Thursday, May 04, 2017

Debating philosophers: The molecular gene

This is my fifth post on the Lu and Bourrat paper [Debating philosophers: The Lu and Bourrat paper]. The authors are attempting to justify the inclusion of epigenetics into current evolutionary theory by re-defining the concept of "gene," specifically the evolutionary gene concept. So far, I've discussed their understanding of current evolutionary theory and why I think it is flawed [Debating philosophers: The Modern Synthesis]. I described their view of "genes" and pointed out the confusion between "genes" and "alleles" and why I think "alleles" is the better term [Debating philosophers: The difference between genes and alleles]. In my last post I discussed their definition of the evolutionary gene and why it is too adaptationist to serve a useful function [Debating philosophers: The evolutionary gene].

The molecular gene doesn't figure prominently in their argument but I'm going to discuss it anyway. Here's how Lu and Bourrat describe the molecular gene in Table 1 of their paper.
A stretch of DNA that contains an open reading frame with a promoter sequence, and functions in transcription and–or translation processes to create a genetic product. (Griffiths and Stotz [2013], p. 73) It is a stereotyped definition of the molecular gene. For more discussions, see Griffiths and Stotz ([2013]) and main text.
The text doesn't really add anything to that definition except to mention that noncoding regions of the gene and alternative splicing create problems with the definition. Nobody who has studied the problem claims to have a definition that covers all possibilities. There are exceptions to every definition of a gene that's ever been proposed.1 Griffiths and Stotz have done an excellent job of explaining some of the complications in their earlier papers. (I don't have the 2013 book referred to in Lu and Bourrat.)

My preferred definition is [What Is a Gene?]:
A gene is a DNA sequence that is transcribed to produce a functional product.
The functional product is RNA. It may be ribosomal RNA, snRNA, mRNA, tRNA, microRNA or a host of others. The point is that only one of those products is translated to make protein yet we still talk about tRNA genes, snRNA genes etc. The molecular gene is not restricted to just protein-coding genes. This is my main objection to the Giffiths and Stotz definition that Lu and Bourrat use in their paper. Genes do not have to have open reading frames.

There are many scientists who make this same mistake even though they should know better. However, if you are going to write a paper that relies on the correct definition of "gene" then surely you should think about it a little harder than the average scientist who might have been taught incorrectly as a undergraduate?

My preferred definition does not include the promoter and regulatory sequences unless they are transcribed. I prefer to think of regulatory sequences as sequences that control the expression of the gene without being part of the gene. Others may use a different definition of gene. For example, in his evolutionary biology textbook Futuyma (2nd ed. p. 188) describes a genes like this ...
The term gene usually refers to a sequence of DNA that is transcribed into RNA, together with untranscribed regions that play roles in regulating its transcription. The term locus technically refers to the chromosome site occupied by a particular gene, but it is often used to refer to the gene itself. Thousands of genes in the human genome encode ribosomal and transfer RNAs (rRNAs and tRNAs) that are not translated into proteins.
While it may be true that there is no universally correct definition of "gene," it's also true that some definitions are wrong. The Griffiths and Stotz definition is wrong because it excludes all DNA sequences that lack an open reading frame and this means that tRNA genes aren't genes by their definition.

1. Some genes are made of RNA, for example, and operons consist of several genes even though there's only one transcript.


  1. I guess the debate then moves to determining the meaning of a 'functional product'. The definition of function in biology has led to even more ink spillage than 'gene'. I don't have a horse in this race, but I thought I'd point this out.

    1. Oh and then this will merge with the debate about ENCODE and how many transcripts are under purifying selection.

    2. Function has to be in there somewhere because otherwise we can't distinguish genes from pseudogenes.

      We also can't distinguish between real genes and spurious transcripts. It may not be easy to figure out if some RNAs are functional but it has to be part of the definition of a gene.

      BTW, if you have a better definition then this is the place to tell us about it.

    3. I'm fine with your definition. Just wanted to point out that the question of 'function' and what is meant by it then becomes important to the definition of a gene. I have a lot of respect for philosophers of biology and I follow some of these debates. However, I think that we can still do a lot of biology without perfect definitions of words like gene. As long as everyone having the discussion agrees on the subject being discussed, then we can often sidestep the definitional issues.

    4. If you threw in a random stretch of DNA into a genome you could probably find an RNA transcript for that random stretch of DNA if you searched hard enough for it. This is why you need to distinguish between just being transcribed and being functional, something that the ENCODE people didn't do.

  2. Stotz and I were following Tom Fogle (a biologist) who argued that 'gene' is defined by ideas about the prototypical gene (our list of features was meant to be like a prototype) and something is a gene if it has 'enough' of the same features (some cases will be borderline). Psychologists regularly treat concepts this way - prototype and similarity. We don't think this is always how biologists think about genes either, sometimes they take a more top-down, functional approach (does the gene job) as Larry suggests.

    1. I don't understand your comment. You define a gene in terms of an open reading frame. Did you mean to exclude all "genes" for noncoding RNAs? I don't have your book. Did you create a new term for those sequences that specify ribosomal RNA, tRNA etc?

      Or did you just forget that there are genes that don't encode proteins? Your earlier papers, which I also critiqued, suggest that such genes weren't on your radar. For example, in your 2004 paper the only genes you and Karola discuss are protein-coding genes.

      But any biochemist or molecular biologist could have told you about tRNA genes—the first gene to be sequenced. I don't understand why you restricted your definition to protein-coding genes.

      This is what I mean about philosophers of biology who seem to be out of touch with the discipline they are studying.

    2. RE: Neo-darwinist philosophers vs empiricist (or physiologist) philosophers!?

      I certainly would concur with your acute observation of the fact that "This is what I mean about philosophers of biology who seem to be out of touch with the discipline they are studying." -- [or more appropriately who seem to be out of touch with the discipline they are reading and inferring!?]

      This is because most philosophers have had No “first-hand training” in the “scientific methods and materials” technigues, at all; but merely “reading and inferring” certain theories of their interests: their interests -- being deeply rooted in the prevailing “physicalist, reductionist, and evolutionist perspectives” -- which have had since 1930s-40s been influenced and hijacked by the then newly formularized, synthesized, and dogmatized Neo-Darwinism (please see analyses @ RE: Neo-Darwinism vs Darwinism: Is Modern Synthesis (MS) effectively dead!?).

      Best, Mong 5/5/17usct19:10; practical public science-philosophy critic (since 2006).

  3. For some reason "Bruces' Philosophers Song (Bruces' Song)" and the "Decomposing Composers" song released on Monty Python's Contractual Obligation Album come to mind.

  4. Generations of undergrads may thank you since they won't have to learn the cis-trans test, but I think that cis-acting elements should definitely be included in your (and Futuyma's) definition of a gene. Mutation in a cis-acting sequence does, after all, lead to a heritable phenotype, which is the most deeply rooted definition.