Big Data Biology: Between Eliminative Inferences and Exploratory Experiments

Abstract

Recently, biologists have argued that data-driven biology fosters a new scientific methodology, namely, one that is irreducible to traditional methodologies of molecular biology defined as the discovery strategies elucidated by mechanistic philosophy. Here I show how data-driven studies can be included in the traditional mechanistic approach in two respects. On the one hand, some studies provide eliminative inferential procedures to prioritize and develop mechanistic hypotheses. On the other, different studies play an exploratory role in providing useful generalizations to complement the procedure of prioritization. Overall, this article aims to shed light on the structure of contemporary research in molecular biology.

What kind of novelties can machine learning possibly generate? The case of genomics

“…Theory-driven practices also imply theory-informed practices; parts of a theory will be used in the theory-driven sense, others as theory-informed. However, in cases such as exploratory experimentation there can be theory-informed practices without theory-driven practices (Waters 2007;Ratti 2015).…”

“…History of molecular biology has led researchers to start investigation about cancer genes selected with ad hoc criteria, such that there was only a handful of genes which have been investigated (Patel et al 2012;Butcher 2003;Weinberg 2014;Ratti 2016) and genes related to cell metabolism were not among the most preferred. Being able to sequence the human genome and have simultaneously data about all genes is a method to generate predictions that our cognitive limitations could not do right away (Dulbecco 1986;Ratti 2015). When we sequence all genes, we are performing an exploratory experiment to overcome our own limitations.…”

“…In this case, ML model by itself provides reasons that are sufficient. Another important instance in which ML has been shown to be important is in (Ratti 2015). Methodologies and tools designed to handle big data sets are used to extend and strengthen eliminative inductive strategies in fields such as cancer genomics.…”

“…In the case of molecular biology, investigating the meaning of those patterns is done with different means than the computational ones. It is the case of several cancer genomics screenings, where we discover cancer genes that we know they might be implicated in cancer because of what the proteins they encode do, but we do not know exactly how, and we have to perform more traditional wet-lab experiments to find out that(Ratti 2015) 9 This is different from saying that something was conceived to do x, and then patterns suggest that it does y, since it is quite the rule in biology that entities or even full-fledged mechanisms are involved in more than one process 10 For a comprehensive account of this episode, seeRatti (2018) …”

Data science and molecular biology: prediction and mechanistic explanation

“…In the last few years, the intersection of data-intensive science and biology have attracted the interest of philosophers, historians and STS scholars (Leonelli 2011;Strasser 2011;Ratti 2015;Boem and Ratti 2016;Stevens 2013; 4 , in particular where cognitive strategies and modes of knowledge production are concerned. However, how data science have shaped the desiderata and the aims of bioscience has not received enough philosophical attention.…”