Assembly Theory and “Selection”: A Failure in Interdisciplinary Science Communication

This past week a very provocative paper was published in Nature on Assembly Theory. What was provocative, was not just the main idea that was advanced, but also how this idea was presented by the authors.

First let’s start off with a discussion of Assembly Theory. Dislaimer: what follows is my primitive understanding of it, given the published work and an online presentation given by one of the senior authors, Lee Cronin. Also see this interview of with the other senior author, Sarah Walker, on Big Biology, and this excellent post on Johannes Jäger’s blog.

The big idea behind Assembly Theory is to understand objects in terms of how they are constructed. In the process, the authors claim that we can make inferences about both their evolutionary history and the complexity of the system they are part of. All it takes are two object-specific parameters: their assembly index (what is the simplest way to build these objects from primary components) and copy number (how many of that particular object exists in the system). This all boils down to a quantity called “Assembly”. Here is the relevant part of the paper:

With ai being the assembly index for object i, ni being the copy number for object i and NT being the total number of objects in the system.

The idea is interesting. That the complexity of an object, and its copy number (as in how many of these particular objects exist), can tell us something about how that object came to be, is potentially an insightful new idea. Their discussion of how these objects evolve is also illuminating. A given object is assembled from simpler objects, but along each step of the assembly pathway, each precursor object must be plentiful. How do you get from one object to the next is a bit fuzzy (for me at least), but the authors describe this step-wise action as “selection”, where a given plentiful precursor object is modified in a particular manner to become a new object with a higher assembly index. If this modification process becomes widespread, then the new object increases in copy number. Thus, evolution, as the authors conceive it, is a process where a precursor object with a certain assembly index and high copy number gives rise to a new, more complicated, object with a higher assembly index, and a high copy number.

This is all fine except that what jumps out of the page of their manuscript, for anyone who is even casually acquainted with concepts in evolutionary theory, is the use of the word “selection”. I want to spend the rest of this post discussing what “selection” means to someone who studies evolutionary processes and why the Assembly Theory authors’ use of the word becomes an obstacle in having a sensible discussion about their provocative new theory. There is another glaring problem with the paper in that they think Assembly Theory has implications to how we understand how life originated within the metabolism-first frame work, but I’ll save this for a future post (maybe a part II?).

Selection, a primer on how to discuss concepts in an interdisciplinary way

Now we could start this discussion off by simply stating that the authors use the word “selection” in a fundamentally different way than how it is used in evolutionary theory. Fine. But the problem is that from the very beginning of their manuscript, the authors conflate the two concepts. They begin their article by writing:

“In evolutionary theory, natural selection [1] describes why some things exist and others do not [2].”

This sentence makes it clear that the authors do not really understand what section (in the evolutionary theory sense) means. Ironically, their second citation is to a paper written by Ryan Gregory that discusses misconceptions around natural selection. You can’t make this stuff up.

So what is selection in evolutionary theory? Selection is a quantifiable measure of the differential “fitness” between two alleles in a population. The use of the word allele is important as it refers to genetic variants that occupy the same locus in a genome. The word fitness must also be properly defined as it often conflated with other concepts. Fitness is a shorthand for the average reproductive success of individuals carrying one allele over individuals carrying the alternative allele. And when I mean that it is quantifiable, I really mean it. The exact quantifiable term is called the “selection coefficient” and it is a strict comparison of the fitness differences between individuals carrying the two competing alleles. When the selection coefficient is zero, then individuals carrying the first allele have on average exactly as many offspring as individuals carrying the second allele. When the selection coefficient of one allele is 0.01, then individuals with this variant have on average 1% more children than individuals containing the other allele. A selection coefficient of -0.01 means the reverse (1% fewer children).

Does selection explain “why some things exist and others do not”? In part. But this naïve interpretation of evolution, which is widespread across the life sciences, is incomplete. Many aspects of why certain biological entities exist are due, in part, to other phenomena such as historical contingencies, developmental constraints, random fixation of neutral mutations, mutational bias and other non-adaptive processes. For a full discussion of the importance of pluralism in evolutionary thought, please see Gould and Lewontin, 1979. And this is not a small issue. Most biological scientists are way too focused on adaptive processes (i.e. positive selection) that ultimately leads to an unrealistic view of biology where organisms, and their various parts, are seen as being ornately designed, like a Swiss watch. In reality, random fixation of nearly neutral alleles is the predominant mode of evolution in eukaryotes. Many parts in our cells evolved by constructive neutral evolution which tends to over-complexify biological systems. The rampant fixation of nearly neutral mutations explain why eukaryotic life forms are messy, with genomes filled with junk DNA, and processes comprised of overly complicated Rube-Goldberg contraptions (see Palazzo and Kejiou 2022). So yes, selection is important, but it is only part of the story, and ignoring (or more likely being ignorant of) these other aspects of evolution gives an incomplete and unrealistic view of how biological systems came to be the way they are.  

