Axe’s experiment was different then the one you are describing. My best take from memory is the protein he was modifying was a sub unit of a two subunit enzyme. He was also testing for sufficient enzyme activity for a bacteria to survive penicillin. Protein specificity depends on the application and that is why there is a range in Hunt’s findings. Again if you read Art’s paper he claims that Axe’s results are in the range of other experiments. about 10^10 to 10^60. From Art’s paper.
Also, many thanks are due to Douglas Axe, who graciously helped me with early drafts of this essay. Please note that all of these ideas are mine, and I make no claim that any of these thoughts represent Axe’s views.
It appears Doug and Art had a respectful relationship at least 14 years ago.
Indeed it was. The one Swamidass is describing is far better because it addresses the question far more directly, as do the thousands on catalytic antibodies.
From Axe’s paper: Combined with the estimated prevalence of plausible hydropathic patterns (for any fold) and of relevant folds for particular functions, this implies the overall prevalence of sequences performing a specific function by any domain-sized fold may be as low as 1 in 10^-77.
It’s not even clear that there is a significant difference in the density of function between 100 and 150 aa proteins. And you still can’t extend Axe’s conclusion to all possible functions for all proteins of 150 aa long. And then there is the question of how that protein first arose, which doesn’t have to have been de novo, but could at least in principle have been evolving from another sequence nearby in sequence space, so what matters there is the interconnectedness of different functions in sequence space, not the proportion of the total space taken up by functional proteins.
It is simply not possible to conclude from Axe’s experiment what is being sold to the ID community (that evolving new functional proteins is impossibly unlikely) by Axe, Meyer, Gauger et al.
I am working on another post right now. I think that Axe created an important single data point but it is a single data point. I think Art’s write up at Panda’s thumb was also very useful.
There are other data points that support Axe’s work such as the work gpuccio is doing at UD. Lets see if @Agauger engages and I will see if I can get Giuseppe Puccio to comment.
All you really need to know is that catalytic antibodies exist along with the size of the immune repertoire.
Bill, do you understand that you are tacitly admitting to approaching this in bad faith because you aren’t looking for all of the evidence while cherry-picking what you wish to generalize?
If you’re looking for the truth, YOU will seek out ALL of the evidence, not demand that others look for you. So what are you really looking for?
John, Why would I reference Art Hunts pandas thumb paper if I was doing this? Why would I ask to have Rumrakets papers referenced? Your claim here is without merit.
This is not my argument. My argument is that all the papers are relevant but there is a range of results based the functional requirements. Have you read the Hayashi paper?
In the Hayashi paper a biologically relevant function capable of sustaining phage infectivity was found among 10 random starting sequences. That’s 1 in 10. Your response to this was to blather that the experiment did not find an adaptive peak with the same level of infectivity as exhibited by the g3p domain of the protein found in wild-type phage. Which is irrelevant, what matters is if it is plausible for evolution to occasionally find novel functions. Of course, we’ve been over this a tiresome number of times.
Whats critical is evolution finding the functional sequences we are observing in nature. What we are observing in this experiment is an optimized protein. How did it get there? Also of note is the protein is approximate 400 AA long and only 25% of it was randomized.
Yeah, and there are many candidate explanations for how that could have happened. And no reason to think they couldn’t.
I don’t know what an optimized protein is.
Yeah a particular domain (D2) of the g3p protein required for infectivity was replaced with a random sequence, the result was 7 orders of magnitude lower levels of infectivity. So the random polypeptide used to replace the D2 domain just happened by chance to still be able to carry out the function of the original D2 domain, but at a much lower level of function.
Shouldn’t it be basically impossible for a random sequence to just so happen to be able to perform the same function as the original domain it was used to replace? Weren’t functions supposed to be found at a rate of 1 in 10^77, instead of 1 in 10?
From what I’ve read, the method can screen a library of 10^10 in each round, with successive amplifications and panning for binding at each round. That was from an older review. Maybe it’s up to 10^12. That’s the number I seem to recall from Szostak’s phage display screen for ATP binding some ten years ago.
The review also pointed out the selected “enzymes” were several orders of magnitude weaker than wild-type enzyme. Perhaps the hapten was less than perfect, but I think it is more likely that the selection did not take into account the need for the enzyme to be flexible in order to be able to change conformation during catalysis. This is suggested by the fact that Km was strong, but Kcat was weak. It is not at all clear that such an enzyme could be optimized to wild-type levels.
It also seems to me that the selection process is unlike any evolutionary process. The protein biochemists had to use all sorts of tricks to remove the “duds” and concentrate the pool to contain phage that actually had the sequence displayed. And selection was strong and unidirectional. That is not the case in natural selection. And fixation of binders happened in just a few rounds of selection. That’s not natural either.