Sure. First here’s a nice blog post by Arlin Stoltzfus on how to understand and contextualize such nubmers:
http://www.molevol.org/the-range-of-rates-for-different-genetic-types-of-mutations/
And here’s a review article on the rates:
4 Likes
Predictive text should be taken out and shot. The renegade apostrophe maybe should be poked with hot pincers first.
It’s possible, and there certainly must be a few. I just point out that the diagram includes a great many paralogs, and that it lack of detectable homology isn’t the criterion.
1 Like
First, thanks for the citations.
I agree that a duplication being fixed is 20k more probable than a specific duplication. The frequent duplicate loss in the population according to the paper is due to the high frequency of duplication being deleterious.
There is a waiting time problem in both cases.
In the case of Behe/Lynch they are looking at observed adaptions such as a new binding site or a disulfide bond. They are only claiming the observation of “bullet holes” and calculating the probability of the observed pattern given the hypothesized cause of gene duplication and variation.
I agree and we can quantify the difference.
The problem is the rate of fixation of changes. I agree that we will see more variation in vertebrates per generation then bacteria. The challenge is new functional gene sequences being formed and fixed in the population. I am extremely skeptical that a new unique gene sequence is likely to form from neutral mutations.
I think @Paul_Nelson1 will eventually be vindicated for his stance on common ancestry specifically his skepticism of being able to explain diversity by reproduction and natural variation. If he is right this is very important for the biological sciences.
What I agree with is Mike Behe’s method of detecting design and I think the Howe diagram may be used to potentially detect design with the pattern of genes. If we consider genes “parts” we maybe detecting their purposeful arrangement.
Behe admitted that a “problem” would only exist if the particular parameters he used in his model were correct, that there were likely other models that could account for the observations without invoking design, and that Lynch had provided one such model. He says as much in his response to Lynch’s paper.
2 Likes
No, there isn’t. At a middle of the range of the rates given in that paper (~5x10^-6 duplications/gene/generation), with the human generation time of ~20 years you’d get over 2 million gene duplications since the last common ancestor of the species in the Howe diagram, in each lineage. 2 million > a few thousand “novel” orthologues.
So in fact there could potentially have evolved many many more new genes than we actually see.
Now clearly for most of that history the generation time was much shorter for every extant species it depicts, and every extinct ancestor.
2 Likes
You are assuming the gene duplications are neutral where most are deleterious. You are also forgetting the time of divergence of each duplicated gene and assuming that it can find function from neutral mutations. Neutral mutations are much more likely to turn the gene into a pseudo gene.
You are about to be burned down for saying something you almost certainly know is wrong. Please reconsider - and maybe cite a reliable source.
1 Like
Hi Dan
From the paper Rum cited.
The fixation of a gene duplicate in a population faces multiple obstacles. First, there is a high probability that the duplicated gene is lost from the population by random genetic drift. Moreover, most gene duplications are probably detrimental to organismal fitness. They can perturb optimal dosage balance between genes contained in the duplicated regions with genes elsewhere in the genome, and increased gene dosage can be costly because of superfluous gene expression (Papp et al., 2003; Veitia, 2004). Empirical estimates of this cost in Salmonella was found to be substantial (3–16%; Reams et al., 2010). In addition to reducing fitness, many gene duplications are inherently unstable, particularly if they are in tandem orientation or flanked by repeat elements (Anderson and Roth, 1981). Lastly, given that most mutations are degenerative, a duplicated gene is much more likely to end up as a pseudogene than to acquire a function that is distinct from the ancestral gene and actively maintained by natural selection.
Well then I may owe you an apology - especially if @Rumraket can verify that. It seems like some pretty important context is missing though - my own fault for not reading up.
1 Like
OTOH, after reading that paper I find that much of what follows your quote goes into the reason duplications might remain in a population despite this. Duplications are common, but most get weeded out before fixation, and that should include all or nearly all deleterious duplications. I don’t think you have fairly represented what the paper is actually presenting.
OK, I apologies for jumping the gun and criticizing you - the paper did indeed contain that statement. 
I still disagree about waiting time and the practical application of these mutation rates. The probability of duplication is relatively large, and so there will be many duplicates available for mutation and selection. What is the probability that nothing beneficial results? (We’ve had that discussion before.)
3 Likes
No, I have made no such assumption. I have merely stated the total number that could in principle evolve given their genome-wide average rate of occurrence.
Completely irrelevant to the Howe diagram, where no claim is made about the functions of the novel genes. They’re considered novel only through their lack of similarity on various scores, not because it is thought they all evolved new functions. Though they might, there is no in principle barrier to that either.
Incidentally over that same period of time there could have been something like 2.2 billion mutations, enough to almost completely scramble 2/3rds of the human genome, at the avg human mutation rate. So there is just no problem with the amount of sequence divergence either.
Neutral mutations couldn’t possibly turn a functional gene with a positive fitness effect into a pseudogene. The duplicate would have to be deleterious to begin with, in which case that just serves to explain why there hasn’t actually fixed 2 million new duplicates! But nobody said all duplications that occur would evolve into new genes, just that there have been enough duplications to account, in principle, for the inferred number of novel genes depicted in the Howe diagram.
Whichever way you put it, there is nothing about the rate of evolution of duplicate genes that constitutes a problem, and you have no basis for claiming there’s a waiting time problem. Even with gene duplications alone there is more than enough to explain every novel gene in the howe diagram. Most duplications being deleterious when they occur then simply explains why there didn’t evolve millions of them, but somewhere on the order of a few hundred to a few thousand, instead.
