Wednesday, May 03, 2017

Debating philosophers: The evolutionary gene

This is the forth post on the Lu and Bourrat paper [Debating philosophers: The Lu and Bourrat paper]. The philosophers are attempting to redefine the word "gene" in order to make epigenetics compatible with current evolutionary theory.

I define a gene in the following way: "A gene is a DNA sequence that is transcribed to produce a functional product" [What Is a Gene?]. This is a biochemical/molecular definition and it's not the same as the definition used in traditional evolution.

Lu and Bourrat discuss the history of the evolutionary gene and conclude,
Hence we suggest a stripped-down notion of the gene that includes only the minimal requirements for it to play the role in formal evolutionary models. Griffiths and Neumann-Held's (1999) conception of the evolutionary gene fits well with our aim. They define the evolutionary gene as a heritable, atomistic unit that causes a difference in phenotype. This definition corresponds to the manner in which formal evolutionary theory treats genes as one of the determinants of trait variance, and also treats genes as the source of inheritance. According to this definition, any physical structure that causes a heritable variation is what we call a 'materialized evolutionary gene.'
Many of you will be puzzled by such a definition but you have to realize that Lu and Bourrat see evolution as synonymous with natural selection. They believe that current "formal evolutionary theory" only recognizes genes that cause a difference in phenotype and that such differences are subject to selection. They don't seem to be aware of neutral or nearly-neutral alleles whose frequency is strongly influenced by random genetic drift. They don't seem to recognize that there are variants in junk DNA that can be fixed or eliminated by evolution. In such cases, it makes little sense to refer to these as evolutionary genes.

As I pointed out in a previous post, Lu and Bourrat tie themselves up in knots trying to make "gene" equivalent to "allele." Here's an example.
Haig (2012), building his notion of phenotypes from the notion of gene, defines a phenotype as 'a gene's effects relative to some alternative', which is not organism-centered, but is coupled to the notion of gene. Two things should be noted. First, as we have mentioned several times, the evolutionary gene is defined by its heritable effects on the phenotype. Hence, to define the phenotype as 'a gene's effects' corresponds well to the concept of evolutionary gene. Second, the requirement of the existence of some alternative seems to mean that if there is no alternative, then there is no phenotype. This is not as problematic as it may first seem to be for three reasons. First, in Johannsen's original definition, phenotypes refer to distinguishable 'types' of organisms, which implicitly assumes a comparison is being made. Second, evolution, as it is classically understood, only occurs when the target population satisfies the condition of variation (Lewontin, 1970). Even in the limit cases where there is no variation in the population at a particular time—because for instance, one variant invaded the population—heritable variations are regularly produced. Finally, under the manipulationist account, causation can only be established when at least two alternatives are compared. Thus, we regard the existence of alternative phenotypes in an evolutionary context as a reasonable assumption.

Inspired by Haig's definition, we define the phenotype of an evolutionary gene as everything that the gene makes a difference to when compared to another gene.
I just don't understand why philosophers put themselves in such a bind by ignoring scientists. Evolution is all about alleles and, by definition, alleles are variants at a certain position in the genome. Evolution is the change in frequency of alleles in a population. If you don't have different alleles then there's no evolution. Alleles are not genes. They are not evolutionary genes nor any other kind of genes. Different alleles at the same locus may or may not produce a difference in phenotype.1 That doesn't make any difference. Evolution will happen regardless of phenotype.

This is not rocket science. Philosphers who want to become experts on evolution can start with any undergraduate textbook on evolutionary biology in order to understand that alleles are important and there's more to evolution than natural selection.

Allow me to help out by quoting Douglas Futuyma from the 2nd edition of Evolution (p. 220).
Fundamental Principles of Genetic Variation in Populations
We embark now on our study of genetic variation and the factors that influence it—that is, the factors that cause evolution within species. The definitions, concepts, and principles introduced here are absolutely essential for understanding evolutionary theory. We begin with a short description of these ideas, followed by an explanation of a very important formal model.

At any given gene locus, a population may contain two or more alleles that have arisen over time by mutation. Sometimes one allele is by far the most common (and may be called the WILD TYPE), and the others are very rare; sometimes two or more of the alleles are each quite common. The relative commonness or rarity of an allele—its proportion of all gene copies in the population—is called the allele frequency (sometime imprecisely referred to as the "gene frequency")....

