in many cases, transformation_of will
hold between disjoint classes. I'm assuming a definition of
disjointness for continuants as follows

C1 disjoint_from C2 iff not exists i, t such that i instance_of C1 at
t and i instance_of C2 at t

[Michel Dumontier]

On Wed, Apr 29, 2009 at 10:15 PM, Chris Mungall <c...@berkeleybop.org> wrote:

On Apr 23, 2009, at 12:14 PM, Michel Dumontier wrote:

> Hi all,
>
> As a starting point, I've always advocated molecular identity on the
> basis of molecular structure. Thus non-covalent complexes are formed
> from molecules that maintain their identity throughout the lifetime
> of the complex. Just as you and I continue to maintain our identify
> while bend or stretch, molecules may also change their conformation
> (but not their configuration!) that yields new functionality. Thus,
> the use of ro:transformation_of seems most appropriate to describe
> the temporal relation from one conformation to another.

What is your definition of molecule -- in particular does it include
large macromolecules? It seems counterintuitive for a single base pair
mutation to cause an entire DNA molecule to cease to exist and be
replaced by a new one.

In chemistry, a structural difference is sufficient to warrant a distinguished identity. I don't see why it shouldn't follow with chemicals of biological interest. A single base change is also sufficient to completely change the functionality of an nucleic acid, nevermind the evolutionary relationships.

I think the bigger problem is that if you don't have a good definition that you can rely on, then our knowledge will be inconsistently represented. If the DNA is composed of 3 base pairs, and you change one pair, is that significant? if it's 100 and you change 5 is that significant ?   coming up with the *significant* part is what i'd like for us all to think about - there may very well be another useful, non-subjective definition !!!!  Please feel free to suggest some other unambiguous possibilities.. 

 

> Unfortunately, the preRNA -> mRNA example provided in the
> ro:transformation_of description [1] is not consistent wrt to
> molecular identity championed above. I would argue that the
> configurational changes that occur between a preRNA to an mRNA is
> best described by ro:derives_from.

I agree this was a poor choice of example

>  Furthermore, I believe that ro:transformation_of should be limited
> to where the changes results in a change in a sub-class membership,

Not sure exactly what you mean

Ok, the idea is that we use the relation to indicate changes in non-natural (or non-primitive) class membership, such as those that involve dependent continuants, and when class membership is temporally qualified. For instance, the "Enzyme" class could be fully defined (in manchester OWL) as a Protein that hasRole some (EnzymeRole that isRealizedin some BiochemicalReaction). When those conditions are met, our protein instances are realized as members of that class. But when we ask about class membership across the lifetime of the proteins, then we do not generate an inconsistency. this is particularly relevant when membership in multiple classes can occur (for instance the protein is a substrate during some post translational modification). So we obviously want to make sure that if we want to specifically annotate the transformation of a protein from one event to another, that the class membership is not fundamentally disjoint.

Hopefully that helps makes some sense of what i'm getting at.

 

> and never between disjoint classes.

I think the opposite - I think in many cases, transformation_of will
hold between disjoint classes. I'm assuming a definition of
disjointness for continuants as follows

       C1 disjoint_from C2 iff not exists i, t such that i instance_of C1 at
t and i instance_of C2 at t

I agree that we have a need to relate the entity when it takes on disjoint class membership - but i would prefer that we have separate relations to describe changes to 1) non-disjoint and 2) disjoint classes.  It's my personal opinion that if the entity undergoes a change of class membership that is disjoint with its previous, then it is a different entity. However, if people want to use such a relation and still maintain identity, we need to know this assumption to accurately and precisely store that knowledge in OWL or any other formalism that makes the distinction.

-=Michel=-

[Barry Smith]

At 01:22 PM 5/3/2009, Michel Dumontier wrote:


>On Wed, Apr 29, 2009 at 10:15 PM, Chris Mungall
><<mailto:c...@berkeleybop.org>c...@berkeleybop.org> wrote:
>
>
>On Apr 23, 2009, at 12:14 PM, Michel Dumontier wrote:
>
> > Hi all,
> >
> > As a starting point, I've always advocated molecular identity on the
> > basis of molecular structure. Thus non-covalent complexes are formed
> > from molecules that maintain their identity throughout the lifetime
> > of the complex. Just as you and I continue to maintain our identify
> > while bend or stretch, molecules may also change their conformation
> > (but not their configuration!) that yields new functionality. Thus,
> > the use of ro:transformation_of seems most appropriate to describe
> > the temporal relation from one conformation to another.
>
>What is your definition of molecule -- in particular does it include
>large macromolecules? It seems counterintuitive for a single base pair
>mutation to cause an entire DNA molecule to cease to exist and be
>replaced by a new one.
>
>
>In chemistry, a structural difference is sufficient to warrant a
>distinguished identity. I don't see why it shouldn't follow with
>chemicals of biological interest. A single base change is also
>sufficient to completely change the functionality of an nucleic
>acid, nevermind the evolutionary relationships.

bear in mind that the definition of transformation in RO rests on the
relation of identity at the level of particular (as for instance
Marietje on Wednesday is identical to Marietje on Thursday; my laptop
on Friday is identical to my laptop on Saturday). If my laptop is
immersed in water on Friday night, it will undergo a complete change
in functionality, but it will still be one and the same identical
particular object.
BS

><http://dumontierlab.com>http://dumontierlab.com
>
>

[Michel Dumontier]

I couldn't agree more - and i explained this further in the email. But what we are talking about is whether the transformation_of Barry can be made from Human to Elephant. The problem is that the very *definition* of transformation of is incredibly vague and subject to arbitrary interpretation.

-=Michel=-

[Barry Smith]

> >In chemistry, a structural difference is sufficient to warrant a
> >distinguished identity. I don't see why it shouldn't follow with
> >chemicals of biological interest. A single base change is also
> >sufficient to completely change the functionality of an nucleic
> >acid, nevermind the evolutionary relationships.
>
>bear in mind that the definition of transformation in RO rests on the
>relation of identity at the level of particular (as for instance
>Marietje on Wednesday is identical to Marietje on Thursday; my laptop
>on Friday is identical to my laptop on Saturday). If my laptop is
>immersed in water on Friday night, it will undergo a complete change
>in functionality, but it will still be one and the same identical
>particular object.
>BS
>
>
>I couldn't agree more - and i explained this further in the email.
>But what we are talking about is whether the transformation_of Barry
>can be made from Human to Elephant. The problem is that the very
>*definition* of transformation of is incredibly vague and subject to
>arbitrary interpretation.
>
>-=Michel=-

The definition of transformation is in fact very precise:

A transformation_of B = every instance of A was once [identical to]
an instance of B

The problem pertains to the 'identical to'. This is so basic that it
cannot be defined -- there are some areas where we can apply it
perfectly well, for example as regards particular human beings
undergoing normal metabolism; and then there are difficult,
borderline cases. Philosophers have learned that there is no way to
precisify the meanings of basic terms such as 'identity' in such a
way as to remove all such cases; fortunately, again in many areas,
the existence of such cases does not prevent us from using a term
like 'identity' perfectly successfully for the many, many cases which
are not borderline.

