Just a few points of clarity, after thinking the discussion, so I can try to make sure I’m not any more confused than I need to be : I don’t see how you could interpret a tree like Jeanson does in terms of population growth unless the tree was relevant to genealogical relatedness. If I can see that, and creationists believe in created variation, lol, I can’t see how they would interpret any tree that way.
I don’t see how those trees could simulate multiple SNPs per generation in the y chromosome and still describe what they say they are describing.
It depends on what sorts of trees you are talking about. Distance trees like Jeanson uses are simply clustering sequences by their similarity. That’s it. Coalescent trees are gene genealogies. Not genealogies of individuals. Gene genealogies exist in populations even when those populations are constant. I explained exactly what a gene genealogy in a growing population would actually look like and why in the video and Jeanson’s approach applied to an actual coalescent gene genealogy would draw a wildly different conclusion from reality.
For example, I use gene genealogies to infer species trees. Think of a species tree as a big bifurcating set of tubes and inside these tubes are branching gene genealogies which may or may not match the branching pattern in the species tree.
These are not easy concepts to wrap your head around. If you think it’s something you can figure out in a chat room or Facebook thread then for most people you are probably wrong.
Put away your device get off these groups for a while and read a good population genetics or molecular systematics book. A good recommendation I have is David Baum’s book Tree Thinking. It’s an excellent undergraduate introduction to genetic trees, basic molecular systematics and evolution, and coalescent theory which is at the heart of most modern molecular systematics and population genetic approaches.
Wow. They probably believe that a WWII bomber was found on the moon, too.
I suspect “most believe” really means “I believe and so most people would also believe”. But it may mean “This isn’t true but none of the rubes will check”.
Right. Got that too. I haven’t watched the video yet but my problem is this: gene genealogies are the more complicated scenario. Even I know not to apply Jeanson’s method for the y-chromosome to them. I can’t imagine how he doesn’t know.
But a tree created from uniparentally-inherited DNA with multiple mutations seems to be a very simple scenario. You could introduce it to high school students and maybe have them create a tree too.
So my questions are:
Does his method work with the high mutation rate? I don’t know yet, but I don’t see why not. I could probably figure it out on pencil and paper.
Does it work with a low mutation rate? (The more important question - because if it does, it proves the high mutation rate is accurate)
Does it work in more complicated scenarios? I already knew the answer to that was no, and that’s been covered here.
It would probably be smart for Jeanson to run some simulations and give the answers to #1 and #2 in a paper.
I like the metaphor. Sounds like a fun job.
Thanks. I would like to understand all approaches so it would be good for me to learn an evolutionary approach more systematically.
Do you not see an obvious conflict between your confession of confusion here and your confidence in claiming that Jeanson is right (while doing nothing empirically), while all of the actual geneticists who produce the data Jeanson is “reinterpreting” are wrong?
Y-chromosomes of course are displayed as gene genealogies. I’m not describing some special case. And in fact virtually all of the papers Jeanson cites regarding human Y-chromosome history use these coalescent approaches and gene genealogies. By gene I’m just referring to any stretch of DNA that is passed on intact from one generation to the next. All the arguments I’m making where I’m referring to gene genealogies are relevant to Y-chromosomes, or W chromosomes in birds, or mitochondrial DNA, chloroplast DNA, or any non-recombining chunks of DNA in the nuclear genome and probably all other sorts of genetic material I’m failing to mention.
I’ll add as an aside that Jeanson acts as if there is NO recombination in Y-chromosomes. There in fact is and it’s likely more than expected. The pseudo-autosomal region of the Y-chromosome for example does in fact recombine with the X chromosome and exchange bits. There is also recombination in the X-degenerate region, which ironically Jeanson is at least in part using in his work while remaining silent on why it is called the X-degenerate region (Hint: it doesn’t bode well for special creationism).
