Extending the Ontology of Physics for Biology with Thermodynamics Daniel L. Cook1 John H. Gennari1 and Maxwell L. Neal2 1 Department  of  Biomedical  Informatics  and  Medical  Education,  University  of  Washington,  Seattle,  WA,  USA    Center  for  Infectious  Disease  Research,  Seattle,  WA,  USA   2   ABSTRACT flows of material, charge, etc. as in chemical reactions, fluid   flows, etc. Third, OPB:Physical properties (Cook, et al, We  have  extended  the  Ontology  of  Physics  for  Biology  (OPB)  to  rep-­‐‑ resent   the   entities   and   relations   of   classical   thermodynamics.   We   de-­‐‑ 2011) are the observable or inferrable attributes of entities scribe   key   subclasses   of   OPB:Thermodynamic   entity   such   as   OPB:   and processes. And fourth, OPB:Physical dependencies Thermodynamic  property,  and  OPB:Thermodynamic  dependency    in  the   (Cook, et al, 2013) are physical laws (e.g., Ohm’s law) and context  of  the  OPB’s  overall  representational  schema.  We  are  motivat-­‐‑ ed  by  practical  utility  of  energy  bond-­‐‑graph  theory,  a  thermodynamics-­‐‑ constraints (e.g., conservation of mass). based   formalism   used   in   the   domain   of   system   dynamical   modeling   and   analysis   of   biological   physical   processes.   We   also   intend   OPB   to   2   NEXT STEP: THERMODYNAMICS extend   available   upper   biomedical   ontologies   to   encompass   entities   and  theories  of  classical  physics  and  thermodynamics.   To formally represent and constrain models of such dy- namical phenomena, we have extended OPB to explicitly 1    INTRODUCTION represent thermodynamic entities, properties, and dependen- cies. Whereas properly derived system dynamical models The Ontology of Physics for Biology (OPB, 2016) ex- will constrain models to the universal rules of thermody- tends available biomedical ontologies to represent the bio- namics (e.g., conservation of energy, in particular), such physics of biological entities, their observable physical constraints are only implicit in model equations. To explicit- properties and the physical dependencies—the laws of clas- ly satisfy both dynamical and thermodynamic laws and con- sical physics—that determine how property values depend straints, thermodynamics-based energy bond graph model- upon one another. OPB is based on engineering system dy- ing was first described for biological systems (Perelson, namics — the study of stocks and flows of material, charge, 1975), adapted to engineering practice (Karnopp, 1979) and etc. — used to qualitatively explain and to quantitatively has been recently formalized by others (Gawthrop and analyze biological processes over domains such as chemical Crampin, 2014; Lefèvre, et al., 1999) to model biological kinetics, fluid dynamics and electrophysiology and spatial dynamical networks. scales from molecular to organismal. To our knowledge, no We have extended OPB to represent the entities and prin- comparable ontology exists. ciples of classical thermodynamics in support of thermody- Our SemGen application1 uses OPB semantics to derive namic-based computational modeling as well as to extend and analyze SemSim models (semantic simulation) to input, the scant representation of physical and thermodynamical parse, and annotate biosimulation model code. Furthermore, concepts in prevailing upper biomedical ontologies. SemGen can decompose SemSim models into reuseable fragments, merge the fragments as a new SemSim model and export new computational model code . A SemSim model is a light-weight OWL ontology that annotates each variable as an instance of an OPB:Physical property and each equation as an instance of OPB:Physical dependency to create a “property dependency graph” (OPB:Property dependency graph). We have recently applied SWRL rules to SemSim models and OPB to infer qualitative changes of model variables on other property values in the model (Neal, et al., 2016). OPB parses system dynamical abstractions into 4 high- level classes. First, OPB:Dynamical entities are energy- bearing physical continuants such as portions of fluid, chemical, charge. Second, OPB:Dynamical processes are * To whom correspondence should be addressed: dcook@uw.edu Fig. 1 OPB:Thermodynamic entity classes. 