An introduction to chemistry ontology Colin Batchelor, Royal Society of Chemistry firstname.lastname@example.org 2009-07-21
About you • Ontology experience • Chemistry experience • What are you doing at the moment?
An ontological toolkit for chemistry • Classes vs. instances • Regular polysemy • Granularity • What classes should go into the ontology? Ontological dependence and dispositions • Identity • BFO’s independent continuants and chemistry • An actually-existing chemical ontology: ChEBI • The Sequence Ontology • Drugs (if we have time)
Tokens Instances Particulars This particular cat, this particular portion of water, this particular skin cell, this particular molecule of nitrogen Types Classes Universals Kinds Cats in general, portions of water in general, skin cells in general, molecules of nitrogen in general Instances and classes
This alpha-D-glucose molecule here is an instance_of… a D-glucopyranose molecule a D-glucose molecule a glucose molecule an aldohexose molecule an aldose molecule a monosaccharide molecule a sugar molecule a carbohydrate molecule a natural product molecule an organic molecular entity Classes and instances
An arbitrary pyridine molecule (in a bottle) is an instance_of… a pyridines molecule an azaarene molecule a monocyclic heteroarene molecule a mancude organic heteromonocyclic parent molecule Why this is confusing
No superclasses • N-hydroxy-L-aspartic acid is_a hydroxamic acids • hydroxamic acids is_a organic functional classes thereforeN-hydroxy-L-aspartic acid is_a organic functional classes (source: ChEBI, May 2008) This has now been fixed.
Regular polysemy Let’s say I have a banana. If I feed it to a child, it will get banana all down its front. One might have one table made of oak, and another made of banana. Last week it was frosty so I took in a plant from the back garden to the kitchen. It is a dwarf banana, so it is a banana but not a banana banana. If it ever grows any fruit, the fruit will be a dwarf banana banana. As opposed to a banana banana banana.
Regular polysemy: banana bananaFRUIT(COUNT) bananaFRUIT(MASS) (grinding) bananaWOOD (grinding) bananaTREE bananaCONTRASTIVE REDUPLICATION FOCUS Other regular polysemies are available (e.g. colour, leaf, brand names).
Regular polysemy: banana “How do you feel now?” “Banana’d out.” “The banana milkshake is still waiting for the bill.”
pyridine fully-specified name closed-world name (anything not mentioned is a hydrogen atom) a pyridine underspecified name open-world name (anything not mentioned could be anything) Regular polysemy:EXACTvs.CLASS
Regular polysemy:A sixfold classification (from Corbett, Batchelor and Copestake, LREC 2008)
“The IR and Raman spectra show that the metal interacts with the oxygen atom of the amide group and allow the vibrations of the complexed CdX2 to be characterized.” BULK for GRAIN. “Pure americium has a silver and white luster. At room temperatures it slowly tarnishes in dry air.” GRAIN for BULK. Regular polysemy:GRAINvs.BULK
Bulk granularity portion of wine has_participant has_part has_participant water–hydroxide + proton equilibrium portion of water portion of ethanol ethanol–ethoxide ion + proton equilibrium has_grain hydrogen atom water CHEBI:15377 hydroxide CHEBI:16234 proton CHEBI:24636 ethoxide CHEBI:52092 ethanol CHEBI:16236 icao icao has_part icbo icbo oxygen atom has_participant proton transfer from ethanol to ethoxide Molecular granularity proton transfer from ethoxide to ethanol
Chloroform, ethanol Bulk chloroform (as sold) is stabilized with small quantities of amylene or methanol or ethanol. Bulk ethanol (as sold) contains small amounts of benzene.
4. What classes should go into the ontology? Dispositions and ontological dependence.
Natural kinds “To say that a kind is natural is to say that it corresponds to a grouping or ordering that does not depend on humans. We tend to assume that science is successful in revealing these kinds;” A. Bird and E. Tobin, “Natural Kinds” in Stanford Encyclopedia of Philosophy (Spring 2009 Edition), ed. E. N. Zalta, http://plato.stanford.edu/archives/spr2009/entries/natural-kinds/
Bad natural kinds • Molecules containing exactly 21 atoms • Molecules where every atom is from a different element • Portions of material that boil at 300 K and 3 atm. • Molecules that are completely surrounded by water. • The largest molecule in a given beaker.
Good natural kinds • Dienes • Carboxylic acids • Diatomic molecules • Ring-containing molecules • Aromatic molecules … why?
