Coyne quickly corrected that omission with a post the next day: The Neutral Theory: a genuinely important supplement to evolution’s “Modern Synthesis”.
Comfort made many mistakes but one of them was praising Philip Ball's book How Life Works. This prompted Ball to respond to my blog post by giving us a link to his own review on the Scientific American website: The Selfish Gene, revisited.
I wasn't impressed by that review either and I pointed out that some of the problem is related to how you define the term "gene." Ball doesn't do a good job in any of his writings - for example, he misrepresents the history of the definition of "gene" by claiming that the original 1960s definition was restricted to protein-coding genes. He attributes that definition to Francis Crick. This allows him to advance his agenda of showing that the old fuddy-duddy scientists were wrong about genes and genomes and recent discoveries have overthrown the old paradigms. (He sees himself as part of the revolution.)
Ball defended his writings by claiming that it's very difficult to write for a general audience and I shouldn't be so picky about how to define a gene.
Well, as it turns out, I've also tried to write something for a general audience only I tried to be historically and scientifically accurate as well.1 Here's what I wrote in my book - please let me know if you think I did an adequate job on a difficult topic.
"Ball is one of the most meticulous, precise science writers out there. He is the antithesis of hypey, "dumb-it-down" reporting. He is MUCH more credible than you are, Laurence."
John Horgan July, 2024
I’m going to be talking a lot about genes so it’s a good idea to describe how genes work and how we define them. Here’s a good working definition of a gene, “A gene is a DNA sequence that’s transcribed to produce a functional product.” I’ll discuss some controversies about the definition in Chapter 6 but, for now, let’s just accept this definition. It’s what most biochemists and molecular biologists mean when they talk about genes. (Chapter 1: p. 21)
Defining a gene
In Chapter 1 I defined a gene as a DNA sequence that is transcribed to produce a functional product and this is the way genes have been described since the 1960s by many scientists who have thought seriously about the subject. There are lots of textbooks that use a similar definition. Kostas Kampourakis in Making Sense of Genes concludes that the most acceptable definition of a gene is the one that includes production of both proteins and noncoding RNAs (see the quotation at the beginning of this chapter).
Therefore, in this book I will consider genes as DNA sequences encoding information for functional products, be it proteins or RNA molecules. With “encoding information,” I mean that the DNA sequence is used as a template for the production of an RNA molecule or a protein that performs some function Kostas Kampourakis (2017)
The idea here is that the primary gene product is RNA and any segment of DNA that’s copied into RNA (transcription) qualifies as a gene as long as the RNA has a biological function. That last bit is important because there are sequences of DNA that are transcribed but the RNA doesn’t do anything. Transcribed pseudogenes are an example but so are many spurious or accidental transcripts that happen quite frequently when the transcription machinery makes a mistake. Those bits of DNA are not genes even though they are transcribed.
There are very few definitions in biology that are absolute. In fact, one of the main reasons why biology is so difficult is because it is full of exceptions to the rule. In the case of genes, the exceptions are well-known; they include the fact that some genes are made of RNA not DNA—the genes of coronavirus are an example. There are also more complicated exceptions such as cases where several different protein-coding regions are transcribed together to produce a long RNA where several different regions can be translated simultaneously. These complexes are called operons and they are common in bacteria. The problem here is that we will often want to refer to each protein-coding region as a separate gene. There are also rare examples of two separate DNA sequences that are transcribed to produce RNAs that are then joined together to make a functional product; strictly speaking, neither of the separated DNA sequences qualifies as a gene.
What this means is that the best definition of a gene is only a “working definition” because there is no definition that covers all cases. This drives philosophers crazy but biologists are used to such ambiguity.
The molecular gene and the Mendelian gene
The gene I’m talking about is the “molecular gene” (a gene that corresponds to a particular DNA sequence in the genome) but geneticists often talk about another kind of gene called the “Mendelian gene.” Both definitions are used in the scientific literature and they do not overlap. I will use the molecular definition in this book but keep in mind that there are lots of functional DNA sequences that are not genes by this definition.
The Mendelian gene concept is frequently used by geneticists and evolutionary biologists. It’s a nebulous ill-defined unit as seen in the definition used by Richard Dawkins in his book The Selfish Gene. To him, a selfish gene is any bit of chromosome that’s subject to natural selection. This would include centromeres, origins of replication, and regulatory sequences that are not genes by the molecular definition. When Dawkins talks about the “selfish gene,” he’s talking about a unit of selection and not the molecular gene. The “selfish gene” is just a convenient metaphor as he makes abundantly clear in his book.
