Team Meeting: Fridolin & Pierre-Luc

Integration as alignment: From measurement to experimentation and models

Research projects in the life sciences typically rely on a combination of different methods, experimental systems, and often sub-disciplines. Most philosophical work on integration in biology has implicitly focused on an ideal of “global integration”, i.e. the idea that within a given domain any piece of data can be directly and meaningfully related to any other. This is quite demanding and can usually only be achieved by sacrificing quantitative aspects of the data. We show that in the practice of data-intensive biology, integration is often achieved locally, in a way that preserves quantitative information. Thus, there is an important trade-off between what we may call the “scope of integration” and the “depth of integration”. In particular, we identify one strategy of local quantitative integration that seems to be applied – explicitly or implicitly – in a variety of contexts. This strategy consists in the alignment of data sets around anchor points. It is applied not only to integrate data sets of the same kind, for instance when the same experiment is performed in different labs, but also across different experimental modalities, different experimental systems, and even across species. In the latter case integration, while still relying on the same basic strategy, comes very close to practices of modeling, which is another interesting result of our perspective.