2 Likes
Hi Dan
Why do you think the Behe/Lynch model represents an adaption of only 2 mutations after a duplicate gene taking up to a million generations? How then do we account for a very different gene sequence with a unique function?
From the paper:
a duplicated gene is much more likely to end up as a pseudogene than to acquire a function that is distinct from the ancestral gene and actively maintained by natural selection.
This supports Rums paper on the low probability of duplicated genes reaching a new novel function.
https://doi.org/10.1126/science.1102033
This paper argues for a much lower duplication rate then Behe/Lynch used.
Which is why 2 million didn’t evolve, but a few hundred instead. Where is the problem Bill? Why do you show no calculation with the given rates?
1 Like
“Up to.” IOW, it can take fewer generations.
You are extrapolating far more from those papers than is warranted. This is what those papers are intended to show.
Behe’s paper intended to show that a new function cannot evolve from a duplicate gene and then become fixed in a population in a reasonable amount of time.
Lynch’s paper intended to show that Behe’s conclusion arose from specific assumptions made in his calculations, and that under different (and more realistic) assumptions Behe’s conclusions were incorrect. That is to say, Behe failed to demonstrate that such a waiting time problem exists in real life.
In his response to that paper, Behe admitted that Lynch was correct (although in his typically obtuse manner that allowed for plausible deniability among his ID Creationist followers):
Our model posited necessary intermediate mutations to be deleterious in the unduplicated gene; Lynch’s model assumes them to be neutral: “all 20 amino acids are equally substitutable in the intermediate neutral state” (Lynch 2005, this issue). All of his objections to our work stem from this difference.
That’s it. End of story.
3 Likes
Those are fixed gene-duplicates in yeast, not the rate of occurrence of duplication mutations “in vertebrates” that you asked for.
I quote that paper for you again like I did a month ago:
Note that by “gene duplication rate we” mean the rate at which a duplicated gene is created by mutation and becomes fixed in the population. The fixation probability of duplicated genes should be largely affected by natural selection (14).
I could of course just refer to that offshoot of the Lenski experiment in which the total genome size for E coli increased by over 20% from hundreds of duplications in less than 3000 generations to show what happens when a locus prone to duplication suddenly has a positive fitness effect when amplified.
Incidentally the method used in the paper you link is phylogenetic, that is they’re using common ancestry to calculate a rate of fixed duplicates over deep time by counting differences between species thought to share common ancestry. A method you explicitly reject when applied to the Howe et al diagram to infer the rate at which novel genes have evolved there(whether by duplication or other mechanisms), and you also reject when used to show de novo gene evolution in yeast too. Back to the old double standard again eh Bill?
What does this paper on de novo genes in yeast species say? Hmmm…
Fig. 2Identification of de novo transcripts in S. cerevisiae .
a Pipeline for the identification of de novo transcripts and other conservation classes. For each of the S. cerevisiae transcripts which were expressed above our threshold (>15 TPM), we estimated their phylogenetic conservation using genomic synteny and homology searches. b Identification of genomic synteny blocks by using MUMs. Diagram illustrating maximal unique matching subsequences (MUMs) across a chromosome in different species of the Saccharomyces genus. The synteny blocks were defined by clustering contiguous MUMs in close proximity. c Examples of different classes of transcripts depending on their phylogenetic conservation. Diagram of a hypothetical syntenic genomic region shared by all 11 species with different classes of genes indicate. d Number of transcripts depending on their phylogenetic conservation. The genes were divided in three classes: ‘de novo’ (213 transcripts), ‘genus-specific’ (251 transcripts) and ‘conserved’ (4,409 transcripts). We found that 213 transcripts were likely to have arisen de novo over the past ~20 million years i.e. there were no homologues in species more distant than S. mikatae (purple). Only transcripts expressed at more than 15 TPM were considered here. Below are the number of transcripts identified at each internal branch in the tree leading to S. cerevisiae , before and after applying different computational filters. e The majority of de novo transcripts are not present in the annotations and are expressed in different conditions. Number of transcripts in each class that correspond to annotated transcripts (light grey) and unannotated transcripts (dark grey). Fraction of transcript expression above 15 TPM in rich media (yellow), in oxidative stress (red), or both conditions (blue). The vast majority of transcripts are either expressed in both conditions or in normal conditions. Source data are provided as a Source Data file.
Take a good look at Fig2 D, bottom:
I guess that refutes common ancestry of yeasts then, right Bill? Or do phylogenetic methods only count when you mistakenly think you can spin the numbers to support this unsupported axiom against novel gene evolution that you have?
2 Likes
It’s wrong because it is based on false assumptions.
Why do you think they didn’t do any actual hypothesis testing?
Why did you remove “probably”?
1 Like
Hi Rum
What do you think the waiting time would be for a new 10 residue active site found in 1 or more of the 10k novel proteins in the Howe diagram? As in the Behe/Lynch discussion you can assume that 6 different AA’s will work at every site? My rough calculation is this observation would signal prohibitively long waiting times.
Because it is answering the wrong question. With so many duplications, the probability that all of them are harmful drops to zero (I don’t have numbers to plug in, but that’s the asymptotic trend). It’s not the probability of a beneficial trait occuring, but how many new traits result.
1 Like
Please show your calculations.
2 Likes