Whose model is it anyways - a fundamental barrier to progress in the age of biology?
We are now well into the ‘age of biology’, which is following the ‘age of physics’ that was the 20th century - a time that marked immense theoretical and experimental progress in fundamental physics. This new ‘age of biology’ has been characterized so far by large leaps forward in experimental technique: the dawn of genomics, epigenomics and proteomics to name a few. Because of these inventions, there has been a sea change in the way we collect data: from simple hypothesis driven experiments where the answer was typically a clear yes or no, to very complicated experiments where the answers are much less clear - very similar to the early experiments of Rutherford and colleagues clearly showing alpha particle scattering (eventually yielding the Bohr model of the atom) as compared to the data coming out from the Large Hadron Collider. The ‘age of physics’ saw it biggest gains when theorists like Feynman and Gell-Mann joined the fray, using primarily mathematical methods to understand complex interactions, and to make predictions to drive future experiments. We contend that the ‘age of biology’ can not yield the results it promises until the analogous situation arises - when quantitative, theoretical biological modelling is fully embraced by experimentalists and trialists.
This realization is not new, indeed many funding agencies are now, and have been for some time, pushing to fund research at the interface between theoretical and experimental disciplines. Surprisingly, despite all this, there are still fundamental misconceptions that we feel are affecting how interdisciplinary research is evaluated and, potentially, seriously hampering its progress.
A recent question from an experimental scientist highlights what we think is a fundamental misconception about the relatively new, interfacial, discipline that is mathematical biology.
He asked:
“What information do you need to put into your [mathematical] model?”
While this is a well-meaning question, and one asked in the spirit of interdisciplinary science, we contend that it illustrates a widely held, incorrect, opinion about mathematical models: that they are fundamentally different models compared to experimental ones. While it is true that the mathematical (or in silico) models that we build look quite different than the models used in a biology lab (in vitro and in vivo), we contend that they are a part of the self-same scientific process, that they also are fundamentally distillations of a central model, and that it is only semantics getting in the way.
To clear the air, let us agree that every scientist doing work in biology has a logical construct (a model) in his or her head to explain the phenomena in question - if they didn’t, they couldn’t possibly be doing hypothesis driven research. This model is, if the project is truly a collaborative effort, the same as the one being held in the minds of each other scientist working on the project, whether they be theorists or experimentalists. While each of these scientists may have a slightly different interpretation or depth of understanding, the central model should be the same. What is different then is the approach that each scientist takes to inform the central model - whether it be an in vitro or in vivo experimental approach, or an in silico quantitative approach.
Further, it is exactly these differences that gives each approach its strength. An in silico model can never show a relationship to be biologically relevant the way that an in vivo one can, and experimental techniques run into difficulty when the relationship or interaction between factors is non-linear (as we believe most are in biology), and it is exactly here that quantitative techniques shine.
So a better question would have been:
“What sort of questions has your [mathematical] approach generated about our model that I could test?”
In order to truly be interdisciplinary scientists, and to realize the promise of the ‘age of biology’, we have to realize that we are all part of the same team, seeking the answer to the same question - trying to understand the same model - just with different tools.
**
n.b. I wrote this piece with help from Philip Maini and Alexander Anderson, two of my DPhil supervisors.