Why is evolution so controversial?

quote:

---

"We are more different genetically from people living 5,000 years ago than they were different from Neanderthals." Genome study places modern humans in the evolutionary fast lane
Any real comments?

---

Yeah -- it was a really bad study. It misrepresented the findings of the Voight et al. study and it didn't take into account how much easier it is to detect recent selection than older selection. The paper should be dropped into the ocean and forgotten.

I haven't been following this thread, and trying to go back and read it just gave me a headache. Is he making some kind of coherent point about linkage disequilibrium?

quote:

---

sfs unbelievable. Why are you even talking to me? I thought you had enough our last go-around.

---

Sorry -- I had no memory of having dealt with you before. I run into a fair number of creationists who don't understand genetics.

quote:

---

I will first ask you your opinion on why how much easier it is to detect recent selection than older selection.

---

Much easier. The long-haplotype tests that the Hawks paper referenced lose all power after about 20,000 years or so.

I still have no idea what argument he thinks he's making about linkage disequilibrium, but whatever it is, it's wrong. There's nothing about human LD that is at all suggestive of a recent origin for humans.

quote:

---

Is this the gist of your opinion?

---

Yes.

quote:

---

I have read some information concerning the topic of classic selective sweeps in humans. The claim is that there might be zero evidence that they happened in recent evolution. I think that would be about 250,000 years in evolution perspective.

​

A young genome of say 6000 years might just work out fine. Please speculate.

---

Speculate about what? Your summary of the claims of unidentified people? Make an argument and then we can discuss it. Right now you seem to be arguing both that there's been lots of selection in recent human history and no selection in recent human history.And no, a 6000 year old genome does not work out just fine. It contradicts everything we know about genetics: LD, heterozygosity, allele frequency spectrum, population differentiation, the recently published 45,000 year old human genome. It has nothing to do with reality.

quote:

---

I disagree, the process of crossing over will degenerate linkage between genes from generation to generation. Hundreds of thousands of years will dissolve links between genes involved in recombination. A young genome will exhibit high orders of linkage disequilibrium an older genome would not.

---

Crossing over reduces LD while new mutation increases it. The two are more or less in equilibrium, with fluctuations resulting from changes in population size and admixture. Admixture increases LD, after which it declines again. This can be seen nicely in that recently sequenced 45,000 year old genome. It showed similar levels of Neandertal DNA as modern humans, but the Neandertal contribution was in much larger chunks back then, since it hadn't been broken down as much by recombination.

quote:

---

The findings are correct by Hawks and there is an apparent acceleration in recent evolution of humans, about 100 times faster than in the past.

​

Your claim that the paper was mistaken in its conclusion based on method.

---

Yes, that is my claim.

quote:

---

Here is a citation about recent selective sweeps not being relavent in recent human history (~250,000 years).[...]
If this is true, the argument that you make about the methodology might be false. Your opinion although informed seems wrong.

---

Sorry, but this makes no sense. Hawks's claim is that positive selection, detected via selective sweeps, has become very frequent in recent human history. Hernadez et al claim that there have been very few classic selective sweeps in recent human history. While the second paper doesn't flat-out contradict the first, it certainly argues in the nearly opposite direction. By introducing the 2nd paper, you're supporting my opinion, not undermining it.(Nevertheless, there are serious problems with the conclusion Hernandez et al draw as well. First, the limits the set actually allow for lots of selective sweeps, especially at regulatory sites, which dominate positive selection. Second, they set no limit on partial sweeps and much selection on standing variation, which would not leave the signature they're looking for. Third, their test actually seems to be highly biased against finding evidence for selective sweeps, as noted in this psper: Genome-wide signals of positive selection in human evolution .)In any case, all of these papers require human genetic history to be vastly longer than 6000 years, so why are you introducing them?

quote:

---

Sorry for the bad detail look at it here: http://www.johnhawks.net/...celeration/accel_story_2007.html
This is from Hawks web site it looks to me like there is no discordance in data, look at ten thousand years (first point is 20,000 years), the trend is already started to decline. The downward trend does look like it continuos uniformly threw and past 20,000 years.

---

First, that plot is of when the "selection" is supposed to have started, not when it contributed to the genetic signature -- selection goes on for a fairly long time. Second, that plot (which I'd forgotten) in fact looks like an almost perfect illustration of the loss of power for older and older events.(The main problem with the paper shows up this plot: they interpret far too many sites as being under positive selection. That's connected to their misreading of the Voight et al paper (a misreading I confirmed with Ben Voight at the time, by the way). They're treating the entire high tail of the haplotype length distribution as representing selected loci, and that's almost certainly wrong.)

quote:

---

You continually move the goal posts, maybe you can claim that 10,000 years has got problems for detection now. If there was a problem of method you would expect an anomaly around 20,000 years (there is none).

