Biology likes to borrow

From genetics to neuroscience, biologists have been borrowing mental models and vocabulary from other scientific fields for decades. The idea of a genetic circuit is based on electrical engineering principles; genetic codes are a framework loaned from computer science; and the concept of proteins operating as machines and cells working like factories reflects the influence of modern manufacturing. These conceptual frameworks have shaped the way we speak, write, and think about biology, they are practically unavoidable when looking through anything in between graduate level textbooks and popular science Youtube videos.

Somehow, the most popular frameworks we have for talking about biology are not actually biologically rooted. Many of these metaphors were established during the mid-19th Century, when engineering and early computing were maturing and gaining cultural acceptance, whereas biology was still finding its footing as a “hard” science. The concepts we use to describe the brain are often tethered to the most advanced technology of the era. The brain used to be described as a telegram, but in recent decades the metaphors have shifted from wiring and circuits towards software and hardware. Descriptions of the genome being the “program for life” are similarly inspired by computer science software development.

These metaphors have been passed on through decades of biology without much dispute. They work to a reasonable degree - the genetic circuit framework can be used to predict behaviour for circuits of a limited level of complexity. By mapping wires as functional connections between a transcription factor and its DNA target, switches as the promoters or repressors that transcription factors act upon, we model the output signal as the gene that is expressed as a result. However, it becomes much more challenging to predict the behaviour of gene expression accurately once the circuit has too many components, let alone to predict whole-cell behaviour using this model.

The limits of borrowing frameworks become even clearer when we think about how little predictive power biology has compared to other scientific fields. Physics is the poster child for a predictive science - physics has developed a rigorous set of theories in thermodynamics, electromagnetism, statistical mechanics, and more that has enabled us to build satellites, radio communication, MRI machines, and power plants, and countless other inventions that have been incredibly useful for civilization. In contrast, it remains surprisingly difficult to predict meaningful outcomes in biology using our current conceptual frameworks. We still cannot reliably tell how a random piece of DNA will behave in a cell - will it fold into a compact shape, will it express a protein, will it gain or lose chromatin marks, will it affect the behaviour of other DNA in the cell? Zooming out even further, will it change the cell’s metabolic state or identity? We know the answers for an infinitely small set of DNA sequences, like the tried-and-true Yamanaka factors which “reprogram” cells to revert to a stem-cell like state, but we’re still missing explanations that are truly generalisable across DNA sequences and cells that we have not encountered before. We understand the mechanisms of genetics, epigenetics, transcription factors, and splicing, but these explanations are often fuzzy and highly context-dependent.

The limits of these frameworks have been pointed out various times¹ and biologists continually acknowledge that these metaphors are not entirely accurate, but we don’t have a coordinated effort to find new ways of theorising biological concepts.

Our overreliance on borrowed frameworks will always suffer from the fact that the fields they were borrowed from are not life-like. Engineering is a discipline designed by humans, it is designed to be exact, sophisticated, and well-defined. In engineering, we can decide every parameter of the system with full awareness of the errors and assumptions made in the system. But life sciences is a different ball game because we don’t have complete knowledge over the parameters - it was not designed by us. Biology has highly context-dependent behaviour and emergent properties, while evolution allows for random and undesirable changes as long as the overall function is maintained. That’s why biology is filled with the weird and the wonderful, like genes that have seemingly no function and organs such as the appendix that don’t do anything for us. This can’t be explained by treating the genome as a “blueprint” and proteins as “machines”. These frameworks continue to be useful for designing simple biological systems, but they have limited usefulness in helping us understand what biology is truly doing.

If we want biology to become truly predictable, to have outcomes that are reliable enough to engineer more civilization-shaping technologies, then we need to start developing new mental frameworks for thinking about biology rather than continuing to borrow them. Our best models shouldn’t be tethered to the latest computer science or manufacturing concepts, we need theories of our own that should be strong enough to stand alone.