That’s because there are no strictly neutral mutations. It’s asymptotic.
If there were a continuous distribution of fitness effects, and above 0 it declined in some smooth way, then it would be very unlikely to have a sharp bend in it anywhere. The sharp corner in the distribution that PDPrice drew is easily explained – it is owing to his drawing an approximate curve. We should seek the source of the sharp corner by interrogating his psyche.
6 Likes
Oh yeah. Here’s another question. You’re saying the distribution differs on the right side because of optimization. But I don’t understand why that would only kick in at the threshold of selection. Let’s keep in mind, these effectively neutral mutations are not strictly neutral. They are in fact making improvements or damages. So why wouldn’t optimization also come into play below the threshold of selection? Why wouldn’t optimization in fact come into play right at the Y axis?
This is too vague to be meaningful. I think the experts would agree that most life is already highly adapted, so to most organisms in their current circumstances, we are in a position approximately like that depicted in the curve.
No it wouldn’t be meaningless at all, it just means models have limitations and you should take them with a grain of salt when you extend them into regimes that differ substantially from those that were approximated when the model was invented.
No, you’re not in agreement with Dr. Schaffner based on what he’s said here. He agrees the sharp corner should be there, and he says it owes its existence to the fact of many generations of optimization by means of selection. Without the sharp corner, you wind up saying that selectable beneficials are equally as likely as selectable deleterious mutations. And there’s a huge body of literature available to refute that idea.
For what it’s worth, the focus on junk DNA strikes me as a bit of sideshow. The amount of DNA in a genome that is likely to be “junk” (for lack of a better word) is dependent on a number of factors and is highly variable among most species (less so mammals). Conversely, genetic entropy is, I believe, purported to be a fairly general phenomenon–at least in the sense that we would expect it to be occurring in species with and without lots of “junk”.
You are correct. GE can and will happen regardless of what fraction of the genome is purported to be junk. You can even use Mendel’s Accountant to test this, with varying levels of junk content.
Earlier I was under the impression that Dr Schaffner was saying that all the effectively neutral mutations were affecting junk only, and that’s why they were equally likely to go in either direction. In fact, I’m fairly certain that’s what he implied in our previous exchange, linked in the top of this thread.
But here he has denied that, and has suggested that a sizeable portion of effectively neutral mutations are affecting functional DNA as well. So that seems problematic, unless he can establish that somehow that effect is in perfect equilibrium on both sides of the Y axis.
Why does it need to be in perfect equilibrium? Maybe the slightly beneficial mutations are more likely.
1 Like
Right. That would show a gradual increase in fitness over time. But that would be counter-intuitive. Dr Schaffner appealed to the concept of optimization in his explanation of why the number of beneficials steeply drops at the threshold of selection.
The problem is, since he has now stated that effectively neutral mutations don’t affect junk only, but also functional DNA, it appears that his own argument has turned back against him. Optimization would come into play, and would suggest we should see fewer effectively neutral mutations of the beneficial variety compared to the deleterious ones.
Biology is usually counter intuitive, especially population genetics.
I think it is worth noting that the function is almost certainly not stable and fixed. Likely, as negative mutations accumulate, positive ones become more likely. As positive mutations accumulate, negatives ones become more likely. This might function somewhat like a thermostat. It would not require any special mechanisms, but be a consequence of the underlying fitness function and dynamics.
This is related to the point that people are making about real genomes being optimized, but it is not exactly the same.
I’d point out that we are talking very theoretically here. There is no way to directly measuring these things. Maybe there is no steep drop off at the threshold of selection. So maybe there is no need to explain it.
You don’t understand why optimization due to selection would only kick in at the threshold of selection?
1 Like
No I don’t. You’re conflating past selection with current selection. See his explanation.
There has to be. We have to account in our diagram for the fact that deleterious mutations are much more likely than beneficial mutations. There is overwhelming evidence for this fact, at the very least as it pertains to mutations which are selectable.
Mathematically speaking that is not true.
There is no reason this fact implies a steep drop off or a kink. One side of the distribution can have a long tail, and the other a short tail. That distribution would be consistent with deliterious>>beneficial, without a steep drop off right at the point of selection.
I am not. Past selection is the thing that did the optimization, and removed from the pool of potential mutations the majority of the adaptive ones. The DFE represents the selection coefficients of currently possible mutations, which would be depauperate for beneficial mutations at the threshold because of this prior optimization.
1 Like
I don’t think so. Even if it was, that does not change my point.
That is an argument that would support my point.
The prior optimization has no logical relationship with the threshold of selection. They’re completely disconnected ideas. Mutations below the threshold can still affect function, and thus optimization should play a role there.
I retracted that comment. I was getting my X and Y axis confused.
But Dr Schaffner’s mentioning of optimization was appropriate. It will affect the distribution in a big way, but I don’t see why the kink should start at the threshold of selection. It should be a totally different-sized curve on the right side of the Y axis.
Your figure fails to include the large number of strongly left-biased lethal mutations. A ‘mostly’ symmetrical DFE at neutrality with this lethal mode as a second exponential curve would both better represent observations and account for the discrepancy. Obviously, these mutations are outside the scope of GE, as they have selective coefficients much greater than the threshold.