Does the genetic load argument really support junk DNA?

Recently I was reading about junk DNA and I was convinced that it is not based on ignorance, but I have a question about the genetic load argument that Lawrence Moran uses, which is that he assumes that all mutations that will occur in any functional DNA will be affected in the same way as the coding regions. Now my question is, is this assumption justified, or can functional DNA be uneven in the effect of mutations on it?

Did you mean coding regions or non-coding regions? Coding regions should be functional in my understanding, so your question makes more sense if you intended to write “non-coding regions”.

1 Like

It would be easy enough to ask Larry. Have you tried posting this to Sandwalk? Is this a question about his book? If so, what page are you referring to? It would be easier to respond if I knew exactly what assumption you’re talking about.

I’m guessing you’re referencing the discussion on page 90-91, in which the assumption is that functional non-coding regions have the same rate of deleterious mutations as is estimated for coding exons, around 10% of mutations. Is that right?


Yes, it’s right, he says " The key parameter is the fraction of all mutations that are harmful since this is what contributes to the “load” that the population must tolerate. It’s actually very difficult to obtain accurate values for this fraction, but most population geneticists think that a value of one-tenth (10 percent) is likely to be a good estimate. This value is based largely on the number of deleterious mutations in protein-coding regions, but it’s reasonable to assume that it applies to all functional regions of the genome, including regulatory sequences, centromeres, and the like.
This number means that only 10 percent of all mutations in functional DNA regions (e.g., genes) are deleterious and that the other 90 percent are effectively neutral, although there are a small number of beneficial mutations occurring from time to time. If the entire genome is functional and there are 100 mutations in every newborn baby, then 10 of these mutations will be harmful, and nobody thinks that our species could tolerate such a high genetic load."

Your question appears to be based on not understanding that “functional” is not synonymous with “coding.” This misunderstanding is actively promoted by IDcreationist scammers.


I have no reason to question that estimate, though I have to say I’m surprised that the proportion of deleterious mutations is so low. It seems that exons should be, on average, be evolving at close to the neutral rate (as in introns, for example) and that isn’t my experience. My intuitive estimate would be that over half of exon mutations are deleterious, including most non-silent mutations. Of course that would decrease the fraction of the genome that could be under selection, given the genetic load aargument.

Thanks for posting the text you are discussing. I think Larry has adequately explained that there is significant uncertainty in that assumption (which he accurately calls an estimate), and I think his overall point (that assigning “function” to the entire genome implies a staggering genetic load) is valid. But:

Yes, I think that “functional DNA” can vary a lot in its tolerance to mutation re fitness effects. This will depend on various things but most notably IMO on the extent to which the sequence in question exerts its “function” by virtue of its specific sequence. That might sound illogical but it’s not: consider a DNA sequence whose function is to be a part of a loop, in which the loop’s “function” is physical but independent of the nucleotide sequence. That sequence could happily absorb most SNPs and even small indels without any consequence, and thus maybe we would assign “function” to the sequence even though deleterious mutations would be very rare (way less than 10%).

Does that mean that the 10% estimate cited by Larry is a lot lower on average for non-coding DNA? I don’t know and I doubt it. But you are right that mutation can have widely varying effects on “functional” DNA sequences.

That’s what I was thinking, the genetic load argument would be a wonderful and strong argument for junk DNA if all functional DNA is affected by mutations no matter what the effect is, but the whole argument falls apart if most functional DNA can absorb most SNPs and indels, but my intuitive is that functional DNA should be affected by mutations

But at the same time, there is no reason to accept this estimate.
The important point I want to discuss is, can most of the functional DNA be unaffected by mutations?

No. Almost by definition.

If the ‘function’ of a region of DNA is unaffected by mutations, then that function logically can’t be sequence dependent. That doesn’t leave you with much.


There’s other factors to consider, which is the functional fraction of the genome, the mutation rate, and genome size.

The non-functional part of the genome can tolerate many more mutations than the functional part. The functional part can still tolerate lots of mutations, but in comparison to the nonfunctional part it is still much more sensitive. The genetic load argument considers all these factors together.

Imagine you hold the mutation rate and genome size constant, but change only the fraction of the genome that is functional. Like the three bars here, meant to represent genomes with the functional fraction marked in green. Say the top one is 50:50 functional/nonfunctional, the middle is 85:15 functional/nonfunctional, and the bottom is 10:90.