The issue, I think, is whether it is appropriate to use
'transformation' as defined above with regard to molecules. Perhaps
it is not appropriate at all. If so, this is not a problem with
transformation, or with its definition, I think.
BS

[Michel Dumontier]

Ok, I think i have made some mistakes in reference to and understanding the RO relations. this is what is said at [1]

[1] http://www.obofoundry.org/ro/#OBO_REL:transformation_of

Relation between two classes, in which instances retain their identity yet change their classification by virtue of some kind of transformation. Formally: C transformation_of C' if and only if given any c and any t, if c instantiates C at time t, then for some t', c instantiates C' at t' and t' earlier t, and there is no t2 such that c instantiates C at t2 and c instantiates C' at t2.

let's see if i have this straight: so the only time this can be asserted is when you know, with no exceptions, that C is a transformation_of C' and that the instance cannot instantiate both classes at the same time. i'm speculating that this could be a transitive relation. The description that accompanies this relation:

When an embryonic oenocyte (a type of insect cell) is transformed into a larval oenocyte, one and the same continuant entity preserves its identity while instantiating distinct classes at distinct times. The class-level relation transformation_of obtains between continuant classes C and C1 wherever each instance of the class C is such as to have existed at some earlier time as an instance of the distinct class C1 (see Figure 2 in paper). This relation is illustrated first of all at the molecular level of granularity by the relation between mature RNA and the pre-RNA from which it is processed, or between (UV-induced) thymine-dimer and thymine dinucleotide. At coarser levels of granularity it is illustrated by the transformations involved in the creation of red blood cells, for example, from reticulocyte to erythrocyte, and by processes of development, for example, from larva to pupa, or from (post-gastrular) embryo to fetus or from child to adult. It is also manifest in pathological transformations, for example, of normal colon into carcinomatous colon. In each such case, one and the same continuant entity instantiates distinct classes at different times in virtue of phenotypic changes.

 i think the first sentence should be rewritten to indicate that since it is true that every oenocyte cell is transformed from a embryonic oenocyte cell, we may then assert that larval oenocyte transformation_of embryonic oenocyte, rather than the current before -> after arrangement.

the thymine-dimer <- thymine dinucleotide example is good (maybe give links to source / wikipedia?)

now let's look at the derivation_of, which unlike transformation_of, puts the formal definition 

C derives_immediately_from C' if and only if: given any c and any t, if c instantiates C at time t, then there is some c' and some t', such that c' instantiates C' at t' and t' earlier-than t and c *derives_from* c'. C derives_from C' if and only if: there is an chain of immediate derivation relations connecting C to C'.

after this description

Derivation on the instance level (*derives_from*) 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. We say that one class C derives_from class C' if instances of C are connected to instances of C' via some chain of instance-level derivation relations.

and the example 

Example: osteocyte derives_from osteoblast. 

also lacks context (add pointer somewhere). I looked up osteocyte and osteoblast to figure out how they were *non-identical*, and here's what i found:

"When osteoblasts become trapped in the matrix they secrete, they become osteocytes" - it pretty much stops contributing to bone formation. i looked around quite a bit, but i don't really see any convincing evidence that they are non-identical. I do certainly see it as transformation_of in this case. maybe a better example is something where there  something either ceases to exist or now comes into existence as a result of a fusion/fission event.

 

The issue, I think, is whether it is appropriate to use
'transformation' as defined above with regard to molecules. Perhaps
it is not appropriate at all. If so, this is not a problem with
transformation, or with its definition, I think.
BS

The challenge is find a case where in every situation membership in class C occurs before C'. I can see it for things like "Product" transformation_of Substrate. or Enzyme transformation_of Protein (so is Protein transformation_of Enzyme ok because they aren't enabling chemical reactions?). These are just specialized terms, so they are fundamentally the same *kind*. so what about more radical transformations where the structure is changed?

let's start with glycolysis - please refer to the image at

http://www.accessexcellence.org/RC/VL/GG/ecb/ecb_images/13_01Glycolysis-Steps_1-5.jpg

In the first step, a phosphate group is added from ATP to glucose, forming ADP and glucose-6-phosphate respectively. In the reaction, the glucose loses a hydrogen but gains a phosphate, whereas the ATP loses a phosphate but gains a hydroxyl group. these both look like "minor" modifications and maybe we can say that they are still the same. At the class level, we would be tempted to assert ADP transformation_of ATP? but, we know that ADP can be formed from ATP or AMP, and that AMP can be formed from ADP or ATP, adenosine + PRPP.... so the complexity of being truly accurate is getting harder nevermind what can be done by a chemist in the lab. So unless you have a pretty complete view of all the possible reactions, making transformation_of assertions on the continuant class may not be feasible or accurate. Pretty much the same applies in using derives_from at the class level.

On binding glucose, the enzyme, hexokinase, changes its conformation from an open form to a closed form, and this puts the ATP and the glucose next to its each for the magic to occur. In this case we can probably say with certainly that "HexokinaseClosedForm" transformation_of "HexokinaseOpenForm", in which would evaluate to some subclass of Hexokinase.

Now, let's take a look at another example in the glycolysis pathway - step 4. Step 4 shows how fructose-6-bisphosphate (FBP) can become split into two forms - dihydroxy acetone phosphate (DHAP) and glyceraldehyde-6-phosphate (GAP). Most people would probably agree that this is a pretty drastic change in molecular structure and would likely lean towards the derives_from relation. This is probably because it appears to be the only object before the reaction occurs.  The proposed mechanism is that F6P makes a covalent bond to the enzyme (Aldolase), losing  a water in the process, then releases GAP. The remainder of FBP is bound to the enzyme, and is later released as DHAP.  This reaction is reversible.

So is it the case that FBP derives_from (DHAP) ? well not without GAP. But what about with the previous step 3, in which FBP is formed from F6P the addition of a phosphate.  So, we would have to say that FBP and ( (derives_from DHAP and derives_from GAP) or (transformation_of F6P) )?  Again, it's possible to generate these molecules with other pathways / reactions, so other derives_from/transformation_of statements might have to be made to remain true (and FWIW the catchall being that X transformation_of Material_entity and X derives_from Material_entity).

