What is the observed human deleterious DFE of coding regions?
What is the observed human deleterious DFE of non-coding regions?
What is the observed human beneficial DFE of coding regions?
What is the observed human beneficial DFE of non-coding regions?
Why does Sanford not use the observed rates in his simulations?
Why does Sanford’s simulation not correctly predict the outcomes of real organisms in real experiments?
I can write a simulation for my bank account to predict I will be a millionaire next month. I can then write a paper about it and post it on my blog. I can convince my gullible friends that I will be a millionaire on the basis of my simulation. I can use the exact fundamental mathematics of banking and investments. All I have to do is plug in ridiculous interest rates.
I still do not understand how this concept is escaping advocates of GE.
Look at the materials and methods portion of the paper you linked:
“Fraction of beneficial mutations = 0.0. While beneficials are desirable in them-selves, they confound selection against deleterious mutations, tending to make the STd problem worse. The effect of beneficial mutations on STd are dealt with in a companion paper.”
“Distribution of mutation effects = Weibull distribution, wherein 0.1% of all mutations reduce fitness by 10% or more. This results in a mean mutation effect which reduces fitness by roughly 0.1%. Altering the shape of the distribution to be either steeper or less steep does not significantly affect the STd phenomenon.”
The sort of mutational effects discussed in your quoted passage are precisely what I explicitly discussed in the second paragraph of my comment, so I assume you didn’t not actually read the comment you responded to. Avoid this in the future.
The human haploid genome is about 3 billion bases. About 90% of that, or 2.7 billion bases, is accumulating mutations at a rate consistent with neutral drift. These 2.7 billion bases would be the areas where nearly neutral deleterious mutations would occur. There can only ever be 2.7 billion of these mutations in the human genome.
If each nearly neutral deleterious mutation reduces fitness by 0.0000000001% then 2.7 billion of them will reduce fitness by 0.27%. So my simulation shows that these mutations aren’t a problem.
Not really an “experiment” when it’s a simulation using Sanford’s physically impossible assumption of a fixed, not to mention dubious DFE of mutations.
This passage and several others in the paper seem to indicate that selection happens to mutations, rather than to individuals. If this is the case, an individual could accumulated many slightly deleterious mutations, without suffering selection, rather than selection acting against the cumulative effect of all mutations within the individual. This just cannot be right, but I don’t see in the paper where Sanford make this meaning clear. Can someone please point me to where he give the correct interpretation?
By slightly silly example, suppose a human mutation to have no left kidney is “slightly deleterious”. Likewise, a mutation to have no right kidney is also slightly deleterious. Is Sanford really saying that an individual could have both mutations and no kidneys(!), but selection could not act because neither mutation is individually selectable? No, this just can’t be right, but I don’t see where he says otherwise.
Then it likely would have been wise to ask for clarification first, would it not? Especially, for instance, before making a second response to an even later comment referencing that paragraph while still not understanding it! But I’m sure you wouldn’t do that, right?
When we say mutations have ‘fitness effects’, what we mean is that mutations cause affect traits and this difference in phenotype, in turn, affects fitness. Some traits, like the amount of junk DNA, might be considered as strictly deleterious, since the organism needs to copy all of that DNA but it gets nothing out of it. However, we know that whatever impact on fitness this might have, it must be infinitesimally small for complex organisms. This is because there are organisms that thrive in their environments in spite of tens to hundreds of gigabases of junk DNA. In other words, such mutations are no help to proponents of GE, because accumulating enough to cause appreciable fitness decline would take so many generations that the sun will likely burn out before it could happen. But Sanford claims that it isn’t these infinitesimal effects that are responsible for GE, but the ones close to the threshold for effective selection. Unfortunately for him, and you, this is equally problematic. Let me lay it out point by point:
Most traits have a fitness peak, a trait value that maximizes fitness for that particular organism in its particular environment.
Most organisms are well-adapted to their environment, meaning the mean trait value is close to the fitness peak for most traits.
Mutations affecting traits will shift the trait value for carriers (by definition).
The mutations from (3) are equally likely to shift the trait in either direction.
Because of (4), equal numbers of mutations in both directions will reach fixation in the population.
Because of (5), the mean trait value for the population will not change.
Because of (6), the mean fitness of the population will not change.
Because of (1) and (2), most of the mutations from (3) will be deleterious.
To conclude: The accumulation of mutations affecting traits with fitness peaks will not decrease fitness (7) in spite of these mutations individually being deleterious (8).