An alteration of the genotype frequencies in one generation will usually alter the frequencies of the alleles carried by the population's gametes when reproduction occurs, so the genotype frequencies of the following generation will be altered in turn. Such alteration, from generation to generation, is the central process of evolutionary change. However, the genotype and allele frequencies do not change on their own; something has to change them. The factors that can cause the frequencies to change are the causes of evolution.
Two of those causes are random genetic drift (Chapter 10) and Natural Selection (Chapter 11).

I'm not sure that Lu and Bourrat understand current evolutionary theory. But they are philosophers of biology and they are writing a paper about evolutionary theory. They are supposed to be experts.

1. Some colleagues don't like to use the word "alleles" to refer to variants that are not part of genes.


  1. Talking about "genes" rather than "alleles" sure leads to weird nonsense. How does a gene cause a difference in phenotype? Difference from what? Very odd.

    1. Actually this is another case where microbes are different. While it is true that, say, humans may have all the same genes as each other and differences are simply due to differing alleles of those genes, this isn't the case in microbes. Pathogenic strains of E. coli literally have genes giving them that phenotype that don't exist in commensal strains.

    2. @Jonathan Badger
      Are microbes really different in this respect? New genes are being generated in or transferred among all organisms, and bacteria have lots of allelic variation. What makes you say that?

    3. Can we talk about "alleles" for bacteria? It is true that there are mobile genes, but isn´t the term allele reserved for eukaryotes?

    4. @Jarle Kotsbak
      No, it is not a reserved term and bacteria have allleles. Alleles are alternative versions of a gene occupying a given locus. The only difference is that a single bacterium, being haploid, has only one allele at a given locus, whereas we can have two. A POPULATION of bacteria will have allelic variation at several genetic loci.

    5. Yes, bacteria have alleles in the sense you can different versions of a gene at a given "locus", but my point is that in bacteria this isn't the only source of variation -- it isn't really meaningful to speak of loci when there is literally no version of a given gene in another strain. The notion of loci (and classical genetics in general) were developed with the idea of organisms where the only difference was in versions of genes, not presense/absence of genes.

    6. @Jonathan Badger
      Yes, I appreciate that. But new genes arise in all organisms so this doesn't really set microbes apart.

    7. Presumably new animal genes must arise at some point (although humans have basically the same set of genes as every other great ape, although there's been some differences as to gene location, so it isn't a very rapid process). But we've sequenced thousands of human genomes at this point. Have we found even a single gene in one of them that isn't in all the rest? Obviously a lot of allelic variation. When you sequence bacteria of the same species, you might get 20% of the genome being completely different.

    8. That is a neat fact. But I wonder whether that is due to the species definition in bacteria rather than any differences in the turnover of genes.

    9. But I wonder whether that is due to the species definition in bacteria

      What *is* the species definition in bacteria?

    10. @judmarc: Most commonly ecotypes, IIRC. Which isn't a species definition, but (as the name implies) a type. Of course there are also still quite a few people who have the cynical species concept ("a species is whatever a competent scientist (like me for instance) decides is a species". I.e a Potter Stewart kind of deal). As far as I can tell this is mostly due to the suspicion that a decent species concept isn't that useful for labeling petri dishes. As far as the data goes it's still possible that all bacteria are a single highly polymorphic species. Or that each becaterium is a species of its own (a single-organismned species).

    11. @Judmarc

      I don't know but had a hunch bacterial species were somewhat broader defined than in macroscopic organisms. A quick scan of wikipedia strengthens that suspicion:

      Escherichia coli is a badly classified species as some strains share only 20% of their genome. Being so diverse it should be given a higher taxonomic ranking

      Yes, that will give quite some polymorphisms in gene presence/absence.

    12. @ Jonathan Badger

      Many human genes have copy number variation. This is where the new genes (or at least paralogous gene families) come from.

    13. @ Simon Gunkel @ judmarc

      I also struggle with bacterial species definition. Ecotypes definitely are used by working microbiologists, and maybe that kind of fits with Van Valen's ecological species concept.

      "A species is a lineage (or a closely related set of lineages) which occupies an adaptive zone minimally different from that of any other lineage in its range and which evolves separately from all lineages outside its range"

    14. Simon, I love the Potter Stewart reference. :-)

  2. “Evolution is all about alleles and, by definition, alleles are variants at a certain position in the genome.”

    Larry, apart from philosopher bashing and language, your point seems to be that Lu and Bourrat fail to take molecular evolution into account because they define evolutionary change as a change in the frequency distribution of variants that differ in phenotypic effect, rather than of variants at a certain position in the genome. Right?

