I get the skepticism. There have been a lot of surprising revelations in biology and I don't think anyone would argue we have every angle nailed down. However, the idea that some DNA is genuinely “junk” is based on more than a hunch. It’s from looking at patterns across species. If a sequence really mattered, then changing it should cause a problem. That would put pressure on the sequence to stay the same, generation after generation. Yet we see big stretches of DNA mutating freely, at rates that exactly match what would be expected from accumulation of random copying errors. That suggests these sequences aren’t under selection for any important function.

This isn’t just “we don’t know what it does, so it must be junk.” It’s more like, “We can’t find any sign that it matters, and everything we know about evolution says if it mattered, we’d see fewer random changes there.” Down the road we might uncover small roles for some of these regions, but at this point, calling them junk is just an honest read of the evidence we have.

But matters in what time horizon. Because it doesnt matter today doesn't mean it doesn't give survial advantage in the future.

What about 1 million years?

That's just not how evolution operates. It can't "look forward" 1 million years and keep something because it will be advantageous in the future.

I'm not suggesting that these sequences can "look forward" in time. However, consider that mutations are constantly occurring. These mutations shouldn't be dismissed as "junk" simply because they seem unnecessary now. In the future, they could become essential.

Over long evolutionary or environmental timeframes, these sequences may take on important functions, potentially becoming critical under conditions we can't currently foresee.

If such a mutation occurs, that sequence would no longer be junk. Until and unless it does happen, it's still junk. But it's silly to get hung up on the sequence, or on the word "junk", based on such a slim chance. What are you trying to prove here?

The weird thing is that some of these lncRNA don’t seem to be under super strong selection pressure, at least at the level of individual nucleotides. Their promoter regions are conserved, which indicates that the cell really does need to produce them, but it doesn’t seem to care much the actual sequence. Very strange.

Anyways there definitely are non-coding regions that just don’t do much and evolve neutrally. I’m hesitant to call them junk but only because that designation has burned biologists so many times.

I think we’re basically on the same page. As you note, a conserved promoter without strong sequence conservation elsewhere suggests functions that might be more structural or regulatory. Still, it’s also true that some (actually many) non-coding regions show no evidence of selection and appear to evolve neutrally.

To borrow an example: an onion likely doesn’t need 5x more DNA than a human, and a lungfish probably doesn’t need 30 times more than we do (and 350x more than a pufferfish). And yet, these enormous genomes exist. It’s very likely that portions of these sequences are what we’d call “junk,” i.e., DNA that doesn’t confer a meaningful functional advantage and can accumulate due to the relatively low cost of carrying it along.

If we want to avoid the term “junk,” we could say something like “areas of the genome for which we assign a very low prior probability of functional importance.” But “junk” is a concise shorthand to acknowledge that, while some non-coding sequences matter, there are also huge swaths of DNA in many eukaryotes that show no signs of being anything other than evolutionary baggage.

Great overview. Worth adding some population genetics: Multicellular organisms typically have small effective population sizes and reproduce slowly in comparison to bacteria. Selection has a hard time “getting a grip” on variants with very weak effects on fitness. Drift becomes much more important.

Bacteria have high population sizes. Selection can be quick and brutal. Low levels of “code of unknown function” in bacteria is perhaps related to replicative efficiency. Fast DNA replication is highly advantageous in nutrient-rich environments. No space (or time) for junk DNA.

Sometimes with lncrnas the structure is what is more important than sequence. You can have two lncrna with different sequence but the same kmer structure. This makes logical sense as while proteins often bind to specific sequence the reasons for that are merely structural. In protein you can also have conservative missense mutations that are tolerated as binding affinities may not have changed swapping out an amino acid residue for another with the same charge or polar properties.

https://pmc.ncbi.nlm.nih.gov/articles/PMC6262761/

> Yet we see big stretches of DNA mutating freely, at rates that exactly match what would be expected from accumulation of random copying errors.

But the rate of mutation is itself subject to selection. There isn't a base rate, just a setting that's different for different parts of the genome. Some parts have more copy errors than other parts. Some are hung out in the sun more often.

So you can conclude from the mutation rate of a particular stretch that it would probably be bad if it started mutating more, and that it would probably be bad if it started mutating less, but not that nothing's influencing the mutation rate.

It's based on certain presuppositions but 1 thing is for sure - we'll know more in the future as more research is done.

Most of the genome is non protein coding. Some is functional, most simply is not functional in any way. It is just empty space.

Rates of mutation in these regions, and lack of conservation are hallmark clues which show that there isn't function in these regions. That doesn't mean totally useless, these non-functional regions provide the raw material for the creation of genes and functional elements. Its just that, right now, those regions aren't doing anything.