---

What on earth are you talking about? I never suggested that long-haplotype tests fail abruptly after 20,000 years. They steadily lose power for older and older sweeps, and beyond roughly 20,000 years the power is low enough that you're not going to detect much. Why should there be an anomaly 20,000 years ago?

quote:

---

You are entitled to any opinion concerning methodology you like, but it is just an opinion.

---

No, it's really not just an opinion. There are a handful of people on the planet who are real experts in this class of test, especially as applied to humans. I'm one of them.

quote:

---

No I am not you are wrong.

---

That's some pretty impressive logic you've deployed there.

quote:

---

Even if you count each indel as a single mutation because those mutations affect coding (my citations) you get ~125 million mutation events (~45 million in chimp and ~45 million in humans the rest exist in both) regardless of bp lengths. This gives you a best similarity of .125/6.4 or ~2%.

​

I regress back to Nachman, Crowell.

​

t= number of generations since divergence (Generation =20 years)
k= percentage of sequence divergence Estimated at 2%
Ne= effective size of population ~10^5
(u)=mutation rate 1.1 x10^-8

​

t= .5(k/u-4Ne) from Estimate of the Mutation Rate per Nucleotide in Humans | Genetics | Oxford Academic

​

You still get a divergence time of ~14 million years.

---

The human and chimpanzee genomes differ in single-base substitutions at a rate of 1.23%. The single-base mutation rate is currently estimated to be roughly 1.1 x 10^-8/bp/gen. Using your other values, that gives t = 7.2 million years.If you want to include indels, you have to increase the divergence by one-seventh. Unfortunately, we don't have a good independent estimate of the indel mutation rate, but a rate 1/7th that of substitutions is completely plausible.

I don't know what point you're trying to make in your response. Using your (correct) formula from Nachman and Crowell, and the values you specified for ancestral population size, generation time and mutation rate, and using the best estimate for human/chimpanzee divergence, the estimated divergence time is 7.2 million years. If you use different numbers, you'll get a different result.

quote:

---

(u) for substations is ~70 per generation. 1/7 (u) makes (u’) = 10 mutation per generation in humans for indels only. (u’) is calculated by (10/6.4x10^9 ~ 2x 10^-9). (u’) for indels is ~ 2x10^-9

---

Or just take 1.1e-8 and divide by 7, giving 1.6e-9 mutations/bp/generation.

quote:

---

With repeats and low complexity DNA is excluded

​

2.37% -1.52% Gives ~.8% for human and chimp divergence concerning indels this seems low but it must be true.

​

Subbing in for indels gives:

​

t= number of generations since divergence (Generation =20 years)
k= percentage of sequence divergence Estimated at .8% (for indels)

---

Sorry, but this is just wrong. As others have pointed out, you're comparing apples and oranges. The mutation rate measures the number of mutations, while the divergence you're using measures the total number of bases changed. Since indels frequently change more than 1 base, you cannot use the Nachman and Crowell formula to predict the divergence (or calculate the number of generations). Not unless you know the mean length of an indel.Note that the estimate I gave for the indel mutation rate comes from the chimpanzee genome paper, in which there are 1/7th as many indel events seen as single-base substitutions. Comparing human and chimpanzee divergence at indels will just give you that 1/7th rate back again, since the mutation rate and divergence are coming from the same data.

quote:

---

sfs has confused the (k) with the (u), sfs can correct me if I am wrong.

---

Consider yourself corrected.

quote:

---

That is number of mutations per generation. not total. The (u).
Divergence is the total percentage of number of bases changed. The (k).

---

Quite. And what you're doing is counting the total number of mutations that have occurred (u x total number of generations since human/chimpanzee chromosomes diverged) and assuming it should give you k, the total number of bases different between the two chromosomes. That only works if each mutation changes one base; that's the assumption in the Nachman & Crowell paper. It's not true for indels. Your equation is wrong for indels.

quote:

---

I know you would like to reject indels altogether, all evolutionists would.

---

Get stuffed. It's possible that you will someday post an accurate statement about genetics or evolution, but if you do, it will be by accident.

quote:

---

As you are using them, (u) and (k) do not have the same units. That's the problem. Number of mutations and number of bases changed are different units. You need to put them in the same units.

---

Exactly. Multiply the mutation rate (= (number of mutations)/bp/generation) by the mean size of the mutation (= (number of bases changed)/mutation) and you'll get (number of bases changed)/bp/generation. Multiply that by the number of generations, and you've got k, the fraction of bases that differ (= (number of bases changed)/bp). For single-base substitutions, it doesn't matter, since it's 1 (base changed)/mutation.