In 2012, Professor Lazebnik published a seminal article titled ‘Can a biologist fix a radio?’ At first glance the title and its contents seem demeaning to the biology profession. However, a close reading shows that this landmark article not only anticipated the rise of systems biology but still has important consequences for the field of bioinformatics in general.
Around the time of writing, Lazebnik’s field of programmed cell death(aka apoptosis) was growing very quickly. So quickly in fact that the large number of publications(est. 10k per year) meant that reviewers were just as overwhelmed as the authors themselves. To be honest I have no idea how a scientist can read and understand one tenth of that number of papers in a year. But, that’s just the beginning.
The terrible part is that key results were not found to be reproducible and with greater evidence models began to fall apart in a seemingly irreparable manner. Naturally, doubt entered the minds of many who were looking for some miracle drug. Moreover, as Lazebnik looked around he noticed that this phenomenon was persistent in many different fields of biology.
Some might say that this is simply because biology is a field with many unknowns and for most that’s the end of the discussion. But, Lazebnik then asks whether the general problem might involve the manner biologists approach problems:
To understand what this flaw is, I decided to follow the advice of my high school mathematics teacher, who recommended testing an approach by applying it to a problem that has a known solution. To abstract from peculiarities of biological experimental systems, I looked for a problem that would involve a reasonably complex but well understood system. …I started to contemplate how biologists would determine why my radio does not work and how they would attempt to fix it.
He then goes through a brilliant thought experiment that surveys the various methods an experimental biologist would employ and their means of describing their findings. After going through the methods of dissection, cataloguing connections, and breaking particular components he concludes that a biologist with their methods of interpreting and analysing the resulting data still wouldn’t have a clue how a radio worked. Indeed, he remarks that a radio like cells and organisms doesn’t simply have connected components but also tuneable components so it may malfunction even if all components are undamaged.
The methodological problem doesn’t have anything to do with the amount of data collected but with the language used by biologists for modelling. He actually says:
…the radio analogy suggests that an approach that is inefficient in analysing a simple system is unlikely to be more useful if the system is more complex.
However, he then describes what appears to be a serious aversion to precise logical models among biologists. In fact, he characterises the average biologist in the the following manner:
In biology, we use several arguments to convince ourselves that problems that require calculus can be solved with arithmetic if one tries hard enough and does another series of experiments.
The rest of the article then focuses on his proposed solution: the development of systems biology for experimental biologists in his field. This basically involves developing powerful predictive models of intra and intercellular interactions using the right level of mathematical abstraction. I must say that he was very prescient for his time as this field has shown great potential since.
The article also helped me appreciate the importance of open source projects like OpenWorm which I have contributed to in the past as these help make sense of the massive amounts of data that come out of laboratories. Clearly, large amounts of data are useless if you don’t have appropriate tools for modelling.