In other words:
Sanford described conditionally deleterious nearly neutrals, not strictly deleterious nearly neutrals. Conditionally deleterious mutations are of no help to Sanford at all. Again, this was clearly explained in my comment.
You may want to refer to this thread for a rather lengthy discussion of exactly this question. The short version is that if the deleterious effects are spread evenly through the population, then the relative fitness impact of the mutations is never selectable.
To rework your analogy: Imagine a series of mutations reducing the activity of some enzyme by 0.001%, with an equivalent fitness effect. While 100 such mutations might be selectable relative to none, it would not be selectable relative to 99. And so the mutations might accumulate. At some point, the reduced absolute fitness might lower fecundity of the populations, but again this would not cause a selectable difference because any individual, relative to its peers, would be ‘nearly’ as fit. Reduced fecundity of the population would reduce the size of the population, and reduced effective population size would reduce the efficiency of selection further resulting in an extinction vortex. Effectively, Sanford is claiming that mutation meltdown should occur regardless of population size. Which is nonsense.
Okay for 1, 2 and 3.
I disagree with 4, 5, 6 and 7. I am pretty that most population geneticists would disagree too.
As for your conclusion, I really don’t see how the accumulation of mutations being individually deleterious could not affect fitness. This seems nonsense to me.
Selection coefficient for a mutation is based on the size of the difference and the distance to the fitness peak. If there is a bias towards one direction, then the population mean will shift in that direction, increasing the selection coefficient of the mutations, both new and old, rapidly shifting the peak back to where it was.
I hope you only disagree with these because you disagree with the four, as they all necessarily follow from it and basic population genetics. Rate of fixation under drift is approximated by rate of mutation, so if drift is the dominant force (which it must be, that’s the point) then the rate of fixation for both directions must be equal if the rate of mutations in both directions is equal. Which, as I just explained, it would be. If the rate of fixation in both directions is the same, then there is no change in the mean trait value for the population, and therefore no change in mean fitness.
Do you think that 1-1=0 is nonsense? Because it is the same idea. The mutations are only deleterious if their effects are applied to the population mean independently. A mutation that slightly decreases a trait is bad, a mutation that slightly increases the trait is bad, the combination of a both is the same as having neither.
Let’s say trait X has a fitness peak at 100 units. Mutation A decreases X by 1. Mutation B increases X by 1. Consider the following:
I think your analysis is highly disconnected from biological reality. Take an enzyme E in organism O. I agree that the fitness of E in O will probably be near to its optimal. So I can indeed see that a mutation that would increase the activity of E can be as deleterious than a mutation that would decrease its activity. But if we are to remain biologically realistic, we have to recognize that the former type of mutations is much rarer than the latter type, for there is obviously more ways to reduce an enzyme activity than to enhance it. And this is even more true when considering complex enzymatic machineries involving many different partners.
You can think whatever you link, but here you would be empirically wrong.
Good.
Changes in enzyme activity are almost certainly selectable, and even still I reject the assertion that there are more ways to decrease activity slightly than increase it slightly. Even if I were to accept that there were a large number of nearly neutral differences in enzyme activity and that such mutations were heavily biased towards decreased activity, you would still be faced with the problem that fitness effect size increases as the trait moves further from the fitness peak (as I described), preventing continuous accumulation in a single direction. And the fact that mutations affecting a single enzyme are certain to be inherited together, meaning you would need to wait for fixation of each mutation individually before the next occurred could become fixed by drift, which would take far too long to be relevant. So protein activity differences are no help to GE.
At the same time, changes in the level or timing of expression may be nearly neutral and may segregate independently, but it is complete nonsense to suggest that there are more ways to manipulate these in one direction than the other. These differences in regulation are in turn responsible for most morphological variation, and it is morphological variation that is has the best chance of producing fitness effects on the scale needed for GE. But again, this is precisely the problem: Fitness effects from morphological differences are always conditional, and conditional effects are no help to GE.
So you still need to provide a class of 1) strictly deleterious mutations (because conditionally deleterious mutations don’t help) with effect sizes 2) less than the selection threshold (because selectable mutations don’t help) but still 3) large enough to matter in relevant timescales (because needing 50 billion years doesn’t help). If you can’t manage three out of three, you can’t support GE.
So far, the best you’ve managed is one out of three.
Sure, but generally not rare enough that enzymes can’t be optimized by selection. And enzyme activity-affecting mutations are generally visible to selection for enzymes that are rate-limiting for growth.
You will not find evidence for GE in the evolution of enzymes.
If you’re alleging that someone is disconnected from reality, do you really think that you can support that point with a hypothetical? Why not a real enzyme in a real organism?