Ok, let's look at proteins. I know that the PIR is trying to do their Protein Ontology, and have started to make assertions about the relationship between modified forms and the unmodified form. So similar arguments can be made - that you don't actually know the whether the modified form derives from the unmodified form or from some other form, potentially going through some incompatible derivation reaction. I would be wary of making such assertions without describing the context in which it occurs.

The "biologically significant portion of matter" for *derives_from* is still too vague when it comes to molecules. while we might get good consensus on the above examples, i can probably find lots of examples where 1/4-1/10 of the atoms are being moved around. That will obviously lead to a lot of problems wrt to identity.

My solution was that any difference in structure was an identity change. Then nobody has to debate any given pair of structural transformations. The alternative is to make a social convention that any change over some arbitrary fraction, like 1/4, uses derives_from, otherwise it's transformation_of (if it can be applied as an instance level relation). I'm not that keen on it, but it might be worth investigating the distribution of atom swaps and if shows some bimodal trend as a plausible alternative.

-=Michel=-

[Marijke Keet]

Colin,

for the S, at first impression, CT14' and CT15' would then apply for the
tautomerism. That is, there are, in your email below, 3 structural
isomers, i.e., same molecular formula but different structural formulas
and because of the different structural formula they can have different
functions, melting points, and whatnot. Let's call these classes Cring,
Cketo, and Cenol; and any of the instances /a/ \in Cring can change to
an /a/ \in Cketo or an /a/ \in Cenol and, at a later point in time,
transform back to /a/ \in Cring (or vice versa).
The characteristic that does not change and, in this case at least,
contributes to identity of a subsuming class C (or a Cringketoenol class
as the union of the three), is the molecular formula, so that diachronic
identity can be guaranteed. If we'd put Cring, Cketo, and Cenol willy
nilly in different branches in the taxonomy, then identity of the
molecules across the transformations cannot be guaranteed (if, on the
other hand, you'd have a taxonomy of, say, functions or roles they play
then they could be at different places).

does this make sense w.r.t. the unmentioned example for the tautomerism?

Best,
Marijke

Colin Batchelor ha scritto:
> (Marijke added to the list as I haven't got round to joining bfo-discuss, sorry.)

>
> Michel Dumontier writes:
>
>
>> let's start with glycolysis - please refer to the image at
>>
>
>
>> http://www.accessexcellence.org/RC/VL/GG/ecb/ecb_images/13_01Glycolysis-Steps_1-5.jpg
>>
>

> I like this example, for reasons I will list below.
>
>
>> [...]

>> My solution was that any difference in structure was an identity change.
>>
>

> What I would observe here is that glucose interchanging between ring and chain forms happens all the time in biological systems, not just in that particular reaction. In your connection table-based system, and let us remember that connection tables, which are based on a ball and stick model of molecules, are an administrative convenience for pharma companies rather than a realistic model of a wide range of molecules, we would have ring-glucose and chain-glucose popping into and out of existence all the time, to say nothing of the transition states between the two.
>
> There are more equilibria to consider: acid--base equilibria for example. Amino acid residues will exist in biological systems in a variety of protonation states. I genuinely wouldn't want to argue that simply gaining or losing a hydrogen atom changes the entire protein. One of the strengths of ChEBI is that it represents these equilibria in a number of biologically-relevant cases.
>
> I need to think a bit about how Marijke's S meta-property applies to the ring--keto-chain--enol-chain tautomerism case.

>
>
>> Then nobody has to debate any given pair of structural transformations.
>> The alternative is to make a social convention that any change over some
>> arbitrary fraction, like 1/4, uses derives_from, otherwise it's
>> transformation_of (if it can be applied as an instance level relation).
>>
>

> I don't think either of these alternatives are helpful. I don't believe this kind of thing can be modelled without thinking. I have some sympathy with the molecular identity criterion, but I would argue that it's naïve for the reasons listed above. Further, we are (non-multiplicatively, obviously) implicitly discussing many of these entities qua their participation in a biological process. Take a messenger RNA molecule M in your body and consider an atom-by-atom-identical molecule M' in a comet. The cometary one isn't, I claim, a messenger RNA. It wouldn't be in a universe where eukaryotes have never existed, so why would it be one in this?
>
> Best wishes,
> Colin.

[Colin Batchelor]

Marijke Keet writes:

> for the S, at first impression, CT14' and CT15' would then apply for the
> tautomerism. That is, there are, in your email below, 3 structural
> isomers, i.e., same molecular formula but different structural formulas
> and because of the different structural formula they can have different
> functions, melting points, and whatnot. Let's call these classes Cring,
> Cketo, and Cenol; and any of the instances /a/ \in Cring can change to
> an /a/ \in Cketo or an /a/ \in Cenol and, at a later point in time,
> transform back to /a/ \in Cring (or vice versa).
> The characteristic that does not change and, in this case at least,
> contributes to identity of a subsuming class C (or a Cringketoenol class
> as the union of the three), is the molecular formula, so that diachronic
> identity can be guaranteed.

There is something a bit stronger than just the molecular formula---nearly all of the molecule is the same in both forms---but yes.

(Incidentally, I've misremembered there being a Cenol! Cchain and Cring will do.)

> If we'd put Cring, Cketo, and Cenol willy
> nilly in different branches in the taxonomy, then identity of the
> molecules across the transformations cannot be guaranteed (if, on the
> other hand, you'd have a taxonomy of, say, functions or roles they play
> then they could be at different places).
>
> does this make sense w.r.t. the unmentioned example for the tautomerism?

Yes. This makes a lot of sense.

[Barry Smith]

Michel,
See quick reaction towards end:
BS

><http://www.obofoundry.org/ro/#OBO_REL:transformation_of>http://www.obofoundry.org/ro/#OBO_REL:transformation_of

seems ok to me

>the thymine-dimer <- thymine dinucleotide example is good (maybe
>give links to source / wikipedia?)
>
>now let's look at the derivation_of, which unlike transformation_of,
>puts the formal definition
>
>C derives_immediately_from C' if and only if: given any c and any t,
>if c instantiates C at time t, then there is some c' and some t',
>such that c' instantiates C' at t' and t' earlier-than t and c
>*derives_from* c'. C derives_from C' if and only if: there is an
>chain of immediate derivation relations connecting C to C'.
>
>after this description
>
>Derivation on the instance level (*derives_from*) 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. We say that one class C derives_from class
>C' if instances of C are connected to instances of C' via some chain
>of instance-level derivation relations.
>
>and the example
>
>Example: osteocyte derives_from osteoblast.
>
>also lacks context (add pointer somewhere). I looked up osteocyte
>and osteoblast to figure out how they were *non-identical*, and
>here's what i found:
>

>"When <http://en.wikipedia.org/wiki/Osteoblast>osteoblasts become

>trapped in the matrix they secrete, they become osteocytes" - it
>pretty much stops contributing to bone formation. i looked around
>quite a bit, but i don't really see any convincing evidence that
>they are non-identical. I do certainly see it as transformation_of
>in this case. maybe a better example is something where
>there something either ceases to exist or now comes into existence
>as a result of a fusion/fission event.