So lets get back to the main topic. Now that we have covered selection, as it is typically used in evolutionary theory, what do the authors mean when they use the word selection? This is unclear to me. But what is clear is that selection, in the sense used in this paper, is not the same as in evolutionary theory. This boils down to the simple fact that their assembled objects (for the most part*) do not increase by replication. The ability of entities to undergo replication is essential if they are to undergo biological selection. Now something in their evolutionary trajectory is occurring. Objects of a certain assembly index are increasing in number at some point. However, this is due to the appearance of a (new?) assembly step that alters a precursor object. It is unclear to me how or why a new assembly step becomes prevalent. Any new assembly step that becomes widespread, I guess, ends up changing the system in a manner that is analogous to an evolutionary change in biology where one allele sweeps to fixation, thereafter permanently altering the genome of the organism. However, without understanding how a new assembly step becomes prevalent, it is unclear how to relate their conception of “selection” with the standard concept of “selection” used in evolution. The fact that they conflate these two processes just confuses the whole issue.

I don’t want to be too negative. Ideas are tools that we use to construct models of the world we live in. New tools, if useful, can generate new insights. In this way, it is sometimes valuable to have naïve individuals come into a field and propose something fresh and potentially innovative. However, if you do this, it is important that you do your homework. It is important that you spend sometime thinking about the field you are attempting to change, and that you chose your words wisely. To paraphrase uncle Bob, you can’t criticize what you don’t understand. And this was very apparent both in the paper and during Lee Cronin’s online presentation.

The Good, the Bad and the Ugly

Beyond misunderstanding the role of selection in evolutionary thought, and conflating his concept of selection with how selection is used in evolutionary theory, there were other statements in Lee Cronin’s talk and in his social media posts that were unnecessary and counter-productive. For example, in his talk he distinctly said that he was going to use “selection” to describe this process and that he didn’t care what other people thought, he wasn’t going to change his mind. So I gather that this whole problem with the word selection has been pointed out to him. He either refuses to address this glaring problem or simply refuses to understand why it is a problem. Another glaring problem was his proclamation over social media that their work constituted a paradigm shift(!), thus explaining why their paper had attracted so much criticism. This just comes across as being arrogant.

If you meet a colleague who has an interesting new idea, which could potentially change our view on a given topic, but who is both careless about the words they use, and clueless about the topic they wish to contribute to, and at the same time arrogant and dismissive of what people are telling them, what would you think? My thought would be that they are their own worst enemy.

I think that these missteps can be fixed. Assembly Theory has some interesting implications and could potentially provide new insights. One aspect that I do not fully grasp (and I would be happy if someone could explain it to me) is exactly how their “selection” process works. Perhaps I’m wrong and it is completely analogous to how evolutionary selection acts. However, it is worthwhile pointing out that one of the reasons that Darwin’s ideas became widely accepted, is that his explanation of how evolution could proceed by natural selection did not require any special metaphysical process. All it required was inherited variation. Since the existence of this variation was apparent (just look at how different each person is and how these traits tend to be inherited), the causal pathway being described by natural selection could be drawn. This was contrary to other proposals, such as Lamarckism, where some mysterious element was acquired during an organism’s life and then transmitted to the next generation. The causal pathway for the inheritance of aquired characteristics required this mysterious element. Yes, Darwin did not have a full understanding of inheritance, and until that was resolved by the modern synthesis, many people still had some lingering questions about whether natural selection was right, but the modern synthesis rooted Darwinian selection in an even more concrete series of causal events. What ever this “selection” process is in Assembly Theory (and I would strongly encourage the authors to rename this process), a deeper understanding of what it is at a fundamental level will be important for its acceptance. But dressing it up as something analogous to natural selection is not the solution (I think).

* I realize that in some cases their objects could be cells or other self-replicating objects. But this particular feature explains why these particular objects behave differently from most other objects.

References:

 Sharma et al., Nature 2023 622:321–328.

Gould and Lewontin, Proceedings of the Royal Society B 1979  205:581-598

Palazzo and Kejiou, Frontiers in Genetics 2022, 13:831068.