Can’t speak for Crispr, but I would say this is rather unspecific in a context where explicitly mathematical models are asked about. Which “population genetics models” feature the math you are asked to show, exactly? What are the governing equations of the model you are pretending is what you were speaking of all along?
1 Like
Bill wants to use the math of evolutionary biology but reject the conclusions. Because he is, as I’ve noted elsewhere, dishonest.
1 Like
I see. What you want is to use hand waving cargo cult “science” as a pretext for supporting whatever conclusions you’ve decided in advance.
Maybe you’re succeeding in fooling yourself but you’re certainly not fooling anyone else.
Can you show this math can validate the common descent of deer?
I can refer you to the methods section of that paper, which will contain the mathematics.
The thing is, when I reference the literature I don’t need to follow it with ‘but the literature is actually wrong, for reasons I can’t explain!’
2 Likes
Presumably you’re talking about the previously cited muntjac paper. That’s simple enough. Figure 1a is the result of two phylogenetic analyses, both of which found the same tree. As in the typical Nature paper, the Methods section is unreasonably terse. But here are the relevant bits:
That was for the 4-fold degenerate 3rd-position sites of all the orthologous protein-coding exons in the respective genomes using the general time-reversible model of site evolution, a gamma distribution of site evolutionary rates, and an estimate of the number of evolutionarily invariant sites. Fit of data to a tree was estimated by bootstrapping, though they didn’t report the results for that; I would assume, given the absence of very short branch lengths, that all branches got 100% bootstrap support.
That was the analysis of the mitochondrial genomes, similar except that all positions and sites were used, there was no parameter for invariant sites, and instead of a gamma distribution of site rates, the sites were divided into rate categories. (I find the last questionable, but doubt it caused major problems with the analysis.)
Both model determine the probabilities of observing the data given a tree and a model of evolution, adjusting both the model parameters and the tree to find the one that gives the highest conditional probability to the sequence data. There are many parameters. Every branch gets a length parameter, there’s a shape parameter for the gamma distribution, an invariant-site parameter, and separate rate parameters for each transition and transversion, e.g. A-G, C-G, etc. All parameters were estimated from the data. That’s the mathematical model. So what’s wrong with it?
The chromosomal changes, on the other hand, were mapped onto the tree without the need for any model, though I suppose you could call it parsimony: suppose that every change happened only once, so you put it on the branch just below the species that have that condition. What’s wrong with that?
And this claim is “vacuous” – “Not properly filled out or developed.”
No Bill. Until your claim is sufficiently properly filled out and developed, with specifics as to what “observed genetics changes” are inexplicable and why (which requires math), then your claim is simply too vague to point to a single model to counter it.
Taking a wider view, “observed genetics changes”, taken without any limitation whatsoever, would appear to be the purview of the entire Theory of Evolution.
Therefore Bill:
I cite the entire field of evolutionary biology, from Darwin’s publication of Origins to the present day, including every journal, paper and model on the subject, as my counter-argument to your claim.
Have you read, and understood, everything I just cited Bill? No? Then you have some work to do. 
The problem is that this “alternative model” is likewise “vacuous” – “Not properly filled out or developed.” It appears to consist of nothing except Bill’s Personal Incredulity at the Theory of Evolution. There is therefore nothing there to explore. If you had any substance to this “model”, you would have brought math demonstrating it – you have no math, so you have no substantive model.
2 Likes
It occurs to me that it might be illuminating to wind back this whole discussion, and substitute “development of human civilisation” for “observed genetic changes” and Chicago for the ancestry of deer.
A hypothetical contrarian, let’s call him “Will Cole”, claims:
There is no model or even the concept of a model that explains the development of human civilisation.
Should the gathered hypothetical sociologists, discovering that they cannot cite one singular model that explains all aspects of the development of human civilisation, or even all aspects of Chicago, agree to explore an alternative model, in which cities magically popped into existence?
Or should they demand that Will Cole be more specific and rigorous in his claim, before they even attempt to address it?
2 Likes
Hi John
The math we are discussing is population genetic mathematics that that estimates if there is time based on reproduction and natural variation to conclude a reproductive connection between the species.