To be strict, this book should be called not “The Selfish Cistron” nor “The Selfish Chromosome”, but “The slightly selfish big bits of chromosome and the even more selfish little bit of chromosome.” To say the least this is not a catchy title so, defining a gene as a little bit of chromosome which potentially lasts for many generation, I call this book “The Selfish Gene.” Richard Dawkins (1976)
Dawkin’s book, The Selfish Gene, has been voted the most inspiring science book of all time in a 2017 poll by the Royal Society of London. This is a problem because the book only deals with the Mendelian gene so the general public (and many scientists) are not familiar with the molecular definition and the distinction is crucial. It’s also a problem because the concept of the selfish gene is highly adaptationist—it’s all about natural selection—and you will never understand genomes if that’s all you know about evolution.
In the debate over the definition of a molecular gene, I favor a definition that focuses on RNA as the functional product because I think this definition captures the essence of what most scientists believe when they talk about genes as functional entities. This includes scientists who may slip up from time to time by restricting their gene talk to protein-coding genes. When you ask them in private, it turns out they are well aware of ribosomal RNA genes and a host of other nocoding genes. Sometimes scientists just forget to be precise when talking about genes and they give the impression that there’s only one kind of gene; namely, the ones that encode proteins. (Chapter 6: pp. 137-138)
1. The top three requirements for good science writing are: accuracy, accuracy, and accuracy.
10 comments :
That's a very nice summary!
Thanks Professor Moran. Fascinating. I hadn't twigged that not everything that is under selection is a gene in the molecular definition.
So what are centromeres etc defined as?
The genome contains functional elements, all of which are under selection to be conserved, though some may be de novo and so not visible in ancestral genomes. Some of these are transcribed - genes- others aren't centromeres etc.
I've also found it confusing that parts of a gene are junk - introns. That confuses me as how can it be a part of a gene (which is defined via function), when it doesn't have a function and is snipped away.
Thanks for educating this confused layman I find it fascinating.
For someone who places such emphasis on accuracy, Larry seems oddly unconcerned about it whenever he talks about my own writings. Essentially nothing he says here (or in the past) about that is accurate. I know he doesn't really care about this, however - he's decided on his story, and pushes it relentlessly. For anyone who doubts whether Crick considered genes to be essentially about encoding proteins, take a look at his 1958 paper "On Protein Synthesis". There Crick says he considers Beadle's "one-gene one-enzyme" hypothesis to be essentially correct, and the only evident exception Crick mentions is the existence of what he calls soluble RNA (which was essentially tRNA), the origin of which he is however uncertain about. In this, Crick was fully justified given what was known at the time.
But I wasn't, in my previous response to Larry, really arguing that one can excuse inaccuracies in writing for a general audience. Nor did I say he shouldn't be picky about defining a gene. (I do think one should be honest about the fact that there is no consensus definition, and indeed Larry mentions that the Mendelian gene is something different from the molecular gene, as I mention too in my book.) What I really meant there was that one must to some extent rely on the assumption of reading skills in one's audience. So when I say, for example, that "The evolutionary biologist doesn’t much care what a gene does, so long as it somehow affects the traits and thus the fitness of the organism", I am not saying "Each and every evolutionary biologist will attest that they care about nothing but the phenotypic and thus fitness implications of a gene". That is simply not what the sentence means. It means "The emphasis regarding genes within evolutionary biology tends to be on their effects on fitness." To suppose otherwise either implies poor comprehension skills or deliberate misreading in order to be able to say "Look how ignorant/inaccurate he is!" Neither interpretation reflects well on Larry.
For those interested in a proper historically accurate view of thinking about genes in the early days of molecular biology in the 1950s and 60s, I recommend Lily Kay's book Who Wrote The Book of Life?, which does not pretend that biologists have known perfectly well what genes (or indeed RNA) are since that time.
@Philip Ball: You could have said that back in the 1950s, at the beginning of molecular biology, scientists thought that protein-coding genes were the only kind of genes, but by the time Watson published his textbook "The Molecular Biology of the Gene" in 1965, it was clear that there were two kinds of genes; protein-coding genes and those that specified functional RNAs such as ribosomal RNAs and tRNAs.
Ever since that time, 61 years ago, all knowledgeable scientists knew about both kinds of genes. You could have said that and been accurate.
Instead, you make a big point about the central dogma and the idea that Crick defined a gene only in terms of making protein. You imply that this definition was the prevailing view in the 1980s and 90s (p. 75) and on page 82 you say, "Crick's definition of a gene as part of DNA that encodes a protein is no longer adequate, for reason that will become more clear in the next two chapters."