I agree with you -- and I hope that this problem will be fixed in the
follow-up to the RO paper, which has finally been initiated. At that
stage we thought that there were examples of derivation which are
1-to-1 as in the osteoblast-osteocyte case; the best example (to me,
at least) is human corpse derives_from human organism. Since then,
however, I have been convinced that the corpse is in fact identical
to the organism which pre-existed it. (There is, in this sense at
least, survival after bodily death.) At the same time I have been
convinced that all the other seeming examples of 1-to-1 derivation do
not work. Hence -- with general agreement, I think -- in the new
version of RO we plan to insist that derivation is always either
n-to-1 or 1-to-n, for n > 1 -- i.e. that all derivation is either
fission or budding, or fusion or capture.
BS

>
>The issue, I think, is whether it is appropriate to use
>'transformation' as defined above with regard to molecules. Perhaps
>it is not appropriate at all. If so, this is not a problem with
>transformation, or with its definition, I think.
>BS
>
>The challenge is find a case where in every situation membership in
>class C occurs before C'. I can see it for things like "Product"
>transformation_of Substrate. or Enzyme transformation_of Protein (so
>is Protein transformation_of Enzyme ok because they aren't enabling
>chemical reactions?). These are just specialized terms, so they are
>fundamentally the same *kind*. so what about more radical
>transformations where the structure is changed?
>

><http://www.accessexcellence.org/RC/VL/GG/ecb/ecb_images/13_01Glycolysis-Steps_1-5.jpg>let's

><http://dumontierlab.com>http://dumontierlab.com
>
>

[Michel Dumontier]

On Tue, May 5, 2009 at 6:34 AM, Colin Batchelor <Batch...@rsc.org> wrote:

(Marijke added to the list as I haven't got round to joining bfo-discuss, sorry.)

Michel Dumontier writes:

> let's start with glycolysis - please refer to the image at

> http://www.accessexcellence.org/RC/VL/GG/ecb/ecb_images/13_01Glycolysis-Steps_1-5.jpg

I like this example, for reasons I will list below.

> [...]

> My solution was that any difference in structure was an identity change.

What I would observe here is that glucose interchanging between ring and chain forms happens all the time in biological systems, not just in that particular reaction.  In your connection table-based system, and let us remember that connection tables, which are based on a ball and stick model of molecules, are an administrative convenience for pharma companies rather than a realistic model of a wide range of molecules, we would have ring-glucose and chain-glucose popping into and out of existence all the time, to say nothing of the transition states between the two.

I suppose you have to ask yourself whether you think two molecular entities that differ even from by a single bond are in fact *different*. What about if they differ by two bonds.... three bonds? four bonds? one bond, two atoms?  so you see how arbitrary it becomes - and applying this kind of rule set is, unfortunately, incoherent. Given that we intend to generate these descriptions of biochemical reactions, etc, computationally, it is not feasible to look them up in ChEBI using some natural language description.

that being said - this approach does *not* preclude one from building an ontology that is *broader* or *narrower* than the criteria i have put forth. For instance, a logic based ontology could assert that closed ring and open ring glucose are members of a more general class of glucose conformers - which, coincidently, can include things like stereoisomers, etc.  Yet in order to do this, we must be able to explicitly mention, by non-ambiguous identity, who all is in this set and/or which criteria they must hold. 

There are more equilibria to consider: acid--base equilibria for example.  Amino acid residues will exist in biological systems in a variety of protonation states.  I genuinely wouldn't want to argue that simply gaining or losing a hydrogen atom changes the entire protein.  One of the strengths of ChEBI is that it represents these equilibria in a number of biologically-relevant cases.

I need to think a bit about how Marijke's S meta-property applies to the ring--keto-chain--enol-chain tautomerism case.

> Then nobody has to debate any given pair of structural transformations.
> The alternative is to make a social convention that any change over some
> arbitrary fraction, like 1/4, uses derives_from, otherwise it's
> transformation_of (if it can be applied as an instance level relation).

I don't think either of these alternatives are helpful.  I don't believe this kind of thing can be modelled without thinking.  I have some sympathy with the molecular identity criterion, but I would argue that it's naïve for the reasons listed above.  Further, we are (non-multiplicatively, obviously) implicitly discussing many of these entities qua their participation in a biological process.  Take a messenger RNA molecule M in your body and consider an atom-by-atom-identical molecule M' in a comet.  The cometary one isn't, I claim, a messenger RNA.

Messenger RNA is a type of RNA that is transcribed from a DNA template, and additionally serves as template for protein synthesis - the discriminant is functional/role based. Both are, however, of the same structural class of RNA, in which the atoms and bonds are indeed identical.

-=Michel=-

 

 It wouldn't be in a universe where eukaryotes have never existed, so why would it be one in this?

Best wishes,
Colin.

DISCLAIMER:

This communication (including any attachments) is intended for the use of the addressee only and may contain confidential, privileged or copyright material. It may not be relied upon or disclosed to any other person without the consent of the RSC. If you have received it in error, please contact us immediately. Any advice given by the RSC has been carefully formulated but is necessarily based on the information available, and the RSC cannot be held responsible for accuracy or completeness. In this respect, the RSC owes no duty of care and shall not be liable for any resulting damage or loss. The RSC acknowledges that a disclaimer cannot restrict liability at law for personal injury or death arising through a finding of negligence. The RSC does not warrant that its emails or attachments are Virus-free: Please rely on your own screening.

[Colin Batchelor]

Barry Smith writes:

> At the same time I have been
> convinced that all the other seeming examples of 1-to-1 derivation do
> not work. Hence -- with general agreement, I think -- in the new
> version of RO we plan to insist that derivation is always either
> n-to-1 or 1-to-n, for n > 1 -- i.e. that all derivation is either
> fission or budding, or fusion or capture.

I've been looking recently at mechanisms of chemical reactions. This all goes back to Ingold's work in the 20s and 30s, but it seems to hold good today.

