Classificatory Theory in Data-intensive Science: The Case of Open Biomedical Ontologies

Abstract

Knowledge-making practices in biology are being strongly affected by the availability of data on an unprecedented scale, the insistence on systemic approaches and growing reliance on bioinformatics and digital infrastructures. What role does theory play within data-intensive science, and what does that tell us about scientific theories in general? To answer these questions, I focus on Open Biomedical Ontologies, digital classification tools that have become crucial to sharing results across research contexts in the biological and biomedical sciences, and argue that they constitute an example of classificatory theory. This form of theorizing emerges from classification practices in conjunction with experimental know-how and expresses the knowledge underpinning the analysis and interpretation of data disseminated online.

Acknowledgements

This research was funded by the ESRC as part of the ESRC Centre for Genomics in Society; by the British Academy, as part of a Small Grant on ‘Data-driven Research in the Biomedical and Biological Sciences’; and by the ESRC/Leverhulme Trust grant for the project, ‘How Well Do “Facts” Travel?’ (grant number F/07004/Z). For exceedingly helpful comments on previous drafts, I thank two anonymous referees of this journal, James McAllister, Jane Lomax, Maureen O'Malley, John Dupré, Thomas Reydon, and Mary Morgan. Warm thanks for generative discussions are also due to Barry Barnes, Hans Radder, Giovanni Boniolo, Jim Griesemer, and Werner Callebaut; the ‘facts’ group at the LSE; the participants in the workshops ‘Making Small Facts Travel’ (LSE, March 2008) and ‘Data-driven Research in the Biological and Biomedical Sciences’ (Exeter, April 2010); the audiences of seminars held at ISHPSSB 2007, Egenis, the Department of History and Philosophy of Science in Cambridge, the Konrad Lorenz Institute, EPSA 2009, and the SEMM; and Sue Rhee, who introduced me to this area of research.

Notes

For a sophisticated analysis of the notion of triangulation, see Wylie (2002).

Of course, there have long been expectations in molecular systematics that there would be highly conserved areas of the genome, and that these will serve to indicate relatedness. Data-intensive methods offer new, efficient ways to spot these areas, thus making it possible to explore their significance.

A useful overview of the various purposes for which bio-ontologies are developed can be found in Keet (2010). Thanks to an anonymous referee for pointing this article out to me.

Note that ‘regulates’ is not organized in the parent/child structure. Also, the Gene Ontology is now in the process of incorporating another two types of relations, ‘has_part’ and ‘occurs_in’, and could potentially adopt many other types of relations, as documented by Smith et al. (2005).

Interestingly, Darden quotes Hesse's older work on analogies as supporting her intuition about the novelty of theoretical language (Darden 2006, 150). As I have shown here, Hesse actually changed her mind on this point.

The importance of models in understanding theories has been widely discussed, for instance in recent work by Griesemer (e.g., 2006) and by contributors to de Regt, Leonelli, and Eigner (2009).

In a similar vein, Schaffner (1993) has offered a sophisticated view of ‘middle-range theories’ as devices for knowledge representation and discovery, in which he even briefly explored the usefulness of object-oriented approaches to programming (the ancestors of bio-ontologies) as a way of expressing theoretical commitments. I see his approach as very congenial to mine, yet I am not considering it in detail here because of his different emphasis and targets (the role of lawlike statements in biology and their relations to theories in physics and chemistry).

I thank the editor for inviting me to stress this important point.

This views contrasts with some aspects of the interpretation of ontology realism given by Smith and Ceusters (2010). A paper discussing the parallels and differences between these two views is in preparation.

This view parallels the endorsement of scientific perspectivism in the analysis of data-intensive science presented by Callebaut (2012).