1 http://sbp.bhi.washington.edu/projects/semgen 1 Doe et al. 2.1   OPB:Thermodynamical entity thermodynamic entities and laws that govern biological pro- In parallel to OPB system dynamical classes that repre- cesses. OPB is a reference ontology of biophysics that ex- sent stocks/flow of material, charge, etc., OPB thermody- tends available "upper ontologies" (e.g, BFO, GFO), com- namic classes represent stocks/flows of energy and entropy plements domain ontologies such as FMA, GO, ChEBI, and (Figure 1). Thus, an instance of OPB:Mechanical solid has provides a computational resource for annotating biophysi- (via OPB:hasThermodynamicEntity) a portion of cal models and datasets for reuse and integration. OPB:Solid potential energy if stretched or compressed and/or a portion of OPB:Solid kinetic energy if in motion. 2.2   OPB:Thermodynamical property Thermodynamical entities have OPB:Thermodynamical properties: (1) rate properties (e.g., OPB:Energy flow rate, OPB:Entropy flow rate), (2) state properties (e.g., OPB:Energy amount, OPB:Entropy amount) and (3) consti- tutive properties (e.g., OPB:Thermal capacity, OPB:Thermal conductivity). 2.3   OPB:Thermodynamical dependency OPB:Thermodynamical dependencies (Fig. 2) define OPB:Thermodynamical properties in terms of other such properties or in terms of OPB:Dynamical properties (e.g., Fig. 3. Overview of OPB schema for relating thermody- fluid volume or pressure). namic classes (left) to observable state and rate dynamical properties (middle) and the constitutive properties of empir- ical dependencies (right). 2.5   Acknowledgements The authors thank Cornelius Rosse and Peter Hunter. This research was partially supported by the National Institutes of Health, grant R01LM011969. REFERENCES Cook, D. L., F. L. Bookstein and J. H. Gennari (2011). "Physical Properties of Biological Entities: An Introduction to the Ontology of Physics for Biology." PLoS ONE 6(12): e28708. Cook, D. L., M. Neal, F. L. Bookstein and J. H. Gennari (2013). "Ontology of physics for biology: representing physical dependencies as a basis for biological processes." Journal of Biomedical Semantics 4(12): 41. Karnopp, D. (1979). Bond graph techniques for dynamic systems in engi- neering and biology. New York, Pergamon Press.Lefèvre, J., L. Lefèvre J., Lefèvre L., and B. Couteiro (1999). "A bond graph model of chemo-mechanical transduction in the mammalian left ventricle." Simu- lation Practice and Theory Volume 7 (Issues 5–6): 531-552. Gawthrop, P. J. and E. J. Crampin (2014). "Energy-based analysis of bio- chemical cycles using bond graphs." Proc Math Phys Eng Sci 470(2171): 20140459. Neal, M. L., B. E. Carlson, C. T. Thompson, R. C. James, K. G. Kim, K. Tran, E. J. Crampin, D. L. Cook and J. H. Gennari (2015). "Semantics- Fig. 2. OPB:Thermodynamic dependency classes. Based Composition of Integrated Cardiomyocyte Models Motivated by Real-World Use Cases." PLoS One 10(12): e0145621. 2.4   Overview and conclusion Neal, M. L., J. H. Gennari and D. L. Cook (2016). Qualitative causal anal- Figure 3 is an overview of OPB classes showing the scope yses of biosimulation models. International Conference on Biomedical Ontology and BioCreative (ICBO BioCreative 2016), Corvallis, OR, and depth of the OPB’s representation of physics of biologi- USA, CEUR-ws.org Volume 1747. cal processes. We have aimed to, first, extend upper ontolo- OPB (2016) http://bioportal.bioontology.org/ontologies/OPB gies to encompass entities and relations of physics as used Perelson, A. S. (1975). "Network thermodynamics. An overview." Biophys by bioengineers and biophysicists, and now, to encompass J 15(7): 667-685. 2