Most of the reactions are 2-to-1 or 1-to-2 as you say, but there are some concerted processes which I would like to model as 2-to-2. One of them is the SN2 reaction:
http://en.wikipedia.org/wiki/Sn2

If we consider the potential energy surface (not shown), the reactants are in one basin, the products are in another basin, and the transition state is the col or pass through which the system moves. In general I would be wary of admitting transition states to a molecular ontology, but I would be happy to admit intermediates (basins in between cols).

[Barry Smith]

At 11:18 AM 5/5/2009, Colin Batchelor wrote:
>Barry Smith writes:
>
> > At the same time I have been
> > convinced that all the other seeming examples of 1-to-1 derivation do
> > not work. Hence -- with general agreement, I think -- in the new
> > version of RO we plan to insist that derivation is always either
> > n-to-1 or 1-to-n, for n > 1 -- i.e. that all derivation is either
> > fission or budding, or fusion or capture.
>
>I've been looking recently at mechanisms of chemical
>reactions. This all goes back to Ingold's work in the 20s and 30s,
>but it seems to hold good today.
>
>Most of the reactions are 2-to-1 or 1-to-2 as you say, but there are
>some concerted processes which I would like to model as 2-to-2. One
>of them is the SN2 reaction:
>http://en.wikipedia.org/wiki/Sn2

I have no objection to 2-to-2, too.
We can just say: transformation is always 1-1; everything else is derivation.
BS

>If we consider the potential energy surface (not shown), the
>reactants are in one basin, the products are in another basin, and
>the transition state is the col or pass through which the system
>moves. In general I would be wary of admitting transition states to
>a molecular ontology, but I would be happy to admit intermediates
>(basins in between cols).
>
>Best wishes,
>Colin.
>
>DISCLAIMER:
>
>This communication (including any attachments) is intended for the
>use of the addressee only and may contain confidential, privileged
>or copyright material. It may not be relied upon or disclosed to any
>other person without the consent of the RSC. If you have received it
>in error, please contact us immediately. Any advice given by the RSC
>has been carefully formulated but is necessarily based on the
>information available, and the RSC cannot be held responsible for
>accuracy or completeness. In this respect, the RSC owes no duty of
>care and shall not be liable for any resulting damage or loss. The
>RSC acknowledges that a disclaimer cannot restrict liability at law
>for personal injury or death arising through a finding of
>negligence. The RSC does not warrant that its emails or attachments
>are Virus-free: Please rely on your own screening.
>

>------------------------------------------------------------------------------
>The NEW KODAK i700 Series Scanners deliver under ANY circumstances! Your
>production scanning environment may not be a perfect world - but thanks to
>Kodak, there's a perfect scanner to get the job done! With the NEW KODAK i700
>Series Scanner you'll get full speed at 300 dpi even with all image
>processing features enabled. http://p.sf.net/sfu/kodak-com

[Michel Dumontier]

+1  

this works for me -> in the case of our molecular transformations, even the gain or loss of a proton, or the opening and closing of a ring, involves multiple parties,  and is therefore a derivation.

-=Michel=-

[Alan Ruttenberg]

On Tue, May 5, 2009 at 11:23 AM, Barry Smith <phis...@buffalo.edu> wrote:
>

> At 11:18 AM 5/5/2009, Colin Batchelor wrote:
>>Barry Smith writes:
>>
>> > At the same time I have been
>> > convinced that all the other seeming examples of 1-to-1 derivation do
>> > not work. Hence -- with general agreement, I think -- in the new
>> > version of RO we plan to insist that derivation is always either
>> > n-to-1 or 1-to-n, for n > 1 -- i.e. that all derivation is either
>> > fission or budding, or fusion or capture.
>>
>>I've been looking recently at mechanisms of chemical
>>reactions.  This all goes back to Ingold's work in the 20s and 30s,
>>but it seems to hold good today.
>>
>>Most of the reactions are 2-to-1 or 1-to-2 as you say, but there are
>>some concerted processes which I would like to model as 2-to-2.  One
>>of them is the SN2 reaction:
>>http://en.wikipedia.org/wiki/Sn2
>
> I have no objection to 2-to-2, too.
> We can just say: transformation is always 1-1; everything else is derivation.

"everything else"

In the puzzle I posed the other day (http://tinyurl.com/dasyrw), one
of the things buzzing around the back of my head was that it feels
like neither derivation nor transformation is appropriate at the
molecular level.

The problem with derivation is that it posits that after some point in
time the previous entities cease to exist. However, in (perhaps only
some) chemical reactions, at molecular granularity, it seems to me
that entities retain identity even as they come together because they
can later come apart and be what they were before.

I assume that existence is such that there can't be temporal gaps.

To argue against that they are the same entity as before would seem to
me to be a losing argument. (at least I can't figure out a way to win
it)

One might argue that the same (that constituents come together and
apart retaining identity) is true for macroscopic arguments. This
certainly seems to be true in some cases. If I build a car and later
take one of the wheels off, it is the same wheel I put on when I was
building the car.

However for many other entities this will not happen. The orange I eat
will not sometimes appear back as a whole orange. Statistically
speaking, it is possible, but very unlikely. Therefore I conclude that
questions of when identity is lost have something to do with entropy
and thermodynamics.

So back to Barry's "everything else"...

There are certainly changes that involve neither transformation nor
derivation. Even aside from the above examples, if we think that there
are thinks that retain identity and class over time (like person),
that nonetheless change in physical constitution, then we have cases
of materials merging and dissociating (n->1, 1->n) without the ceasing
of existence posited by derives_from, nor change in classification
posited by transformation_of.

In the case of my earlier post - proteins, protein complexes, ligands,
and implicitly chemical reactions (molecular granularity), and in the
case of car parts (some macroscopic granularity), are there not other
relations that are more appropriate than the two we have in hand
currently?

-Alan

[Colin Batchelor]

Michel Dumontier writes:

> I suppose you have to ask yourself whether you think two
> molecular entities that differ even from by a single bond
> are in fact *different*.

What about hydrogen bonds? Some weak covalent bonds are weaker than some strong hydrogen bonds, and in any case hydrogen bonds may well have some covalent character. Your definition of ChemicalBond, "The attractive interaction between atoms.", includes van der Waals forces and other interactions which are not significant at body temperature.

My feeling is that biologists would consider the breaking of all of the hydrogen bonds between two strands of DNA to be a more significant change than the gain or loss of a proton. Energetically it's massively more significant.

