Paradigm shifts no more. During our bioinformatics workshop in Leuven, Roland pointed out that I tend to use the phrase “paradigm shift” too liberally. In fact, the concept of paradigm changes in science was made popular by Thomas Kuhn, an American physicist, historian, and philosopher of science. Kuhn believed that scientists work within a given set of paradigms and believes that they don’t really question them – until everything falls apart. Let me take you on a brief journey through the philosophy of genomics starting with Kuhn’s nemesis, Karl Popper.
Back to high school. I remember that we discussed Karl Popper in my high school philosophy class and that his concept of science was both clear and simple enough for me to remember. According to Popper, the very essence of science is falsifiability. We cannot prove a scientific hypothesis to be true, but we can only exhaust ourselves trying to prove this hypothesis wrong. While this statement might sound slightly dry and theoretical, this actually has profound implications: this is how we work in biomedical science today. For example, if we would like to demonstrate that the ME2 gene predisposes to epilepsy, we need to do the opposite: we aim to show that there is no association (the null hypothesis) and are subsequently surprised if our finding is not in line with our initial assumption. And the level of our surprise is the p-value. This is as good as it gets according to Popper: a surprise that our initial hypothesis of no difference in allele frequencies in cases and controls is quite unlikely. And if we want to get more secure, we have to do the same experiment over and over again. This is the nature of falsification – scientific theories cannot be proven right, they can only be proven wrong. All we can do is generate hypotheses that can then be shattered by the data. I picked ME2 as this gene demonstrates a common dilemma in genomic science: what should we believe if the data is conflicting, if we could falsify the null hypothesis once, but not a second time?
Falsifiability and beliefs. There is sometimes a fine line between scientific hypotheses and beliefs. The theories of Freudian psychoanalysis, for example, look scientific at first glance. However, upon closer inspection, you realize that – at least according to some supporters – there is nothing you can do to prove Freud wrong. Some would even argue that your attempt to prove it wrong is simply a support of the initial theory. If you encounter theories like this, they may be many things, but they are not scientific. They are simply not falsifiable. The same is sometimes true for theories in the field of epilepsy genetics when we feel strongly attached to a particular candidate gene, claiming that the causative variation has simply not been picked up yet in the studies that failed to show an association. Popper’s ideas are quite sobering. All we can do as scientists is to take one hypothesis after the other and prove it wrong. This is the emotionless, neutral scientist who is driven by nothing else but applying the scientific method correctly. You might already realize that this might not be an entirely accurate snapshot of what science is.
Thomas Kuhn. I must admit that I remember Kuhn’s name, but not what he stood for. In contrast to Popper, his concept of normal science and scientific revolution suggests that we usually operate within a framework of paradigms that we don’t question. We simply go with the flow. People with results that question the overall trend are rather accused of having produced flawed data, which is a clear opposite to what Popper would have suggested. At some point, the anomalies are too numerous for the old paradigm to hold and the entire framework collapses. New paradigms are discovered and the scientists start piecing the puzzle together again. Kuhn’s idea of science reads much more negative than Popper’s idealistic version of the constantly falsifying researcher. However, his concept of normal science captures the “group dynamics” that we see in many fields of science: we perform exome sequencing because it is the current method of choice; we aim to be competitive when racing to identify a particular gene. All these phenomena don’t really have a place in Popper’s universe.
Genomic paradigm shifts. We are currently in a phase of transition in the field of genetics that I refer to as the “genetics vs. genomics” transition. Time will tell whether this change is an actual paradigm shift or simply a mild transition. Nevertheless, I believe that there are various concepts that are changing, most of them related to the fact that we become more and more aware of the vast amount of genetic variation in our genomes. Old concepts that find evidence for genetic causation in a small association study, identification of rare variants, or a single de novo mutation are not applicable any more. We have gained a new sense of false positive findings, and it will probably take some time for this idea to fully settle in. As mentioned in previous posts, genetics is transformed from a formerly data-poor field of science to a big data science.
Different paradigms for different scientists. Kuhn was also aware that paradigm shifts and novel levels of “normal science” might asymmetrically apply to different fields of science. This might result in different levels of understanding between related fields. For example, there is an increasing gap between the genetics community and scientists working on functional aspects of the epilepsies, because the genetics field is slowly drifting away from the candidate gene driven approach to broader types of studies such as hypothesis-free genome-wide or exome-wide analyses. It’s simply not easy anymore to find a quick genetic validation for a gene that shows particular epilepsy-related properties on the functional level.
Lessons learned. I misused the term paradigm shift a few times in the past on our blog and felt that I should write a brief explanatory post on what I mean by this. The two philosophies by Popper and Kuhn outline the boundaries of our scientific studies, ranging from the highly methodological and emotionally detached studies of Popper to the trendy normal science of Kuhn that eventually leads us into the next scientific revolution. And we might be experiencing a slow scientific revolution in the field of genomics at the moment.
I wouldn’t really compare Khun’s idea of paradigm shift to the change in methods we currently experience. His ideas were more focused on entire change in models in light of new observations. Like switching from geocentric view of the universe to heliocentric or switching from DNA world to RNA world theories of origin. Even though our methodology is different and small scale studies are not applicable, what we do now is reversely compatible with the old methods. Meaning that you can theoretically replicate your findings in a small scale study and reach the same conclusion. So I’d still be more conservative with the use of the word ‘paradigm shift’. At least in Khun’s terms
Dear Ogan,
thanks for the thoughtful comment. I perfectly agree that Kuhn’s “paradigm shift” would probably be more than our transition from genetics to genomics (if such a transition is actually taking place). However, I also believe that some of the older history of genetics may need to be re-written as some studies would not be considered significant any longer with current day standards. With respect to the epilepsies, this might actually mean that some of genes that were proposed in the past might actually not be implicated at all. Happy to hear your thoughts about this!
Pingback: Three reasons why we need a new genetic literacy to understand epilepsy | Beyond the Ion Channel
Pingback: Building the Epilepsiome | Beyond the Ion Channel