Everybody reading that will assume that Crick only knew about protein-coding genes until recently (he died in 2004). Now you are pointing out that it "is no longer adequate," which definitely implies that things were only about to change in 2023 when your book was published. I'm pretty sure that "no longer adequate" doesn't mean 61 years ago.
Then you say on page 83, "There is no consensus about whether or when such noncoding DNA sections warrant being called (noncoding) genes - simply because there is no longer any agreement about what, at the molecular level, 'gene' should designate."
Do you actually believe that? Textbooks and Wikipedia define the molecular gene as a DNA sequence that's transcribed to produce a functional product. A non-coding gene is a gene if it produces a functional product. That's the consensus view even if there are some unknowledgeable people who don't understand it.
Here's a quote from the Alberts et al. textbook "Molecular Biology of the Cell" (1994 edition, 32 years ago). It's on the Wikipedia site where you could have easily found it.
"The primary function of the genome is to produce RNA molecules. Selected portions of the DNA nucleotide sequence are copied into a corresponding RNA nucleotide sequence, which either encodes a protein (if it is an mRNA) or forms a 'structural' RNA, such as a transfer RNA (tRNA) or ribosomal RNA (rRNA) molecule. Each region of the DNA helix that produces a functional RNA molecule constitutes a gene.["
There are other quotations there that conflict with the view you express in your book. Why didn't you check the textbooks before publishing your book?
At this point I’m responding just in case there are others still reading the exchange (unlikely I guess), so that they are not misled.
It’s clear Larry is imagining that no one is going to go to my book and check the things he says about it. For example, when I talk on p75 about “the prevailing view in the 1980s and 90s”, it is not with reference to the idea that genes only encode proteins.
It is obvious to anyone reading my book by that point that not every single functional sequence in DNA codes for proteins, because there is also rRNA and tRNA involved in protein synthesis. Larry’s own wording about this is constantly misleading, as he talks about “functional RNAs such as rRNAs and tRNAs”. Sure, we’ve established that. And what else? Well, Alberts et al. are pretty clear about that in that edition of their textbook: they point out that there are also catalytic RNAs involved in ribosome function, discovered in the 1980s (as mentioned on p109 of my book). And that’s it. Indeed, Alberts et al. say that “The expression of the genetic information stored in DNA involves the translation of a linear sequence of nucleotides into a co-linear sequence of amino acids in proteins”, they too recognizing that this is an acceptable shorthand given that elsewhere they’ve also mentioned those (presumed rare) exceptions regarding tRNA, rRNA and ribozymes.
So why is Larry still unhappy? I think it’s fairly clear: he seems to be dedicated to obfuscating the fact that it was only since the late 1980s and especially the 1990s that we began to recognize the importance of other noncoding RNAs, which is the story I tell in that part of my book. For anyone who doesn’t appreciate what a significant shift that was, I suggest looking up the citations for the Nobel prizes awarded in 2006 to Andrew Fire and Craig Mello for their work on siRNAs and especially to Victor Ambros and Gary Ruvkun in 2024 for microRNAs. (e.g. https://www.nobelprize.org/prizes/medicine/2024/press-release/). The importance of functional long ncRNAs also became apparent in the 1990s. Tom Cech, who won the Nobel for the discovery of ribozymes, has called this discovery of the diversity of roles of ncRNA since the 1990s a revolutionary change in understanding. The distinguished geneticist Adrian Bird, who has been rightly cautious about assuming widespread functionality in all transcribed ncRNA, nevertheless refers to this change as a paradigm shift. In all this, Larry is now an outlier. I'm not sure why he takes this position: this new knowledge is (as long as it's not overhyped) very exciting. Isn't new knowledge what science seeks?
There are all kinds of reasons why there is still plenty of discussion about what is and isn’t a molecular gene (and which bits of the sequence are in fact “the gene”), and one reason for this, as I explain in my book and elsewhere (and above), is that there is some ambiguity about what counts as functional. Larry did not engage with this. Nor, each time I point out another misrepresentation of what I have written, does he seek to defend it. He just moves on to something else. This is the well known tactic of what historian Naomi Oreskes calls “merchants of doubt”.
Here's my version.
There are two kinds of genes; those that encode proteins and those that specify functional RNA (non-coding genes). Scientists have known about non-coding genes since the 1960s but back then the only ones known were the different genes for ribosomal RNAs and the 20 or more genes for transfer RNAs.