    It strikes me that you point to an important conceptual problem in contemporary evolutionary theorizing, that never occurred to me before. Thanks very much for pointing this out!

    However, I don't think it is fair to blame Lu and Bourrat for overlooking that problem on the ground that they miss something that can be found in any undergraduate textbook on evolutionary biology and I don't think that this failure detracts anything from their paper: the problem seems widespread and Futuyma’s exposition exemplifies it too, as is clear from your quote form the second edition were ‘allele frequency’ is defined as the proportion of gene copies in the population.

    According to the third edition (2013) (I have the second one too but it is in the attic) an ‘allele is a particular form of a gene usually distinguished from other alleles by its effect on the genotype’ (p. 218).

    It is true that Futuyma allows for the possibility that different alleles of the same gene occasionally do not differ in their effect on the phenotype but that doesn’t solve your problem, for that problem is that alleles are defined as gene-variants, while a lot of molecular evolution concerns changes in parts of the genome that aren’t parts of genes.

    This problem occurs if genes are defined in terms of their phenotypic effect (as Lu and Bourrat do), but also if they are defined in terms of their role in development (as you do).

    As you yourself explained in several blogposts (e.g. this one) current biology employs at least two different notions of gene. This was first pointed out by Lenny Moss (2002), as you mention in these posts. Moss refers to them as Gene-P en Gene-D and I will adopt these terms for the sake of this argument (sorry Paul). P-genes (Mendelian genes) are defined in terms of phenotypic effects. D-genes in terms of their role in development.

    If genes are defined in the latter way, differences in regulatory sequences don’t count as different gene variants (for those sequences don’t count as d-genes). This is one of the reasons many philosophers as well as historians of biology think that the notion of gene relevant to evolutionary biology isn’t the developmental one, but the phenotypical one, for this solves the problem with regulatory sequences.

    As I read your post above, it points to a new problem (new in the sense that it has escaped the attention of both philosophers, historians and evolutionary biologists) namely the problem that any definition of evolution in terms of a change in frequencies of gene variants fails to view changes in junk DNA as evolutionary changes.

    So if changes in junk DNA are to count as evolutionary changes and evolutionary changes are changes in allele frequencies, alleles cannot be defined as gene variants (as all evolutionary textbooks in my bookcase do!).

    Your proposal to define allele variants in terms of their position in the genome might solve that problem, but it might require a lot of work to integrate this new definition of allel (lets call it 'Allel-D' ;-) ) in current evolutionary thinking.

    1. Arno-
      You helped with a confusion I was having- I think the neutral theory has to apply to the entire DNA sequence, but when I read the term ‘allele’ is often used to mean ‘altered gene’. It is a confusion, not just mine.

      How about this-
      Evolution is changes in DNA sequences and happens all the time everywhere there is life (due to imperfect replication and mutation).
      Occasionally we see the results of evolution in a change in phenotype.

    2. I think of alleles as variations in base-pair sequences in the given region of the genome. Variations in protein-coding, regulatory, junk, or any other kind of DNA sequence are all alleles of those DNA sequences.

      We mostly recognize alleles by variations they cause in the phenotype, so we mostly miss the majority of the variants, but all the kinds of sequence variants are the alleles. We may choose to limit our interest to alleles that cause phenotypic variations. Our preference doesn't limit what alleles can really be, though.

  3. "Many of you will be puzzled by such a definition but you have to realize that Lu and Bourrat see evolution as synonymous with natural selection."

    It seems to me that this particular misunderstanding of theirs isnt related to their push to define 'gene'. I think they'd say that the original notiion of 'gene' or 'gemmule' became narrowly defined as we learned about about transcription and translation. They're just saying our understanding of epi means we need to go back to the original more expansive definition.

  4. I thought this was going to be about Williams genes. A concept I'm very fond of, but which would have been better off not being called a gene. In the Williams sense a gene is anything that is heritable, where the rate of endogenous change is low enough (less than one expected change per generation), which is local in the sense that there is an upper bound to the number of copies per individual and which has an at most countable number of possible alternative versions. Alleles then are the currently existing alternative versions.
    That definition is far broader than the molecular gene, but it is eminently useful. As for coding regions, it is worth noting that in most cases the Williams definition tells us that a nucleotide sequence and the AA sequence it codes for a different genes, although these are in linkage disequilibrium.