I took the time to watch your and @dsterncardinale video on his Creation Myths channel. I did know about some of his methods because of the questions I asked on the forum last year. But I learned exactly how the screenshot was used via your descriptions, so that was amusing
However, if you criticize his approach by saying an exponentially growing tree has to look like the bottom section of your figure below, this is obfuscation. (I’m not saying it’s purposeful on your part, because Jeanson’s distance-based genealogical tree is so, so, so simple and you’re used to working with coalescent trees and extremely complicated scenarios.)
For example, the Karmin et.al. distance-based tree that Jeanson uses does not look like those coalescent-based trees that have exponentially growing populations. Instead, there are long branches at the top and many short ones at the bottom.
There is no reason Jeanson’s approach shouldn’t work in a tree that is a genealogy of individuals, like it has to be with a mutation rate that high. My only question is what happens in a distance-based tree with a slower mutation rate - does this genealogical approach still work? If not, he doesn’t disprove the slow mutation rate. However, if he keeps showing that his approach keeps working, then it provides more support for the higher mutation rate.
Not exactly. Though neighbor-joining is a clustering method, it actually approximates a least-squares fit. Still, I see no good reason to use it here in preference to ML or MP.
Figure 1. The phylogenetic tree of 456 whole Y chromosome sequences and a map of sampling locations. The phylogenetic tree is reconstructed using BEAST. Clades coalescing within 10% of the overall depth of the tree have been collapsed.
You beat me too it. I forget exactly what Jeanson does here off the top of my head but I think he tries to recreate the tree from Karmin et al by using neighbor joining. Either way Karmin et al is using Bayesian coalescent based methods in BEAST and fastsimcoal in their analysis and Jeanson’s method of counting branches as a proxy for population size is silly with pretty much any genetic tree.
Totally missed Figure 2 huh? The one with the Bayesian skyline plot and all those little short branches that Jeanson thinks occur during expansion actually coinciding with a population bottleneck. Figure - PMC
However that tree is created, they have to collapse the tree because there are many of short branches at the tips. Go ahead and tell me that tree represents a constant population if you want. You may as well as also say the sky isn’t blue on a sunny day.
Again, read @Herman_Mays words carefully. When I did, I realized all he is saying is that Jeanson’s method doesn’t work for coalescent trees. Yes, we all know that. And he has said Jeanson isn’t using coalescent trees. So he is implicitly admitting he hasn’t actually critiqued Jeanson’s method at all. It’s so simplistic Jeanson even called it a kindergarten point in his video introducing the idea. It’s not hard to understand, so I only see obfuscation here.
What you’re saying boils down to “I understand that Jeanson used the wrong method, but he could still be correct”.
The word in dispute, for you, might be “wrong”, but rest assured, he uses the wrong method to answer the question he was trying to answer. You can argue that his answer is nonetheless correct. But you can’t do so because his results indicate so.
I mean take the time to look at the papers you are talking about. Jeanson (2019) takes the Karmin et al (2015) tree and treats it under the exact same parameters he treats the neighbor joining trees he derived from their data.
“I utilized two types of Y chromosome phylogenetic tree data from Karmin et al. (2015). The first type of data was taken from their Figure S3 and from their Table S7, which lists all the branch points and associated time values. The second type was derived from their accompanying primary data in their online VCF file, which I converted to FASTA form, and then used the FASTA file to draw a neighbor-joining tree (see accompanying Jeanson and Holland 2019 paper for details and references on how I derived the tree from the VCF file).”
So it’s simply not true that Jeanson is drawing any distinction at all between a Bayesian tree generated using a coalescent approach and his distance-based neighbor joining trees. He’s taking both trees and submitting them to the same bizarre analysis. He clearly doesn’t understand what coalescent processes are at all despite the fact these concepts are at the very heart of the population genetic approaches he claims to be an expert in.
The “time values” Jeanson refers to in the Karmin et al coalescent tree are a function not just of the mutation rate but also the population size. He has no clue about this and is treating the two sorts of trees in precisely the same way.
: I don’t see how you could interpret a tree like Jeanson does in terms of population growth unless the tree was relevant to genealogical relatedness. If I can see that, and creationists believe in created variation, lol, I can’t see how they would interpret any tree that way.