Suppose the functional part can tolerate 50% of mutations, but the nonfunctional part can tolerate 99.9% of mutations. Now suppose all three of these organisms suffer 100 mutations every generation.

Clearly the top organism with can tolerate more mutations than the middle one one. Roughly half of those 100 mutations can be tolerated, as approximately 50% will land in the nonfunctional part with can tolerate 99.9% of mutations(meaning 1 in 20 individuals will have a deleterious mutation here). Then 50% of the 50% that land in the functional part can be tolerated too, leaving 25 mutations being tolerated and 25 being deleterious here.
So all in all there will be 25.05 deleterious mutations on average to individuals with a genome like the top organism.

In the middle organism there’s 85 mutations landing in the functional part and 15 in the nonfunctional part on average. 50% of of those that occur in the functional part (85 mutations) are deleterious so 42.5 mutations are deleterious in the functional portion. Then another 15 in the nonfunctional part, 0.1% of which are deleterious. So that’s 42.515 deleterious mutations in total.

42.515 > 25.05

With these numbers, if the functional fraction of the genome is 10% (and holding the other numbers constant), then roughly 5% deleterious mutations will happen every generation. The remaining 90% nonfunctional part will suffer 90 mutations on average, 99.9% of which have no effect (so 0.09 mutations will be deleterious, on average). Which means it suffers 1 deleterious mutations every ~11,1 individuals on average.

So for 10% functional it’s 5.09 deleterious mutations.

42.515 > 25.05 > 5.09

So clearly the functional fraction makes a large difference.

Your intuition is right. Most functional DNA is affected by a considerable fraction of mutations. But the difference in the rates of deleterious to neutral/beneficial mutations in functional vs nonfunctional DNA is very large.

That’s why, all else held equal, the functional fraction of the genome makes a large difference for genetic load.

1 Like

Not sure why that’s an important point. If the argument underestimates the number of deleterious mutations, then upping the estimate makes the argument stronger.

1 Like

Important to note that a deleterious mutation imposes the same load, whether it is severely deleterious or moderately deleterious. That is the essential result found by Haldane (1939) and Muller (1950). Feels counterintuitive, but it is true.


But in the end it remains intuitive, for we are faced with two options: either most of the genome is functional or most of it is junk. If most of it is functional, then this implies that most of the functional DNA is not affected by mutations, and mutations in it are more like silent mutations (i.e. do not affect regulation, transcription, etc.).

Otherwise the mutational load would be too high, yes.

But then that implies most functions would be sequence-independent, as if most DNA is just spacer or filler-DNA (maybe there needs to be at least a certain distance between some elements, and the filler can have basically any sequence).

The mutational load argument can’t disprove a spacer-DNA hypothesis. That is correct.

For that there are other lines of evidence, which are the large variations in genome size between even closely related species, the actual constituents of junk DNA (mostly dead transposons, viruses, and pseudogenes etc.)


And most of it is junk.


Yes, you are absolutely right, and in addition to what you said, in light of this hypothesis, if we assume that there is a sequence that regulates the X gene, then according to this hypothesis, most of the possible sequences can perform this function!! If we assume that the length of the sequence that regulates the X gene consists of 10 nucleotides, this means that the same function can be performed by about 40 different sequences!! This is clearly counterintuitive, in addition to that the different functions could be performed by the same sequences, I don’t think anyone would say this, and therefore according to Occam’s code the DNA must be mostly junk, otherwise the genetic load will be high.

You’ve got the math wrong there. It’s much, much worse (many more than 40). I assume you just multiplied 10 by the number of different DNA bases (4).

The correct calculation is 410 for DNA because there’s 4 possible bases at each position in the length of sequence and you’ve defined the length to be 10, and 20length of sequence for proteins because there’s 20 possible amino acids at each position.

So 410 = 1 048 567

That said, I get your overall point and I agree.


The Twitter thread Paul Nelson alluded to in his piece below may have some relevance to this discussion

I had a hard time finding that discussion from @Paul_Nelson1’s link, but I was able to track it down from the author.

However, I don’t see the relevance. Mutations impose a genetic load whether or not they are random. A discussion of randomness is a different topic, and one we have covered previously.

1 Like