Anyway, even taking the bonds that the MDL mol file knows about, consider ozone. Consider isotopically-substituted ozone, 17O16O2. The resonance structures for this are (in SMILES notation):

[17O-][O+]=O and [17O]=[O+]O.

Their bonding is different. Yet they are the same molecule. It's a case of mesomerism. Munchnones (http://goldbook.iupac.org/M04068.html) are the same. Nitro groups are the same. There has to be something to point to that is more subtle than an OWLification of the mol file.

> computationally, it is not feasible to look them up in ChEBI using some natural language description.

If ChEBI had natural language descriptions of (more of) its terms it would be much more useful!

But yes, obviously we want a computable version of ChEBI and I think your contributions have been tremendously helpful.

[Colin Batchelor]

Alan Ruttenberg writes:

> However for many other entities this will not happen. The orange I eat
> will not sometimes appear back as a whole orange. Statistically
> speaking, it is possible, but very unlikely. Therefore I conclude that
> questions of when identity is lost have something to do with entropy
> and thermodynamics.

I wonder whether the important quantity here isn't kT. What counts as a chemical entity at 3 K certainly doesn't count as a chemical entity at body temperature, and very little inside me would count as a chemical entity inside a star.

The only difference between changes that break bonds and changes that rearrange atoms in space but preserve bonding is one of energy, and what we think of as a bond is conditioned by the temperature we do our experiments at.

This is a bit like granularity. Could it even be a different sort granularity?

[jennifer fostel]

totally agree with you about transition states -- not sure they
actually exist outside of textbooks.

[Barry Smith]

there is a good case to be made to the effect
that the existence of artifacts can have temporal gaps:
Poland in the era 1795–1918
your watch when it is lying in bits in different places in the repair shop
BS

[Michel Dumontier]

What happens if you pick up different parts than you had before?  In only a select few cases would the same atoms actually be reincorporated into the molecule... 

-=Michel=-

[Alan Ruttenberg]

There are enough cases to warrant fixing.

[Alan Ruttenberg]

> there is a good case to be made to the effect that the existence of
> artifacts can have temporal gaps:
> Poland in the era 1795–1918
> your watch when it is lying in bits in different places in the repair shop
> BS

I can't reason well about countries. The case of the watch is
interesting, though I note that your language still identifies the
watch with the bits. "your watch when".

What happens to the parts when they are part of the watch. Do they
cease to exist? If not, continuing my question, what is the relation
between the bits and the watch that is part of the family of relations
that include derives_from and transformation_of?

In the case of molecules we aren't dealing with "artifacts". Did you
mean to suggest that the compelling argument doesn't apply in that
case?

-Alan

[Barry Smith]

At 02:00 PM 5/5/2009, Michel Dumontier wrote:

>On Tue, May 5, 2009 at 1:33 PM, Barry Smith
><<mailto:phis...@buffalo.edu>phis...@buffalo.edu> wrote:
>
>At 12:06 PM 5/5/2009, Alan Ruttenberg wrote:
> >On Tue, May 5, 2009 at 11:23 AM, Barry Smith
> <<mailto:phis...@buffalo.edu>phis...@buffalo.edu> wrote:
> > >
> > > At 11:18 AM 5/5/2009, Colin Batchelor wrote:
> > >>Barry Smith writes:
> > >>
> > >> > At the same time I have been
> > >> > convinced that all the other seeming examples of 1-to-1 derivation do
> > >> > not work. Hence -- with general agreement, I think -- in the new
> > >> > version of RO we plan to insist that derivation is always either
> > >> > n-to-1 or 1-to-n, for n > 1 -- i.e. that all derivation is either
> > >> > fission or budding, or fusion or capture.
> > >>
> > >>I've been looking recently at mechanisms of chemical
> > >>reactions. This all goes back to Ingold's work in the 20s and 30s,
> > >>but it seems to hold good today.
> > >>
> > >>Most of the reactions are 2-to-1 or 1-to-2 as you say, but there are
> > >>some concerted processes which I would like to model as 2-to-2. One
> > >>of them is the SN2 reaction:

> > >><http://en.wikipedia.org/wiki/Sn2>http://en.wikipedia.org/wiki/Sn2

> > >
> > > I have no objection to 2-to-2, too.
> > > We can just say: transformation is always
> > 1-1; everything else is derivation.
> >
> >"everything else"
> >
> >In the puzzle I posed the other day

> (<http://tinyurl.com/dasyrw>http://tinyurl.com/dasyrw), one

> >of the things buzzing around the back of my head was that it feels
> >like neither derivation nor transformation is appropriate at the
> >molecular level.
> >
> >The problem with derivation is that it posits that after some point in
> >time the previous entities cease to exist. However, in (perhaps only
> >some) chemical reactions, at molecular granularity, it seems to me
> >that entities retain identity even as they come together because they
> >can later come apart and be what they were before.
> >
> >I assume that existence is such that there can't be temporal gaps.
>
>there is a good case to be made to the effect
>that the existence of artifacts can have temporal gaps:
>Poland in the era 1795–1918
>your watch when it is lying in bits in different places in the repair shop
>BS
>
>
>What happens if you pick up different parts than
>you had before? In only a select few cases
>would the same atoms actually be reincorporated into the molecule...

I can well imagine that artifacts have special features in this respect;
BS

[Alan Ruttenberg]

On Tue, May 5, 2009 at 6:09 PM, Barry Smith <phis...@buffalo.edu> wrote:

> At 02:21 PM 5/5/2009, you wrote:
>>
>> > there is a good case to be made to the effect that the existence of
>> > artifacts can have temporal gaps:
>> > Poland in the era 1795­1918
>> > your watch when it is lying in bits in different places in the repair
>> > shop
>> > BS
>>
>> I can't reason well about countries. The case of the watch is
>> interesting, though I note that your language still identifies the
>> watch with the bits. "your watch when".
>>
>> What happens to the parts when they are part of the watch. Do they
>> cease to exist? If not, continuing my question, what is the relation
>> between the bits and the watch that is part of the family of relations
>> that include derives_from and transformation_of?
>

> remember derives_from is defined at the type level;

There are both class and instance level relations. As usual, I'm interested in the instance level relations.

From the relations paper:

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 sig-

nificant 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. 

> but it seems that we have
> watch derives_from scattered bits
> is an all-some relation

The class level is stronger than the instance level, so from this we can conclude that the bits cease to exist when the watch does. It is this inference that concerns me. It seems wrong. The spring doesn't cease to exist when it gets incorporated into the watch. When the tire is assembled into the car, the wheel doesn't cease to exist.

