Drs. Sanford and Carter respond to PS Scientists

And since mutations are approximately random, we should expect variance. If Sanford is proposing variance so low it is unselectable, that’s something he needs to show support for empirically, and cannot just assert.

It’s inherent to his argument. Mutations accumulate without being selected against, and then at some point the population goes extinct. That means at some point you have to go from “no fitness cost” to “high fitness cost”. There’s an inflection point there, a fitness cliff.

2 Likes

7 posts were split to a new topic: Comments on Sanford and Carter respond to PS participants

You should read the book.

Worse, not all mutations have the same probability. If a C→T mutation (one of the most common mutations we see) happens, the reverse [T→C] is less likely to occur. Thus, while we are waiting for our rare T→C back mutation to happen, other C→T mutations should be occurring in the near vicinity. The same is also true of A→G vs G→A, where the former is much more common than the latter. This is all based on simple chemistry. But these are all examples of transition mutations, where a swap has been made between a purine and a purine (A and G) or a pyrimidine and a pyrimidine (C and T). Transversion mutations (swapping a purine for a pyrimidine and vice versa) are even more rare, so while the organism is waiting for a transversion to reverse itself, transition mutations (mostly C→T and A→G) will be happening all around it.

Thanks @PDPrice this is interesting to know.

This below seems key. Does anyone here actually disagree with any of it?

The selection threshold, that is, the strength a mutation must have before it can be ‘seen’ by natural selection, has been quantified, for the first time, using Mendel’s Accountant .17 Any claim that new mutations are equally as likely to be beneficial as they are deleterious flies in the face of both expectation and measurement. The human genome demonstrates a huge load of rare, recessive, and deleterious mutations. These have been preserved by our rapid population growth. In rapidly expanding populations, new mutations are less likely to be lost, so they show us something more closely approximating the real mutation spectrum. We have also been able to see the effects of mutation accumulation on the human H1N1 influenza virus. After nearly 100 years in circulation, it became less deadly and less infectious as the number of spontaneous, random mutations climbed inexorably. At the same time, codon usage randomized, and nucleotide composition began to approach thermodynamic equilibrium. In fact, long-term mutation accumulation patterns followed a predictable pattern that closely matched the chemical probabilities of the various mutations. That shows us selection was not a significant factor in affecting which mutations accumulated.

How can they say we don’t know the distribution of fitness effects when we can certainly know what mutations are accumulating in genomes over time? Evolution cannot handle this. By definition, natural selection must be able to overcome the effects of thermodynamic pressures in chemistry. If not, all life would have already devolved into simple chemicals. Sanford and colleagues also studied the effects of strongly beneficial mutations. The models indicated that selection for a mutation also carries along the nearby deleterious alleles. Thus, ‘linkage’ is a major problem. In order for their idea to work, beneficial mutations must vastly outnumber deleterious ones.

1 Like

Not, I think, for nearly neutral mutations.

Just not true. It’s a completely gradual accumulation of reduced absolute fitness, and again there’s a cost to absolute fitness but no significant differential in relative fitness. I don’t think you understand Sanford’s model.

3 Likes

Let’s take a look…

Um, no. Running a simulation tells you about the selection threshold in your model – it doesn’t reveal something new about the real world. And this is far from the first time that anyone has quantified the selection threshold in a model. I’ll label this one ‘grandiloquently false’.

In the context of mutations of very slight effect, this one is false regarding both expectation and measurement. ‘Doubly false’, then.

All true. And also all irrelevant, since no human mutations have been shown to have the very slightly deleterious effects needed for GE. Since this is presented as somehow supporting GE, I’d have to give this bit a ‘misleading’ label.

I’ll leave it to others to deal with the specifics of H1N1 flu, since I haven’t paid much attention to the subject. The last sentence, though, is completely wrong – the mutations in any virus are highly constrained by natural selection. A charitable designation would be ‘deeply confused’, but ‘breathtakingly false’ is more accurate.

We can say that because knowing what mutations are accumulating tells us nothing about the fitness effects of those mutations. I would call this ‘not even wrong’.

I’m having trouble making sense of this. Combined with the statement about chemical properties above, it appears to be saying once again that selection hasn’t been acting on H1N1 at all, so I guess this should be lumped together with the previous breathtakingly false statement.

This one is true. Of course, it’s also a phenomenon that’s well known by everyone in the field, known as ‘hitchhiking’. The fact that it’s introduced here as if it were a discovery suggests either a deep ignorance of the field or an attempt to mislead. Which would you prefer?

That one is merely false.

So to answer your question… yes.

