Your assertion is insufficient.
How did you calculate that probability without knowing what the P. vivax chloroquine resistance route was?
You appear to be pulling numbers out of thin air.
2 Likes
The interesting thing is that, almost immediately prior to that sentence, @Giltil wrote something that would seem to indicate he understands this:
No, not very hard at all.
3 Likes
We know that for P falciparum, the per-parasite occurrence of de novo resistance is 1 in 10^20. From this, we can deduce that no pathway to resistance exists that is more probable than that. So if the P vivax pathway to resistance is possible for P falciparum, then it is at least as improbable as 1 in 10^20, and most probably more improbable since it has never been observed.
No. See above
Hi Rum
Do you have any examples of unconstrained adaptions you can point to?
Hi Bill.
Do you have any example of me saying any adaptation is unconstrained you can point to?
4 Likes
So what do we call it if they stay similar to each other but are different in identical ways from their ancestors?
And within a breeding population, is every mutation that happens more than once to be considered convergent? What if one site converges but other sites diverge? How do we count the genotypes? If we’re talking about individual sites, there’s a simpler term: homoplasy.
Wrong. All we know is that is a very rough guesstimate made by one author in one paper of how often this resistance has arisen in human subjects being treated for malaria. In another paper the same author uses the number 1 in 10 19. Maybe rats are teeming with resistant parasites that we know nothing about.
Now, this raises an interesting question. Your reasoning here seems more or less sound, and as it happens CQR does exist for vivax, but is much less common than for falciparum.
Here is the question: If Behe had written his book about CQR in vivax rather than in falciparum, would he not have defined this as the “edge of evolution”? I can’t see why not. But then, following his reasoning, CQR should not be more common in falciparum, because this would require that it occur beyond Behe’s “edge.”
And yet it is.
Can you explain this finding that should be impossible if Behe’s reasoning is correct?
1 Like
The example Behe gave you describe it is constrained as it takes a long time to reach an adaption. What is causing the constraint and would this condition be different in other bacteria. @Giltil any thoughts?
Plagiarism.
2 Likes
No, we don’t. That’s a very rough estimate based on the number of observed incidences of apparently unrelated resistance to chloroquine.[1] The actual occurrence rate could be much higher, with the trait being lost due to it being selected against in the absence of chloroquine; it could be much higher because the occurrence of different chloroquine resistance mechanisms in already-resistant organisms would be masked; it could be much lower, if there were fewer incidences[2] or some of those incidences were actually related; and it could be either higher or lower simply because that’s an extremely low sample size for a very rare event. Note also that this estimate was made before we knew much about the genetic basis for chloroquine resistance, so reflects only the outcome, and not the prior probability, which could be very different. [3]
You might, but anyone who understands probability would immediately recognise this as fallacious, since something may be more probable but still occur less often, especially for rare events in a single trial.[4] Deriving a priori probabilities from actual outcomes is unreliable.
Never having been observed is not the same as never having happened. It could have happened many times in P. falciparum that wasn’t subjected to chloroquine. It could have happened many times in P. falciparum that was already resistant to chloroquine via a different route. It could have happened many times in P. falciparum but been lost due to being outcompeted by P. falciparum that was resistant via the route discovered. It could be more common but less effective, more likely but have occurred less in practice, or more common and more likely but with side-effects that lead to it being outcompeted.
Yes. There are so many unwarranted assumptions and extrapolations in your reasoning that your conclusion is unjustified.
“Resistance to chloroquine in P. falciparum has arisen spontaneously less[5] than ten times in the past fifty years (14). This suggests that the per-parasite probability of developing resistance de novo is on the order of 1 in 10^20 parasite multiplications.” (source) ↩︎
(14) from the above describes only two original outbreaks, which is a long way from “less than ten”. (source) ↩︎
If I roll four dice and get three 2s and one 5, that doesn’t mean the probability of rolling a 2 is 0.75. You need far more trials to get a decent estimate of these probabilities, and the same is true of de novo chloroquine estimates, for which we will only ever have a single trial, and possibly only ever two incidences. ↩︎
If I roll two dice 36 times and get a total of 10 only once, that doesn’t mean there is no way of rolling 10 that is more probable than 1/36. But that’s the form of argument you are using. ↩︎
[pedant] Fewer! [/pedant] ↩︎
3 Likes
I would say that any improbable adaption is constrained, by definition. The more improbable, the more constrained. Now, as to what is causing the constraint, I think it has to do with the fact biological systems are highly complex and specified. Without intelligence, such systems can change only marginally.
This is incoherent even by Bill’s standards. The only thing I could discern is that he thinks malaria is caused by bacteria.
2 Likes
Are you sure that CQR for vivax is much less common than for falciparum ? To claim this, you have to know what is the per-parasite occurrence of CQR in vivax. It is more or less 1 in 10^20 for falciparum. What is it for vivax?
Since it has come up a couple of times now, it seems worth pointing out that P. falciparum does not infect rats. While there can be some spillover to other apes, it is largely a human parasite. Same for P. vivax. There are Plasmodium species that infect rodents, but they don’t infect people.
(While I’m here, an update on my attempts to probe the behavior of AminoGraph. I thought I’d try some small toy alignments to see if I could get an intuition for the kinds of sequence patterns it determines require a DAG vs those it considers a tree adequate for. But even a 10 sequence, 10 residue alignment takes roughly 24 hours to analyze, and that’s pegging 4 cores* the whole time. I’m not interested enough to wait weeks for a batch of results when it might require iterating on several such batches to get answers. *Obviously YMMV if you’ve got more or faster compute resources to throw at it.)
In the scenario you describe they have independently evolved in identical ways, so apparently they are staying equally distant to each other(even if that distance is zero) while putting more distance to their ancestor. If they really are evolving independently, yet are not diverging, it’s parallel evolution. I haven’t stated anywhere that we should not employ the term parallel evolution.
Possibly, yes.
You can count them on a per locus fashion (constrained to a coding region, intron, or whatever you want), or you can count them genome-wide for a net total. Depends on what you’re doing.
Some genes or loci can converge, while others diverge, and you can count those individually when and if that is useful (if you want to analyze and understand why that locus in particular converged), and genome-wide in other situations. People already do that. Parts of the prestin gene seems to have converged on similar amino acids, while the overall genome diverged in net total, between bats and echolocating whales, with implications for phylogenetics, functionality, etc.
And I would consider homoplasy a type of convergence.
That sounds absurd. How in the world can it take that long? It must be extremely badly optimized. Inferring trees for thousands-basepair loci with dozens of species and 100 replicate bootstraps is considerably faster than that, on my computer, and tree space is vast.
Ah, so it would seem that what Ewert shows in his figure is in fact his full alignment. Perhaps it’s just an assembly of all the informative sites, not an alignment of any section of the protein.
Make up your mind.
In any event, since CQR is less likely to be encountered in vivax than in falciparum when treating malaria, and vivax is the most widespread species of Plasmodium, it logically follows that CQR is likely less probable to rise in vivax than it is in falciparum. That is, if we accept the metric that Behe relied on to determine his 10-20 figure.
So, with that settled, will you answer my question? If CQR in malaria determines the “edge of evolution”, why will that “edge” differ depending on the species of malaria parasite? For that matter, why rely on this specific trait in this specific species at all? Why not, instead, define the “edge” by the evolution of the ability to anaerobically metabolize glucose in E. coli? According to some of Behe’s colleagues, this can evolve in as few as 100 generations.
Please explain.
Thanks for the correction.
That IS absurd. Could Ewert have tuned his code to solve one specific optimization at the expense of general utility?
Could be a side-effect of using Rust to implement it.
1 Like