Monday, August 6, 2018

The biologist's arrow

Guest post by Caroline Bartman

How do we understand biology? “Mutant IDH2 <arrow> 2-hydroxyglutarate <arrow> hypermethylation <arrow> cell proliferation (?),” I scribbled at the top of a paper I read this week. My mind requires linear relationships, direct chains of cause and effect, to retain the findings of a paper I read.

Evidence suggests that this is not how biology in general operates. For example, Pritchard’s ‘omnigenic theory’ synthesizes many years of work to show that most polymorphisms contribute to the total phenotype in a significant but barely detectable way. Identifying each genetic variant that contributes to a phenotype requires many years of costly effort and will culminate with a long list of polymorphisms that incrementally contribute to a phenotype. (Exceptions to this rule- PCSK9- are valuable but rare.) Not only are most contributions miniscule (median contribution of significant height SNPs is 0.00143 meters according to Pritchard), but many polymorphisms play a role in a wide range of traits, by influencing broadly expressed genes. Our search for cause <arrow> effect reveals a tangled thicket of partial causes and modest effects.

Human genetic studies are not the only realm where such complexity dominates. We perform RNA sequencing of wild-type and knockout cells, find a thousand differentially expressed genes, and then focus on a single target gene. We do a screen and follow up on a single hit. It boggles the mind to understand that all of the hits, probably even some below the significance threshold, contribute to that biological process every time it occurs. So we ignore this tangle in order to tell a story, to write a paper, to give a talk that other scientists will appreciate.

This struggle to understand continues as we try to finish a study. Many scientific projects reach an uncomfortable stage where we have a phenotype in hand, a dramatic finding with some relevance to an open biological question, but we require a bit of mechanism for the last figure. (We use the phrase ‘bit of mechanism’ with a half-ashamed laugh.) A bit of mechanism? A handle to give readers, to reassure them that biology is not random, there is a reason for our finding, there is ultimately something to understand? How many of these last figure gambits are quickly abandoned by the relevant subfield as future studies fail to support these ‘mechanisms,’ or change their interpretation beyond recognition?

How do we as humans with limited intelligence, limited bandwidth, limited attention span understand complex biological processes?

Does understanding biology even matter? Don’t we do biology to help patients, to solve problems, to cure disease? But one of the most attractive things about biology for me was that there is a truth outside oneself. Unlike consulting, or writing, or reporting, which are all ways humans can talk about humans, or operate in artificial systems constructed by humans, I believed that science was the way to escape from navel-gazing, the way out of the closed loop. It is not all about humans and feelings and opinions! There are truths outside our selves that we can understand! Just look at ribosomes, or whales, or frogs, or the lac operon and you see a truth that does not require humans as an origin but that humans could find a logic behind. But can we actually understand that logic?

This concern does not lend itself well to selecting and starting a new biological project. The papers that are most beautiful and elegant to me are the simplest. But they leave me with a disquieting feeling that they have achieved beauty by denying complexity.

1 comment:

  1. I would argue that even if "we do biology to help patients, to solve problems, to cure disease", we still need to understand it unless we merely want to interpolate between dense datapoints. Such interpolation is rather limited, even before we consider that datapoints can hardly be dense in high-dimensional space.