7 Likes
  1. This is pretty rich coming from who it’s coming from.

  2. Read, highlighted, annotated.

7 Likes

I’ve rewritten one of the paragraphs:

We can use a simple word analogy here. Start with the word UNDERKING and add a typo to make it UNDERLING. An intelligent mind will immediately see the problem and correct the mistake. But what about unguided chemistry? First, the system would have to recognize there is a problem, and biology has no way to do that. If an organism dies because of a genetic spelling error, it is dead and cannot correct the problem. If an organism lives with a spelling error, it does not know that a problem exists. If the organism got really lucky and a back mutation happened that fixed the wording, all would be well. But what are the chances? It does not know there is a problem in general, and it cannot know which letter is the problem even if it knew there was a spelling error in that one word. Would another mutation ever be expected to fix it when there are so many other possibilities, like UNDERWING? And what would happen if a few more mutations happened before the fortuitous back mutation arose? We might go from UNCOCKING to UNCORKING, which is unhelpful. Back-mutations are only helpful if the original context is preserved.

English words are far more interconnected than creationists using them as examples realise. As noted here, almost half of nine-letter words have single-letter changes to other words. Note also that in words every letter is fixed, which isn’t the case in genetics where variation is possible with little or no effect.

There’s another problem with this paragraph, though. As the last sentence makes clear, Sanford et al are still working with the assumption that any deviation from the original genome is going to be worse, even if it moves the genome closer to another peak in the fitness landscape. Why should UNCOCKING->UNCORKING be unhelpful?

And in the original example, why should the reverse mutation BOUTNHELDE->HOUTNHELDE be unhelpful? First, it reduces the number of subsequent mutations needed to return to the adaptive peak. Second, Sanford’s genetic entropy is supposed to be about mutations with very small unselectable effects, so why is he assuming in this paragraph that any mutation is fatal? If HOUTNHELDE->BOUTNHELDE is one of a series of very slightly detrimental mutations, then BOUTNHELDE->HOUTNHELDE is very slightly beneficial, so it is helpful.

That section has other problems that can be highlighted simply by changing the examples used:

Worse, not all mutations have the same probability. If a T→C mutation (one of the rarer mutations we see) happens, the reverse [C→T] is more likely to occur. Thus, while we are waiting for our common C→T back mutation to happen, other T→C mutations won’t be occurring in the near vicinity.

That UNDERMINES the passage and UNDERLINES the problem.

Finally, Sanford’s comment at the end of the section serves only to illustrate that Sanford hasn’t understood the claim, and has nothing but assertions.

5 Likes

Oh yeah.

That shows us selection was not a significant factor in affecting which mutations accumulated.

I hope Sanford isn’t suggesting that influenza mutation is somehow exempt from selection!? The papers below say otherwise, including for synonymous mutations. I have much more if you want. Bold mine.

In our dataset, 31.6% of all mutations were lethal. Approximately 40% of all viable mutations were highly detrimental (<0.85), 50% mildly detrimental or neutral (0.85–1.05), and only seven were beneficial (>1.05). The fitness among all viable mutations ranged from 0.26–1.13 with a mean of 0.82. Nonsynonymous mutations were more deleterious than synonymous mutations, consistent with reduced genetic constraint at the level of RNA relative to protein. Two of the three noncoding mutations were lethal, consistent with the conserved roles of these regions in RNA packaging and genome replication.

Mutation and Epistasis in Influenza Virus Evolution

Non-synonymous mutations occurring in immunogenic epitopes can undergo positive selection driven by host immunity, and may lead to the virus escaping e.g. neutralizing antibodies. The major epitopes of IAV, also termed antigenic sites, are located on the globular head of the hemagglutinin (HA) molecule, which is encoded by the HA1 domain. … For humans, positive selection in these sites has been documented, and as little as one mutation in an antigenic site has been shown to affect the vaccine effectiveness. Human influenza vaccines are therefore, evaluated twice a year to prevent mismatches between vaccine strain and circulating strains.

Substantial Antigenic Drift in the Hemagglutinin Protein of Swine Influenza A Viruses

Here, we have constructed 53 single random synonymous substitution mutants of the bacteriophages Qβ and ΦX174 by site-directed mutagenesis and assayed their fitness. Analysis of this mutant collection and of previous studies undertaken with a variety of single-stranded (ss) viruses demonstrates that selection at synonymous sites is stronger in RNA viruses than in DNA viruses. We estimate that this type of selection contributes approximately 18% of the overall mutational fitness effects in ssRNA viruses under our assay conditions and that random synonymous substitutions have a 5% chance of being lethal to the virus, whereas in ssDNA viruses, these figures drop to 1.4% and 0%, respectively.

The Fitness Effects of Synonymous Mutations in DNA and RNA Viruses

4 Likes

I got turned off by the title. @PDPrice, do you consider yourself an expert or an ‘expert’? If it’s the latter, the title is inappropriate.

Actually a reversal is fairly likely. Gene conversion is heavily biased towards the GC allele.

Any claim that new mutations are equally as likely to be beneficial as they are deleterious flies in the face of both expectation and measurement.

The real question is whether the nDFE* is less than the bDFE, since the dDFE can be ignored here. Since the quoted statement is only clearly true as it applies to the full DFE, the point is irrelevant. The nDFE for normal effective population sizes is poorly characterized at best, largely due to how slight those effects need to be.

