A modelling pattern for multi-track dispositions for life-science ontologies

Abstract

Motivation: This paper extends Röhl & Jansen's (2011) model of dispositions by introducing a relation of parthood between dispositions to formalize multi-track dispositions. Results: We suggest axioms for parthood relations between dispositions and discuss possible applications to life sciences.

INTRODUCTION

Dispositions are realizable entities - properties of material objects that are realized by certain types of processes (Jansen, 2007; Arp, Smith and Spear 2015) . In particular, dispositions are realizable entities that provide their bearers with certain causal powers. The Basic Formal Ontology (BFO) defines a disposition as a realizable entity “that exists because of certain features of the physical makeup of the independent continuant that is its bearer” (Arp, Smith and Spear, 2015, 178) .

The biomedical world is full of such powers: many organisms have a disposition to proliferate; enzymes have the disposition to facilitate, accelerate or inhibit reactions; drugs have both therapeutic dispositions and dispositions for adverse effects. For this reason, Röhl and Jansen (2011) suggested a modelling pattern for dispositions for use in biomedical ontologies. This pattern, however, was restricted to single-track dispositions, i.e. to dispositions that show exactly one kind of realization given one specific bundle of triggering events and/or background conditions. This falls short of the complexity of the biomedical world.

In this paper we suggest a pattern to represent more complex dispositions, taking into account both the possibility of several kinds of triggers as well as of several kinds of realizations. The challenge is that triggering conditions and realizations come in well-defined pairs. For example, a magnet has the disposition to attract another magnet when facing an unlike pole, and it has the disposition to repulse the very same magnet when facing a like pole. Both dispositions are explained by the very same physical structure and laws. We suggest to model this situation by treating these two single-track dispositions as composing a complex disposition, related by a special parthood relation.

DISPOSITION-PARTHOOD

In the following, terms for instances and relations will be written in bold, and classes will be italicized. Consider the fragility of a glass g, which can be manifested by cracking (R1) in case of light shock (T1) and breaking (R2) in case of more serious shock (T2). This can be formalized by the glass having a disposition d1 instance of a universal D1 to R1 in case of T1, and a disposition d2 instance of a universal D2 to R2 in case of T2. Following Röhl and Jansen’s model (2011), we would have: • D1 subClassOf has_triggerD only T1 • D1 subClassOf has_realization only R1 • D2 subClassOf has_triggerD only T2 • D2 subClassOf has_realization only R2 Note first that such axioms do not violate the actualist philosophical stance of BFO. Even if an instance d1 of D1 is never triggered (and thus, there is no t1 instance_of T1 such that d1 has_trigger t1), the class T1 has some instances at some time (even if they do not trigger d1) and therefore it exists. However, “fragility” is arguably one single entity – which may be seen as a disposition d0 instance of a universal D0 to R1 in case of T1 and to R2 in case of T2. In a first attempt, we may want to introduce D0 such that D1 subClassOf D0 and D2 subClassOf D0, characterizing D0 with the following axioms: • D0 subClassOf has_triggerD only (T1 or T2) • D0 subClassOf has_realization only (R1 or R2) This way, however, we would lose a lot of information, like the information that an instance of R1 is always triggered by an instance of T1.

Therefore, we introduce instead a relation d-part_of between dispositions, such that, at the instance level, d1 d-part_of d0, and d2 d-part_of d0. At the universal level, we would have D1 subClassOf d-part_of some D0, and D2 subClassOf d-part_of some D0. This allows to formalize dispositions that are multi-triggers and multi-realizations: D0 can be said to have several types of triggers (T1 and T2) and several types of realizations (R1 and R2) – or, more generally, several tracks (T1 leading to R1 and T2 leading to R2).

One question is whether d-part_of is an authentic parthood relation. There does not seem to be an apriori objection for a parthood relation to hold between dispositions, as parthood does not hold necessarily between material entities only, as underlined by Varzi (2016) . The definition of part_of (BFO_0000050) in the Relation Ontology (RO) (Arp, Smith and Spear, 2015) associated with BFO is generic enough also to comprise relations between dependent continuants; hence we can introduce d-part_of as a sub-relation of part_of.

This relation is reflexive, antisymmetric and transitive, and thus a partial order, like classical relations of parthood. As an example for transitivity, consider the disposition of aspirin to relieve headaches, which is d-part_of the disposition to relieve pain, which in turn is d-part_of aspirin’s overall disposition to physiological effects, of which the first one is also a d-part. Defining has_d-part as the inverse relation of d-part_of, we can then model the magnet example by stating that Magnet subClassOf has_disposition some Ferromagnetic_Disposition, and Ferromagnetic_Disposition subClassOf has_d-part some D1 (resp. some D2). Now the pattern from Röhl and Jansen (2011) can be used to model the two partial dispositions D1 and D2: • D1 subClassOf has_bearer some Magnet • D1 subClassOf has_triggerD only ‘Unlike pole approximation process’ • D1 subClassOf has_realization only ‘Attraction process’ D2 can be formalized along similar lines, with the ‘Like pole approximation process’ as trigger and ‘Repulsion process’ as realization. We consider introducing the additional following axioms (with d0 and d1 being particular dispositions): • If d1 d-part_of d0, then: x bearer_of d1 iff x bearer_of d0. • If d1 d-part_of d0 and d1 has_realization r, then

d0 has_realization r. • If d1 d-part_of d0 and d1 has_trigger t, then

d0 has_trigger t.

Note that not all relations between dispositions can be dealt with in this model. E.g., the disposition of the candle wax to melt surely contributes to the disposition a candle to burn when lighted, but it is not captured by d-part_of, because both dispositions have different bearers. 3

DISCUSSION

If d-part_of is to be a parthood relation, it should also satisfy principles of decomposition and composition (Varzi, 2016) . If PP is the relation of “proper_part_of”, P the relation “part_of” and O the relation “overlaps”, the following variants of decomposition principles are distinguished (Varzi, 2016) : • (Weak supplementation) PPxy → ∃z (Pzy • (Strong supplementation) ¬Pyx → ∃z (Pzy ¬Ozx) ¬Ozx) • (Complementation) ¬Pyx → ∃z ∀w (Pwz ↔ (Pwy ¬ Owx)) • (Atomistic supplementation) ¬Pxy → ∃z (Az ¬Pzy) where “Az” means that z is an atom, i.e., an entity that cannot be decomposed further.

Arguably there are no atomistic dispositions, for both realization and trigger types can always be more fine-grained. However, dpart_of seems to satisfy the principle of complementation, and therefore of weak and strong supplementation.

A more debatable question is whether arbitrary sums of dispositions also do exist. That is, for any dispositions d1 and d2, is there always a disposition that is the mereological sum of d1 and d2? It might be strange to say that a patient not only has the disposition to vomit and the disposition to develop an allergic reaction, but also the complex disposition to vomit or to react allergically. However, this kind of entity might be useful for some computer applications, and it is not absurd to say that they exist as particulars - the same way that there might exist so-called “junk particulars”, such as “the mereological sum of Bush’s right knee and the pain in Clinton’s left leg” (Smith, 2004) . However, such dispositions run counter to the intuition that a bona fide complex disposition should show some kind of unity. Along this line, one can point out that the two dispositions of the patient (to vomit and to react allergically) can vary independently from each other; the patient can lose or gain any of them without losing or gaining the other. The situation is different in the case of the magnet: It is the very same physical structure that leads both to the magnet’s attraction power and its repulsion power. In order to change the one, we need to change the other. Hence it might be a viable strategy to account for the unity of complex disposition in terms of the physical structure that constitutes it, the so-called ‘categorical base’ of a disposition (Röhl & Jansen, 2011) .

As dispositions are continuants, the dispositions of a particular object can therefore change over time. In order to account for this, we suggest the adaption of the temporally qualified continuant approach (Grewe & Jansen, 2013) to formalize such changes. Finally, various probabilities might be assigned to the various tracks of a multi-track disposition, adapting the methodology developped by Barton, Burgun & Duvauferrier (2012 ). 4

APPLICATIONS TO LIFE SCIENCES

This model could be useful in life sciences, in particular to formalize diseases and drugs. First, a disease is formalized as a disposition in OGMS (Scheuermann et al., 2011). However, a disease can have different kinds of disease courses – which are the realizations of the disease. Second, a drug has several dispositions to have various physiological effects, in various situations. In sum, the proposed pattern should enhance expressivity for the modelling of dispositions in biomedical ontologies.

ACKNOWLEDGEMENT

The authors thank Johannes Röhl and three anonymous reviewers for valuable feedback on earlier versions.

Funding: Research of AB was supported by the “bourse de fellowship” of the medical department of Sherbrooke University.

References

Arp , R. , Smith , B. , & Spear , A. D. ( 2015 ). Building ontologies with basic formal ontology . MIT Press, Cambridge MA.

Barton , A. , Burgun , A. , & Duvauferrier , R. ( 2012 ) Probability assignments to dispositions in ontologies . In Proceedings of the 7th International Conference on Formal Ontology in Information Systems (FOIS) , pp. 3 - 14 , IOS Press, Amsterdam.

Jansen , L. ( 2007 ). Tendencies and Other Realizables in Medical Information Sciences. The Monist , 90 ( 4 ), 534 ‑ 554 .

Jansen L. and Grewe N. , ( 2014 ) Butterflies and Embryos: The Ontology of Temporally Qualified Continuants . In Ontologies and Data in Life Sciences (ODLS 2014 ), p. E1 - 5 , Universität Leipzig , Leipzig.

Röhl , J. , and Jansen , L. ( 2011 ). Representing dispositions , Journal of Biomedical Semantics , 2 ( Suppl 4 ), S4 .

Scheuermann , R. H. , Ceusters , W. , & Smith , B. ( 2009 ). Toward an ontological treatment of disease and diagnosis . In Proceedings of the 2009 AMIA Summit on Translational Bioinformatics , p. 116 ‑ 120 , San Francisco CA.

Smith , B. ( 2004 ) Beyond concepts: ontology as reality representation . In Proceedings of the 3rd international conference on formal ontology in information systems (FOIS) , pp. 73 - 84 , IOS Press, Amsterdam.

Varzi , A. ( 2016 ). Mereology . In Edward N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2016 Edition).