-Alan

>
>> In the case of molecules we aren't dealing with "artifacts". Did you
>> mean to suggest that the compelling argument doesn't apply in that
>> case?
>

> I am just not sure.
> BS
>
>
>

[Alan Ruttenberg]

On Tue, May 5, 2009 at 12:36 PM, Colin Batchelor <Batch...@rsc.org> wrote:

Alan Ruttenberg writes:

> However for many other entities this will not happen. The orange I eat
> will not sometimes appear back as a whole orange. Statistically
> speaking, it is possible, but very unlikely. Therefore I conclude that
> questions of when identity is lost have something to do with entropy
> and thermodynamics.

I wonder whether the important quantity here isn't kT.  What counts as a chemical entity at 3 K certainly doesn't count as a chemical entity at body temperature, and very little inside me would count as a chemical entity inside a star.

The only difference between changes that break bonds and changes that rearrange atoms in space but preserve bonding is one of energy, and what we think of as a bond is conditioned by the temperature we do our experiments at.

This is a bit like granularity.  Could it even be a different sort granularity?

Doesn't quite feel the same. With granularity we have different perspectives on the same stuff, whereas with temperature you are either hot or cold, so to speak, but not both at the same time. The arrangements of matter can be different at different temperatures (or in solution), and as objects are subjected to these different conditions, changes to them may occur.

-Alan

[Colin Batchelor]

Alan Ruttenberg writes:

> Doesn't quite feel the same. With granularity we have different
> perspectives on the same stuff, whereas with temperature you are
> either hot or cold, so to speak, but not both at the same time.
> The arrangements of matter can be different at different temperatures
> (or in solution), and as objects are subjected to these different
> conditions, changes to them may occur.

Agreed.

The conundrum as I see it is that, disregarding the thermal background, at the molecular granularity, all changes are reversible. Irreversibility only arises when we start looking at ensembles. But if we pick a temperature then we can look at the energy landscape for a particular system and point to relatively-shallow, symmetric barriers which we expect to be recrossed all the time, like tautomerisms, and barriers which are asymmetric and relatively steep on one side which we don't expect to be recrossed. I'm not suggesting this is a good way of modelling this! But I am suggesting that this might be one of the many unstated assumptions in the way that people try to model molecular entities in biological systems.

Best wishes,
Colin.

--
Dr Colin Batchelor MChem MRSC, Team Leader, Informatics R&D
Informatics Department,
Royal Society of Chemistry, Thomas Graham House, Cambridge UK CB4 0WF
batch...@rsc.org t: +44 1223 432280

[Barry Smith]

>Did Korea cease to exist when divided into North and South?
>Would the USA have ceased to exist if the Civil War had ended differently?
>
>
>I'm having trouble figuring out how to reason
>from your examples here. Are you suggesting that
>these are cases of derivation?

I think South Korea is derived from Korea
I think East Germany, God bless its soul, derived from Germany

>I was hoping that I was getting clearer on what
>the problem is, and that you might address it
>more directly. But I'll try again. We have two
>types of relations for representing temporal
>change in entities in an overall process where
>total material is approximately conserved:
>derivation and transformation. According to the
>documentation of these relations, each has
>consequences - in transformation there is a
>class change and in derivation entities cease to exist and come into existence.
>
>Neither of these seem appropriate for
>representing the kind of change where something
>becomes a part and yet retains its identity, as
>evidenced by the ability to have the part later
>become dissociated and be the same thing as it was before.

I think on the instance level (where we can talk
simply of identity, rather than transformation), the matter is clear:
A heart undergoes two successive transplants; the
same heart (instance) in 3 successive ; so
identity. Can you think of a case where the story
you tell above would not realizably be described in terms of identity?
BS

>Given the above, I see two choices currently.
>Either a) such a change is not represented by a
>relation at all, as neither derives_from or
>transformation_of appears appropriate or b)
>There is some other relation to be used.
>
>If the choice is v, the other relation should
>imply (at least) that one of the relata goes
>from being a part to being not a part, and that
>the now-not-part-then-part continues to exist
>and retain identity. Subtypes of this relation
>could be created for the case where the other
>relatum a) comes into existence during the
>transition, as is the case with the synthesis of
>small molecules or b) retains identity during
>the transition, as is the case with cars and
>wheels, or c) Undergoes a transformation (not
>coming up with a good example atm, but I suspect this is a valid case).
>
>There would be a corresponding relation going in
>the other direction, for the now-part-then-not-part case.
>
>-Alan
>
>
>
>BS

[Alan Ruttenberg]

OK. Consequentially Korea no longer exists, nor did Germany, when East Germany came into existence. Nor is the reunited Germany the same entity as the original Germany. These conclusions all courtesy of the portion of definition I quoted earlier. For my sanity, will you affirm that you agree?

We'll ignore the "biologically significant" part of the definition of derivation, presumably we'll change that in the next RO.

 

I was hoping that I was getting clearer on what the problem is, and that you might address it more directly. But I'll try again. We have two types of relations for representing temporal change in entities in an overall process where total material is approximately conserved: derivation and transformation. According to the documentation of these relations, each has consequences - in transformation there is a class change and in derivation entities cease to exist and come into existence.

Neither of these seem appropriate for representing the kind of change where something becomes a part and yet retains its identity, as evidenced by the ability to have the part later become dissociated and be the same thing as it was before.

I think on the instance level (where we can talk simply of identity, rather than transformation), the matter is clear:
A heart undergoes two successive transplants; the same heart (instance) in 3 successive ; so identity. Can you think of a case where the story you tell above would not realizably be described in terms of identity?

No, but that's not surprising - all of our stories (derivations, transformations) are told in terms of (identity and existence over time. What's missing is the relation that summarizes the change. We need relations, I think, that parallel the definition of transformation and derivation, but for different configurations of identity and existence in time. Here's an attempt, following the language in the relations paper.

c assembled_from c1, c2, .. cn 

assembled_from holds between material continuants when one comes into existence at a certain time in such a way that it has the others as parts. Thus we will have axioms to the effect

that from c assembled_from c1 and c2 we can infer that c1 part_of c at t, c2 part_of c at t, etc, and that the spatial region occupied by c contains the spatial regions of c1.. cn at t

c gains_part c1 

gains_part holds between material continuants when one becomes part of the other at a certain time. Thus we will have axioms to the effect that from c gains_part c1 we can infer that c1 part_of c at t, that the spatial region occupied by c contains the spatial region of c1 at t , and that for any e > 0, there exists a time te st. t-te < e and it is not the case that c1 part_of c at te and the spatial region occupied by c does not overlap the spatial region of c1 at te.

c loses_part c1 

loses_part holds between material continuants when one ceases being part of the other at a certain time, but both entities continue to exist. Thus we will have axioms to the effect that from c loses_part c1 we can infer that not(c1 part_of c at t), that the spatial region occupied by c does not overlap the spatial region of c1 at t , and that for any e > 0, there exists a time te st. t-te < e and c1 part_of c at te and the spatial region occupied by c contains the spatial region of c1 at te.