How can they say we don’t know the distribution of fitness effects when we can certainly know what mutations are accumulating in genomes over time?

This is another rejection of basic mathematics. If we don’t know the DFE, then it ‘could’ be that the nDFE>=0, which would obviously result in accumulation of mutations without deleterious effect.

*Neutral = absolute fitness effect less than the selective threshold.
DFE = distribution of fitness effects
nDFE, bDFE, dDFE = selectively neutral, beneficial, and deleterious portions of the DFE

2 Likes

The word “expert” in the title refers to the PS participants, not to me, and not even to Dr Sanford or Carter.

It certainly does when the parameters are realistic. We can’t run an experiment over millions of years, so this is the best way we can “empirically” verify the claims of evolution.

I still think you’re being misleading with this claim. As addressed in the article, we certainly can learn valuable information about the effects of mutations–the overall spectrum–and that’s mostly consistent of effectively neutral ones, by everybody’s account, including yours.

You’ve appealed to junk DNA to justify your claim that effectively neutral mutations are balanced in their effects. But you’ve also denied that effectively neutral mutations occur exclusively in junk DNA. How do we put these things together?

You understand you just said your model is accurate if your model is accurate, right? Whether or not the parameters are realistic and whether or not they are appropriately modeled is precisely what is in question.

6 Likes

I assure I’m well versed, but thank you for the gratuitous comment. I feel like I should just snap a picture and change my avatar to a highlighted and annotated page so we can skip the “I don’t think you’re actually familiar with Sanford’s work” bit whenever someone disagrees with me.

We kind of lost the thread when you stopped commenting on the premise there. Sanford assumes that fitness within populations will be equal enough so as to preclude selection. This is demonstrably not the case. We can document purifying selection. Carter and Sanford actually did this in their H1N1 paper, simultaneously showing “a declining ratio of non-synonymous to synonymous codon changes (dN/dS)” while also arguing that selection wasn’t occurring. If Sanford wants to assert that the baseline scenario is that everyone is so similar there are no selectable differences in relative fitness (which, to be clear, is what he asserts), he needs to provide empirical support for that premise.

We should be able to agree that this premise is simply not realistic, which means there can be no “gradual accumulation of reduced absolute fitness” without a reproductive cost.

If you think otherwise, that Sanford’s premise is valid, then by all means, what data do we have that support it?

3 Likes

This is an objectively false claim that demolishes the whole thesis.

Why are you regurgitating it? Have you even bothered to look at the fatalities over time?

And please don’t defend this by regurgitating their handwaving falsehood that “H1N1” refers to a specific strain. The HxNx designations are subtypes.

4 Likes

The central premise of GE isn’t justified: How do you know the mutations are overwhelming deleterious?

GE often refers to the works of Kimura and Ohta–predominately interchanging definitions for “nearly-neutral” even though Kimura and Ohta’s definitions refer to different underlying assumptions.

It seems that GE proponents want to make fundamental claims about the nature of “nearly-neutral” while simultaneously claiming the effects cannot be measured. This is problematic because GE is premised on the assumption that these mutations are, indeed, deleterious.

The effects of “nearly-neutral” mutations, which Kimura later distanced himself from Ohta’s working definition, are contingent upon the effective population sizes under question. Neutral Theory attempts to explain allele diversification through genetic drift–as opposed to a greater emphasis on selection. It also serves as a framework to understand how some deleterious mutations can be propagated when the effective population size is small. Nearly-neutral theory does not propose a run-away accumulation of deleterious mutations which become impervious to selective pressures.

GE does not seem to consider the underlying context of “nearly-neutral” and effective population size–rather, it asserts “nearly-neutral” is in effect at all population sizes.

13 Likes

Yes, exactly. One of Sanford’s implicit assumptions is that all mutations have a fixed, inherent fitness effect. What is “slightly deleterious” in the here and now will always have that effect, regardless of effective population size, environment, etc. Another impossible assumption for the list.

8 Likes

Welcome Christopher.

Great points.

I think he also ignores pseudoreversions or compensatory mutations, which could attenuate the fitness costs of these slightly deleterious mutations.

3 Likes

With an undergraduate-level understanding of genetics and a rejection of the creationist falsehood that junk=noncoding, perhaps?

1 Like

One of Sanford’s implicit assumptions is that all mutations have a fixed, inherent fitness effect.

This is true, but I think this is an orthogonal issue. Even if we assume fixed fitness effects–regardless of environment or population size–GE still does not establish that those fixed effects are inherently deleterious.

GE assumes that the effects are both immutable and negative for the organism. The former claim is demonstrably falsified through data. I would argue that the latter is also demonstrably falsified through large-scale variant databases like gnomAD. However, GE posits that the latter claim is simultaneously true, but otherwise unverifiable.

I’ve regularly brought this up on Reddit and with Paul, but have not received any cogent response to the issue–barring the misquoting of Eyre-Walker’s commentary on coding-region variants.

6 Likes