I agree the papers infer a reproductive connection based on phylogenetic analysis.
The papers conclude tandem fusions and fissions again this is based on inferring a reproductive connection between all the species mentioned. Parameters like the mutation rates, rates of deleterious mutations, neutral mutations and positive mutations is not part of the model.
I see that a reproductive connection is inferred yet I do not see a strong case for this inference which is simply based on the pattern of the data.
Why is that not a strong case? You asked for a model. I showed you the model. How, absent common descent, can you explain the pattern? Are you perhaps claiming that there’s somehow a dependency graph of 4-fold degenerate sites?
There is no maths to discuss, because you refuse to provide any, claiming instead - falsely - that none is required.
This statement seems to walk face-first into a question I’ve been contemplating over the last day or so – what scale of modelling would be required to answer Bill’s claim with the degree of detail he seems to be demanding?
I’m not an expert in this, but it would seem that a separate population genetics model would not only be required for each species, but for each distinguishable time period (in terms of population size, sub-populations, competitive environment, etc) that species found itself in over time. This could easily result in multiple (dozens of?) models for each species.
This would appear to have some startling consequences:
-
It would appear to be likely to produce enough output to fill an entire scholarly journal, let’s call it the Journal of Answering Bill’s Deer Ancestry Question, for decades.
-
The cost of this enterprise would likely run well into the millions of dollars. Do we know of anybody with that much money that cares that much about deer ancestry?
-
The models would likely require an enormous amount of data. Do we, for example, know enough about the circumstances of muntjac deer during the last ice age, to formulate a detailed population genetics model for that species and time period?
As I said, I’m not an expert. Would any of the experts like to weigh in out what level of scale we’re contemplating here?
Speaking as a semi-expert on the evolutionary biology side of the question, I have to say this is more of a question about Bill’s psychology, on which there are no experts. But based on the data we have so far, I would judge that no model would ever satisfy Bill, because the request for a model is just an excuse not to believe the conclusion, and that can’t be changed.
1 Like
Yes, as I specified in an earlier post above, Bill’s “request for a model” is vague to the point of being essentially vacuous. But based on his rhetoric, it would seem that he is at least after models that would (i) give an estimate of the number of gene differences expected in one of his Venn diagrams, and (ii) the expected number of changes in chromosome counts. That would seem to give a lower bound on the level of modelling he is demanding. Of course if, by some miracle (including a quixotically-generous benefactor) we were able to produce this modelling, I have no doubt that he’d demand more – but it would seem to provide sanity-test on whether what he is demanding is somewhere within a few light-years of reality.
The question has always been about using population genetics to reconcile deer common ancestry.
Phylogenetic studies set a very low bar of showing the pattern of the data and not how the genetic changes occurred.
Initially prior to seeing the chromosome divergence and different genes I never thought generating a population genetics model of genetic differences between deer would be a problem.
It appears population genetics studies, in the case of deer, are limited to specific populations such as white tailed deer or musk deer.
Reconcile deer common ancestry with what? What data would you test your model with? I don’t think you have any idea what you’re asking for.
But the GTR+gamma+I is in fact a model of how the genetic changes at 4-fold degenerate sites occurred. It’s a model that more or less assumes neutrality + purifying selection at different sites. How does this not fit your requirements?
What about the data changed your mind? You must have some vague model in your head under which the data are unlikely. Otherwise you couldn’t make any claims.
They usually are, sure. That’s because population genetics is the study of the genetics of populations. Is this somehow surprising?
6 Likes
With a model that takes the genetic changes into account with parameters that take empirical data into account.
I am not sure what this means.
Initially the large number or chromosome changes where chromosome mutations are not frequent and are often deleterious.
Second a paper you posted that shows non shared genes between the species.
I assumed that deer could be treated as a single population but I was clearly wrong.
Then you have no business claiming that it’s inadequate. Go ahead: ask for clarification of the parts you don’t understand, specifically. Most of it was explained just previously.
The family Cervidae??
I did not claim it is inadequate. I am interested in your argument here.