What is a disposition? Says BFO: “A realizable entity that essentially causes a specific process or transformation in the object in which it inheres, under specific circumstances and in conjunction with the laws of nature. A general formula for dispositions is: X (object has the disposition D to (transform, initiate a process) R under conditions C.”
Realizable entities For every realizable entity (dispositions, roles, functions, powers, virtues, tendencies, and so forth) there must be a process. The process must be alienable (there must be no mutual ontological dependence between the bearer of the entity and the process).
Alienable and inalienable processes Alienable processes: • Breathing • Swimming • Sleeping Inalienable processes: • Being somewhere (location) • Pointing somewhere (orientation) • (of pairs of objects) being a given distance apart (relative location) • (of pairs of objects) pointing in different directions (relative orientation)
Ontological dependence X is ontologically dependent on Y if X cannot exist without Y existing. Qualities cannot exist without the entities that exhibit them. Roles cannot exist without their players. Functions cannot exist without their exercisers. Dispositions cannot exist without the entities that realize them.
Mutual ontological dependence (1) George Best cannot exist without George Best’s life existing. conversely George Best’s life cannot exist without George Best existing.
Determination, specialization and dispositions (1) Dispositions are relational entities. They depend ontologically on both their bearer and the conditions in which they are realized.
Determination, specialization and dispositions (2) Material Xhas dispositionDto undergo processM (melting) under conditions pressure P and temperature T. Here the conditions are determinable.
Determination, specialization and dispositions (3) Material Xhas dispositionDto undergo processM (dissolution at ratek) under conditions pressure P and temperature T. Here the conditions are determinable and the process is a quantified specialization of the process of dissolution.
Surefire dispositions vs. tendencies A more thorough account of dispositions: Xhas dispositionDto undergo processPat ratekwith probabilitypunder conditionsC. For surefire dispositions, p = 1.0. For tendencies, p < 1.0. The disposition D(C) and its associated rates kD(C) and probability pD(C) can be complicated mathematical functions of C.
Antidotes, finks and fuses Antidotes: Poison has the disposition to kill people unless the antidote is applied. (Careful: does this mean X has disposition D to undergo process P except where it doesn’t?) Finks: where the conditions C are exactly those that prevent X from undergoing process P. Example: a fuse.
Mutual ontological dependence (2) There are many mutually ontologically dependent dispositions in chemistry. Being an acid is the disposition to donate a proton to (or to receive an electron pair from) a base. Being a base is the disposition to receive a proton from (or to donate an electron pair to) an acid.
Grounding dispositions There are no bare dispositions. All dispositions ought to have a categorical (= quality or independent continuant) basis. Metals conduct because of their band structure (= quality). Dienes take part in Diels–Alder reactions because of their electronic structure (= quality). Objects fall under gravitation because of their mass (= quality).
What grounds many of these natural kinds? Structure or, better, Parthood
Exercise • What dispositions do the qualities inhering in the parts that define these classes ground? CHEBI:22479 amino cyclitol glycosides CHEBI:47788 3-oxo steroid CHEBI:33726 canonical amino acid residue anion CHEBI:25384 monocarboxylic acid CHEBI:33791 gold coordination entity CHEBI:30879 alcohol
Chemical entities and identity criteria • Synchronic (class) Synchronic identity conditions are for different objects i and j at time t. • Diachronic (instance) Diachronic identity conditions are for what may be the same object i at times t1and t2.
Synchronic identity for small molecules What follows is an account of identity criteria as understood by chemists expressed in terms of qualities.
Synchronic identity for small molecules Cyclohexane = A molecule consisting of six carbon atoms and twelve hydrogen atoms with the carbons joined up in a ring. Hexene = A molecule consisting of six carbon atoms and twelve hydrogen atoms with the carbons joined in a chain.
Boats and chairs Chair cyclohexane = A cyclohexane with the carbon atoms in the chair conformation. Boat cyclohexane = A cyclohexane with the carbon atoms in the boat conformation.
Qualities vs. realizable entities Qualities are those entities that have no temporal parts, inhere in objects, and are present (though may change) if the objects they inhere in exist at all. Realizable entities may never be realized. The only way to realize them is as an alienable process.
Qualities are (indirectly) multiply-realizable There are many tests, many processes that can be realized (many dispositions) grounded on qualities, such as: • Triangularity • Being 2 m tall. • Containing six and only six carbon atoms
Qualities that determine synchronic identity • Constitution (of a molecule) • Being cyclic (of a part) • Being linear (of a part) Hence cyclohexane and the hexenes are different synchronically, despite each molecule containing the same number of each kind of atom. Loss or gain of an atom or atoms changes the class of the molecule.