-Alan

[Alan Ruttenberg]

I neglected to include one relation -

breaks_into holds between material continuants when one ceases to exist at a
certain time and some of its parts come to be self standing.
Thus we will have axioms to the effect that from c disassembles_into

c1 and c2 we can

infer that c1 part_of c just before t, c2 part_of c just before t,
etc, that the spatial region
occupied by c contains the spatial regions of c1.. cn just before t,
that c does not exist
after t, and c1, c2.. exist at t.

-Alan

[Alan Ruttenberg]

It's about tracking how entities participate in processes. To me this
seems entirely analogous to derivation and transformation. i.e. if you
have those, you need these.

-Alan

On Fri, May 8, 2009 at 3:58 PM, Werner Ceusters <ceus...@buffalo.edu> wrote:
> Sound all more like processes to me, rather than relations.

[Barry Smith]

Except that 'breaks-into' (or better: is_result_of_disassembly_from
-- since otherwise we lose the all-some relation) sounds just like a
fission subcase of derives_from.
BS

[Alan Ruttenberg]

On Fri, May 8, 2009 at 4:14 PM, Barry Smith <phis...@buffalo.edu> wrote:
>
> Except that 'breaks-into' (or better: is_result_of_disassembly_from
> -- since otherwise we lose the all-some relation) sounds just like a
> fission subcase of derives_from.

Yes, it sounds a little like this, except that the fission case
implies that the entities that are created in the fission process
didn't exist as themselves before the fission process started, whereas
in the case of is_result_of_disassembly_from they did.

Fission: A drop of water splits into two upon bouncing on the ground.
Disassembly: The 3 parts of a yoyo body (side1, side2, post) come apart.

As I understand these, a process might result in there being both
derivation *and* disassembly relations, e.g. in the case of container
made of two parts that snap together holding some material which
expands rapidly causing to come apart with the contents fissioning.

-Alan

[Barry Smith]

At 07:24 PM 5/8/2009, Alan Ruttenberg wrote:

>On Fri, May 8, 2009 at 4:14 PM, Barry Smith <phis...@buffalo.edu> wrote:
> >
> > Except that 'breaks-into' (or better: is_result_of_disassembly_from
> > -- since otherwise we lose the all-some relation) sounds just like a
> > fission subcase of derives_from.
>
>Yes, it sounds a little like this, except that the fission case
>implies that the entities that are created in the fission process
>didn't exist as themselves before the fission process started, whereas
>in the case of is_result_of_disassembly_from they did.

See section VI of http://ontology.buffalo.edu/smith/articles/16Days.pdf

>Fission: A drop of water splits into two upon bouncing on the ground.
>Disassembly: The 3 parts of a yoyo body (side1, side2, post) come apart.
>
>As I understand these, a process might result in there being both
>derivation *and* disassembly relations, e.g. in the case of container
>made of two parts that snap together holding some material which
>expands rapidly causing to come apart with the contents fissioning.

Disassembly, it seems to me, is a subcase of derivation
BS

[Alan Ruttenberg]

2009/5/8 Barry Smith <phis...@buffalo.edu>:

>
> At 07:24 PM 5/8/2009, Alan Ruttenberg wrote:
>
>>On Fri, May 8, 2009 at 4:14 PM, Barry Smith <phis...@buffalo.edu> wrote:
>> >
>> > Except that 'breaks-into' (or better: is_result_of_disassembly_from
>> > -- since otherwise we lose the all-some relation) sounds just like a
>> > fission subcase of derives_from.
>>
>>Yes, it sounds a little like this, except that the fission case
>>implies that the entities that are created in the fission process
>>didn't exist as themselves before the fission process started, whereas
>>in the case of  is_result_of_disassembly_from they did.
>
> See section VI of http://ontology.buffalo.edu/smith/articles/16Days.pdf

Yes. A richer account than we have in RO. In those terms assembly =
unification, disassembly = separation. I don't see the cases
corresponding to gains part/loses part. Here's a summary, in short
hand. -> marks the temporal divide. If you don't see a variable on one
side, the entity it denotes doesn't exist at that time.
The {} is a kind of signature, even shorter version to emphasize the
changes in existence. "," marks the temporal divide. "-" means
something ceases to exist. "*-" means all denoted entities ceases to
exist. "+" means something comes into existence.

derivation {*-,+}
a -> b c
b c -> a

transformation {,}
a,C(a) -> a,C'(a)

gains part {,}
a,b, not has_part(a,b) -> a,b, has_part(a,b)

loses part {,}
a,b, has_part(a,b) -> a,b, not has_part(a,b)

unification/assembly {,+}
a,b -> c, has_part(c,a),has_part(c,b)

separation/disassembly {-,}
c, has_part(c,a),has_part(c,b) -> a,b

>
>>Fission: A drop of water splits into two upon bouncing on the ground.
>>Disassembly: The 3 parts of a yoyo body (side1, side2, post) come apart.
>>
>>As I understand these, a process might result in there being both
>>derivation *and* disassembly relations, e.g. in the case of container
>>made of two parts that snap together holding some material which
>>expands rapidly causing to come apart with the contents fissioning.
>
> Disassembly, it seems to me, is a subcase of derivation

Still don't see it. In terms of the signatures I've listed above,
derivation is {*-,+} and disassembly/separation is {-,}.

-Alan

[Barry Smith]

I guess my proposal is that derivation should be viewed as the
superordinate relation, of which all cases except transformation [and
(perhaps) loses part and gains part] would be subcases.
BS

[Alan Ruttenberg]

So, in that case, the condition that in derivations the previous
entities cease to exist would be dropped, but the condition that a new
entity be created is retained?

The only issue I see with this is that the relation "derives_from" is
already out there and this would be an incompatible change.

-Alan

[Barry Smith]

At 01:31 AM 5/10/2009, you wrote:

>So, in that case, the condition that in derivations the previous
>entities cease to exist would be dropped, but the condition that a new
>entity be created is retained?

yes

>The only issue I see with this is that the relation "derives_from" is
>already out there and this would be an incompatible change.

We have started work on drafting the follow-up publication to
"Relations in Biomedical Ontologies". There are a couple of things
that need to be fixed with the original set of relations, but of
course the most important thing is to add a stable set of new ones.
BS