No biologist calls it "junk DNA". That is just a simplified layman's term for media press releases.

I would say that, on balance, one must prove DNA to be not-junk.

The idea that you can have DNA of some critter and there aren't some errors, unused bits, and so on, after what must be trillions of copies, well, I would find it statistically unlikely. Like saying you have a program with millions of lines of code and it is completely error-free.

As said below, the sequence of the junk DNA does seem to mutate a lot faster than parts of the DNA which is known not to be junk, heavily suggesting that at least the exact sequence is not so important.

Furthermore, in rats and mice, large swaths of junk DNA have been experimentally removed, without any detectable effect on the phenotype.

If the junk DNA has any positive effect, it may be to protect against viruses or transposons inserting themselves into random areas of the genome. Keeping the majority of DNA "useless" may decrease the risk that these insert themselves into vital parts of the DNA.

Basically ASLR for DNA?

I read once (maybe in a Dawkins book) that large gaps also help preserve genes during sexual reproduction when chromosome crossover happens. Basically if the crossover point happens in the middle of a gene, that gene doesn’t survive the meiosis… having large gaps increases the odds that crossover happens in junk DNA. I’m not sure how true/oversimplified this is though.

Lots of recombinations within “genes”—-however you define a gene.

What I read was from “The Selfish Gene”, I went and dug up the section:

> A gene is defined as any portion of chromosomal material that potentially lasts for enough generations to serve as a unit of natural selection. In the words of the previous chapter, a gene is a replicator with high copying-fidelity. Copying-fidelity is another way of saying longevity-in-the-form-of-copies and I shall abbreviate this simply to longevity. The definition will take some justifying.

> On any definition, a gene has to be a portion of a chromosome. The question is, how big a portion—how much of the ticker tape? Imagine any sequence of adjacent code-letters on the tape. Call the sequence a genetic unit. It might be a sequence of only ten letters within one cistron; it might be a sequence of eight cistrons; it might start and end in mid-cistron. It will overlap with other genetic units. It will include smaller units, and it will form part of larger units. No matter how long or short it is, for the purposes of the present argument, this is what we are calling a genetic unit. It is just a length of chromosome, not physically differentiated from the rest of the chromosome in any way.

> Now comes the important point. The shorter a genetic unit is, the longer—in generations—it is likely to live. In particular, the less likely it is to be split by any one crossing-over. Suppose a whole chromosome is, on average, likely to undergo one cross-over every time a sperm or egg is made by meiotic division, and this cross-over can happen anywhere along its length. If we consider a very large genetic unit, say half the length of the chromosome, there is a 50 per cent chance that the unit will be split at each meiosis. If the genetic unit we are considering is only 1 per cent of the length of the chromosome, we can assume that it has only a 1 per cent chance of being split in any one meiotic division. This means that the unit can expect to survive for a large number of generations in the individual’s descendants. A single cistron is likely to be much less than 1 per cent of the length of a chromosome. Even a group of several neighbouring cistrons can expect to live many generations before being broken up by crossing over.

[end quote]

I think a reasonable extrapolation from this is that “genes” (things that are subject to natural selection in our genome) can survive more generations if they are surrounded by genetic material that are not “genes” (ie. Their copying fidelity is not subject to any selection pressure.)

If a gene in this sense is recombined in a way that makes it no longer the same gene, it most likely isn’t going to be beneficial to the organism (and thus the gene’s longevity.) Most genetic mutations aren’t.

Funny, I read that same section two weeks ago. Dawkins’ definition is not what geneticists typical consider a gene. His “gene” is more like what I would call a haplotype in that he divorces “gene” from “protein-coding” region. But he has a good point and I like this section. But he wrote it in the genetic dark ages and we now know quite precisely where and how often recombinations occurs, and for how long haplotypes are preserved as a function of generation numbers.

Highly recommend David Reich’s book for are good overview of the math of recombination in humans.

https://www.amazon.com/Who-Are-How-Got-Here/dp/1101873469

Thanks for the recommendation! I’ll have to check it out.

I got into Dawkins’ books initially from The God Delusion (as I suspect many laypeople), and heard about The Selfish Gene from there, so evolutionary biology is not my area.

It makes sense that TSG is considered the dark ages as it’s such an old book. I was always curious to read more about the topic from other — and hopefully more recent — biologists, since Dawkins sometimes feels like he’s more of a communicator than a practicing biologist (and one with a particularly anti-religious chip on his shoulder, not that he’s wrong.)

Teleost fish that you and I would have a hard time telling apart, can have a genome sizes that range from 0.5 billion basepairs to 50 billion base pairs. It would be difficult to explain this huge range as due to selection acting at a fine grain on 49.5 billion nucleotides.