Genes for regulatory RNAs were discovered in the 1970s and more non-coding genes were identified in the 1980s (snRNAs, snoRNAs, siRNAs, and piRNAs). There are also some unique examples of functional RNAs such as 7SL RNA, XIST, and RNase P.
Later on, classes such as miRNAs and lncRNAs were added to the list.
All eukaryotes have roughly the same number of catalytic RNAs Complex multicellular eukaryotes, such as humans, contain variable numbers of the regulatory RNAs but the exact number is not known. The current estimates from some knowledgeable scientists in the field are that there are at most a few thousand non-coding genes in total and possibly fewer than that.
This number is disputed by other scientists who argue that there may be many more non-coding genes than protein-coding genes in the human genome. Some of these estimates go as high as several hundred thousand.
Nobody questions the fact that there are thousands of low level transcripts in human cells. The debate is about whether most of these transcripts have a biological function since the definition of a gene specifies that the product must have a function. There is good evidence that most of these transcripts are spurious RNAs produced by errors in transcription.
There are some scientists who skip the 'function' part of the gene definition and insist that the mere presence of a transcript means that it comes from a 'gene' and that's why they claim that there are hundreds of thousands of non-coding genes. Some even go so far as to say that this represents a 'paradigm shift.'
The so-called 'paradigm shift' is not about the existence of non-coding genes since those have been known for more than 60 years. It's also not about the expansion of the types of non-coding genes that occurred in the last half of the 20th century since none of those discoveries shifted any paradigms.
The so-called 'paradigm shift' that's being advertised today is a whole order of magnitude expansion of the number of non-coding genes in some complex multicellular eukaryotes with genomes full of junk DNA. The onus is on those claiming a paradigm shift to come up with solid evidence that this number of functional non-coding genes actually exist.
So far, after two decades of intense investigation, that evidence is lacking but that hasn't stopped scientists and science writers from claiming that a paradigm shift has already occurred.
I'm pretty sure that readers aren't going to rush out and buy Phil Ball's book so here's what he says on page 75.
"The Problem(s) with Genes
Despite all these caveats, complications and conditions, Crick's point remained: a gene was a means of getting a protein. But what then? The prevailing view up to the 1980s and 90s was that somehow the activation - the technical term is expression - of a gene to make a protein leads to some specific physiological or developmental outcome ..."
Ball then goes on to discuss some complicated point about the function of developmental genes.
The point I was making was that Ball established a solid link between the idea that a gene only encoded proteins (attributed to Crick) with the prevailing view in the late 1980s and 90s.
Philip Ball says, "There are all kinds of reasons why there is still plenty of discussion about what is and isn’t a molecular gene (and which bits of the sequence are in fact “the gene”), and one reason for this, as I explain in my book and elsewhere (and above), is that there is some ambiguity about what counts as functional. Larry did not engage with this."
The debate over how to identify and define function (Function Wars) has been covered in numerous publications over the past 20 years or so. I devote several pages to the issue in my book (pp. 302-307) followed by a critical discussion of "Scientific revolutions."
I've also posted lots of stuff on my blog about the function wars, including detailed analyses of the best ways of identifying function. Almost all of that information was available to Philip Ball before he published his book.
The bottom line is that there is no logical defense of the claim that there are hundreds of thousands of non-coding genes. The best you can say is that such a claim is extremely unlikely to be true given all that we have learned over the past 60 years but it can't be definitively disproven.
In spite of all that, here's what Philip Ball says on page 125 of his book.
"ENCODE initially identified around 16,000 long ncRNA genes (yes, we can justifiably cell them genes; the redefinition I presented on p. 83 came out of ENCODE's revelation, explicitly to remove the emphasis on proteins). By 2020 the project had proposed that there are almost 37,600 such genes, which is nearly twice as many as those that encode proteins."
The function wars are over
Ah, gradually we're getting there. It's now clear that the debate is in the weeds, and it comes down to the fact that Larry disagrees with what some other (often heavy-hitting) scientists say. I'm fine with that. That's what science is about. But it's evidently not at all about "It's in all the textbooks and everyone knows that and you should have just looked it up." (And by the way, this applies too to definitions of the gene. For example, according to the Human Genome Nomenclature Organization in 2002, a gene is “a DNA segment that contributes to phenotype/function. In the absence of demonstrated function a gene may be characterized by sequence, transcription or homology." Hmm, sounds a little more complicated than what Larry tells us about the state of play in 1994, right?)
The quote from my book speaks for itself, so I don't need to defend that any more.
On past experience it'll be back to square one the next time my name crops up, but such is life.
Post a Comment