Terra Incognita of Science

“Drug Development: When Scientists Try to Build Things”

In 1968, Gunther Stent, a prominent biologist and part of the “band” that included Watson and Crick, wrote a famous paper, subsequently followed by a book. In it, he bemoaned the fact that everything there was to know about molecular biology had already been discovered. That there was to be no more progress. [www.nytimes.com/2008/06/16/health/research/16stent.html and www.leeds.ac.uk/heritage/Astbury/bibliography/Stent_1968.pdf]

Of course, he was wrong. If you ask a molecular biologist today, he would tell you that it is only now that we have achieved a relative complete understanding of molecular biology…

And of course he would be wrong. Even today, we probably understand a tiny sliver of molecular biology. Fifty years hence, they will surely wonder how we could have believed what we believe today.

But how does this relate to drug development? Let me take a step back and explain.

When I was applying to colleges, I didn’t really understand the difference between liberal arts colleges, like Harvard, and engineering schools, like MIT.

Later on, I was taught about the distinction between science, such as biology and physics, and engineering, like mechanical or electrical engineering. I was taught that science was the pursuit of knowledge and engineering was the application of the science. Theory vs. practical application. Discovery or the truth vs. invention.

Truth be told, there is a bit of rivalry between scientists and engineers. I don’t know how engineers feel about scientists, but among scientists, there is a bit of disdain for engineers, and even for inventors and tool-makers, as opposed to scientists who pursue pure research. This despite the fact that many major advances in sciences come about when people invent new tools, like microscopes, polymerase chain reaction, and genetic engineering.

One of the very unfortunate things in drug development is that almost everyone in the field is a scientist, not an engineer. This is a problem, because there is very little true, basic science in drug development. It’s almost all applied science, or engineering. We’re not discovering new science; we’re applying it. We’re trying to build things, namely drugs.

And the problem is that we as scientists are not trained to build anything, to apply the science.

What do I mean by that? Well, if you look at how engineers solve problems, they do several things we’re not trained to do. First, they approximate. They figure they don’t have all the data, and that they don’t have time to do all the calculations, so they use rules of thumb. Second, they employ trial an error. Third, they use practical models (now simulations), because they know that systems can be complicated and that the theory they work from can be wrong. I don’t mean mental models. I mean physical models, like little bridges. (I don’t consider mouse and rat models to be practical models – I will explain in another blog, but that’s because they are not practical models meant to facilitate drug development by reflecting the real world but rather scientific models specifically meant NOT to resemble the real world.) Fourth, they consider the entire system. Fifth, they build in a margin of error. If a bridge is meant to carry 100 tons of traffic, they build it to carry 150 tons.

In other worlds, engineers are trained to solve real problems.

The real difference between scientists and engineers is that scientists are trained to take all the data available, and to come up with a hypothesis that explains all the available data. Then, we consider that hypothesis to be the best available approximation of the truth, and we work off that hypothesis. Three dots? Triangle. Five dots? Pentagon.

Scientists are NOT trained to ask, “how much data is missing?” We’re not trained to evaluate the completeness of our data set. We’re trained to cook the best dish we can with the ingredients we have, sort of like the show Iron Chef.

Scientists are also NOT trained to worry about non-conflicting, non-testable data. Often, there is data that neither contradicts nor supports the dominant paradigm. An example would be “junk DNA.” When I was in college, we were taught that over 90% of the DNA was useless, that they had no function. This was called junk DNA. The dominant paradigm was that DNA encoded genes. The junk DNA didn’t encode genes, but at the same time, they didn’t contradict the then-current theory of how DNA and genes work. So we said that they had no function. A similar example is “cell debris” that we now call exosomes. Scientists are trained to ignore a lot of data, because scientific method consists of hypothesis-testing. If something doesn’t lie in the path of a hypothesis, if it’s lying in the side of the road, we don’t swerve our theory. Only if the data contradicts the dominant theory do we change it.

There are four kinds of problems in the world: simple, complicated, complex, and complicated and complex. Simple problems are problems like what to eat for lunch. You just pick an answer. Complicated problems are problems like how to send a rocket to the moon. You have to do a lot of calculations, but there is an answer. Complex problems are problems with incomplete data or fluid situations such that there is no way to calculate an answer. An example is a presidential election. There are too many unknowns and too many variable to be able to come up with a step by step guide to win an election. For a complex problem, you need to take a pragmatic, trial-and-error approach.

Scientists are trained to solve complicated problems, not complex one. We solve for the equation with the data we have, and ignore the terra incognita.

This is why biologists always think we know 90% of all there is to know about something, but then are surprised every ten years or so when we discover the “other 50%” of biology. RNAi, innate immune system, miRNA, exosomes, epigenetics, transposons, and CRISPR are only some of the entirely new systems that we have discovered recently. Doubtless, there will be more. We always know enough to explain 90% of the data that’s in our viewscreen, but we often don’t stop to consider what is not on our viewscreen. It’s like the story about the man looking for his key under the street lamp because that’s where the light is, though that is not where he lost the key.

So the problem is that we as biologists have a theory of how cells work, and how a disease occurs. And the way that modern, target-based drug development works is that based on that understanding, we try to develop a drug. A kinase gene is mutated and this causes cancer. So we develop a molecule to block that kinase. Beta-amyloid plaques develop in people of Alzheimer’s disease. So we try to clear the plaques. Unfortunately, these approaches rarely work as planned. Sometimes it turns out there are alternate molecular pathways. Other times, it turns out that we didn’t even understand that a type of biological process critical to the disease even existed. For example, Angelman’s syndrome is due to epigenetic imprinting. Our understanding of the disease was obviously incomplete when we didn’t even know there was such a thing as imprinting.

So what does this mean? How can we improve drug development? I think there are a handful of things we can do. First, a little humility about how much we actually understand about biology is probably advisable. Second, we might want to take a bit more pragmatic, trial-and-error approach to drug development. Third, we really need to take a step back and look at biological systems, and the emergent phenomena, rather than approaching everything from a reductionist approach (I will blog about emergent phenomena). Finally, I think we should make a real effort to recruit engineers into drug development.

%d bloggers like this: