As @BJB noted in this thread, it seems as if the changes that Behe et al. latched onto are actually not found in all, or even most, animals, but rather are variants found in one or a few of the many genomes that were sequenced. I have asked some of the authors of Liu et al. about this, and they tentatively confirm this suggestion. The way to nail it down would be to download and align a few terabytes of sequencing reads from the individual bears, something I have neither the inclination nor the disk space to do at the moment.
I believe this really calls into question the polar bear discussion in Darwin Devolves, since the changes Behe uses to frame his argument are not actually fixed in populations but rather are relatively rare variants that may be positively selected. I haven’t tried to figure out how Behe et al. could spin this.
In this post, no, it doesn’t look like anything new, except an even stronger backpedal on “first rule.” Now they’re saying “many” and “a large proportion,” of adaptive mutations are protein-damaging, which, of course, have little mathematical meaning so those statements are so weak they can’t possibly be wrong.
Also, this line made me laugh out loud: “The concept of getting an adaptive function from genetic breakdown is a counterintuitive thing. How can the loss of a genetic function lead to a new phenotype — a new behavior or appearance or ability — for an organism? That’s the key to understanding Behe’s new book.” First of all, “genetic breakdown?” Oh jeez, are they coining yet another sciency term that has no actual scientific meaning but sounds like it does to the credulous? Secondly, looking past the sciency diction, nothing about that statement is new or surprising to anyone who works in molecular biology or biochemistry. It does prey on misconceptions about evolution proceeding through continual perfection. It tells us this idea is counterintuitive, which is a rhetorical trick to convince unquestioning readers, but it isn’t at all.
Anyway, I read it quickly and that’s all the time I can give this right now - busy time of the year. Of course I understand why they’re trying so hard to recover from this, but just because they keep repeating points we’ve already dismantled, doesn’t mean that we are obligated to continue engaging. For them, this is their day job, they are paid to try to prop up the Behe book and ID by extension. (Did you catch the use of “we” when discussing the choice of polar bear in the Behe book? Was DD actually a group effort?) But for Art, Josh, and me, we have day jobs that we took time away from to point out the bad science in DD. We did it as a public service, but we can’t keep going back and forth, mostly repeating ourselves, without neglecting the work we are actually paid to do, which is to say teaching and doing research.
As Art mentioned above, there may be even more fundamental problems with Behe’s use of the polar bear APOB variants, but that’s not something even a thorough reader of the Liu article could have known, so I’m not counting it as a error as much as bad luck on Behe’s part. Others may argue that he failed to do due diligence, but as an author of a book that, like Behe, tackles a broad range of topics and examples, I won’t fault him for not going above and beyond what Liu did in terms of looking up the allele frequencies of all the variants. The point is that his argument is on even weaker ground than we’ve already shown. (Some, most, or even all of these variants may be rare anyway.)
Also, we’ve been in contact with the authors of the study and they had a hard time even understanding Behe’s arguments and how their work is being used. They see little value in bothering to respond to him and I’m starting to understand the wisdom of their position.
Yep. That fact has been known for decades from work in bacterial genetics. Function knock-out based adaptions are among the most common found in a population. They tend to arise on the order of 1E-6 or less.
I agree this isn’t something to fault Behe on, however…
That is actually the point. We won’t fault him for missing this, but it severely undermines what is left of his argument.
Very true. I think, nonetheless, @art, @nlents and myself (@swamidass) should do a final article that ties up the loose ends. The fact that ENV is getting repetitive tells us they don’t have much more to add.
Yes, I’m fine to do one more post, summarizing the issues with his writings about polar bear evolution in DD and chloroquine resistance in EoE. Then I really am done with Behe and the DI, at least for the foreseeable future. I have a new book to get started on and new research projects that are at a critical point and really need my full attention.
Im at a loss why so much time is being spent, by @Agauger et al., to describe that evolution frequently involves a loss of function which had emerged EARLIER, during an earlier phase experienced by a living population, which helped fix the function in question to begin with!
The fact that fishlike tetrapods arrived on land WITH GILLS AND FINS should have been enough to convince anyone! Why? Because amphibians, reptiles and mammals no longer have gills and fins!
Fins can only become limbs if the fin expression is corrupted.
And gills had to go… if only to keep from losing moisture or to prevent blood networks to the gills from being compromised.
This is obvious stuff.
The weird twist is Behe implying that when God is helping evolution along through design … that loss of function will NOT (necessarily?) occur!
Why weird? Does the fact Adam does NOT have gill slits mean that even God’s design sometimes leads to loss of function?
Or does it mean that it would have come out differently if God had been working that particular case?
Actually amphibians do in their first phase (tadpoles), and we all do in embryonic phases. But yes, I agree with your point and I do think it can be applied to biochemistry as well as anatomy. The reason why Behe’s “first rule” hasn’t impressed any scientists is because insofar as it is a “rule,” it’s false, and insofar as it describes something that happens sometimes, it’s obvious and long-known.
Here’s a parallel. I could come along and declare “Lents’s first rule of gene expression: the 25th triplet after a start codon will code for the 25th amino acid of the protein.” This statement is false in many instances, but when it’s true, it’s completely boring and pointless as a “rule.” I’ll admit this example is way more pedantic than Behe’s first rule, but the logic is parallel. It only describes how nature behaves some of the time and in ways in which we have long known. This is why I read quickly through the Polar Bear example in my first reading of DD. I assumed Behe at least had that example right because examples that follow his rule are not hard to find. It turns out, he picked a bad example, but even if he hadn’t, it wouldn’t matter. I could trot out dozens of examples of proteins that obey Lents’s first rule, but that wouldn’t change the wrongness of the rule in other cases nor does it somehow make the rule novel. Behe’s first rule has the odd distinction of being both wrong and long known. Or to say it Pauli-style, "Not only is it not right, it’s not even wrong. (Actually, the Pauli principle applies better to ID as a theory, because, since it’s not testable, it therefore can’t be proven wrong.)
A few comments here. Some have made the wild claim that PolyPhen-2 does not measure the likelihood of biochemical disruption but rather it measures adaptive function. Wow. As the PolyPhen-2 literature shows, this is completely wrong: PolyPhen-2 is designed to detect disruption and does a pretty good job of it. In a study where it was tested against positive and negative controls, PolyPhen-2 had an accuracy of .72, a sensitivity of .8 and a specificity of .7.
The one thing PolyPhen-2 cannot do is predict which amino acid changes might be adaptive. At the protein level, the disruption is going to be … disrupting. Damaging, by all the criteria PolyPhen-2 measures. The program is cueing off of disease-causing mutations, not beneficial ones. It looks for biochemical and structural problems . Its function is to identify SNPs that are likely to have a damaging effect on structure or chemistry or both.
I haven’t seen it before, but it may be that I simply didn’t notice it before.
She just doesn’t understand the literature on PolyPhen-2. Honestly, seems like it is acting, because they do not really have a scientifically grounded response. PolyPhen does not predict biochemical damage, period. It is trained on (ironically for Behe) phenotype, and it is merely a quirk that is says “damaging.”
We might even contact the authors and ask them to change it to reduce confusion. If they do make that change, what happens to the ID argument? Their argument is based merely on quote mining that word, so the ID argument would just evaporate. What would they do then?
Driving home the point against them, @Agauger cites a study that undermines her own point:
The validation study is looking not at biochemical damage, but exclusively negative phenotypic affect in humans. It is a profound misunderstanding of biology to think that this is a predictor of “biochemical damage” in selected genes. It is so wrong, it is hard to know how to respond more clearly than we have already.
Well, I think it’s new in the sense that she’s claiming that critics have said something (that PolyPhen “measures adaptive function”) that I don’t think any of us have actually claimed. But others may have.
I tend to take a more practical approach to this question. Let’s say we used various ape genomes to construct an ancestral genome for the common ancestor of humans and chimps. From this, we could find all of the potential non-synonymous mutations that occurred in the human lineage. What would happen if we put those mutations into PolyPhen? The algorithm would label them as benign or damaging. None of them would be considered beneficial simply because that is not an option in the algorithm. That doesn’t make much sense. Either modern humans are “damaged” apes, or there really are beneficial mutations that resulted in modern humans. I lean towards the latter.
I think it is also worth mentioning that even an intelligent designer would not be able to change a genome in a way that PolyPhen would consider beneficial.
Sorry it has taken me so long, but this illustration really helped.
That’s an interesting (and slightly comedic) point!
It is such an obviously broken analysis they are making I am genuinely surprised they are continuing to press it. There is some good news in this as it makes it very obvious to students that they are wrong, stubbornly so.
From the abstract of the article that @Agauger cites:
automated prediction of whole-genome/exome non-synonymous variant functional consequences.
The article itself is not talking about “biochemical degradation” or “biochemical damage”, but ability to predict negative phenotype (not biochemical!). This blatantly contradicts their understanding of PolyPhen 2, and it is fairly striking that her own citation shows their understanding is wrong.
I have to wonder if they know they are painted into a corner here, and just do not know the way out.
The same could be said of IC and that’s the problem with their whole approach. If you are starting from a conclusion that you feel you must support (rather than letting the data lead you to a conclusion), it’s basically impossible to interpret data in an unbiased way.
No, not really discussing. I don’t think Bilbo wants to do that. Really interesting thread, though, thanks.
@NLENTS I was struck by the addendum to your post a few months ago, in which you show cherry picking of the supplemental data in the Cell paper. Has any one responded to this? I ask because if this “discussion” is about frantic efforts on the part of the DI to clear a train wreck, it’s less interesting to me. My impression was that Behe was taking a position on an open question, admittedly a position that favors his thesis, but still a position that’s defensible in principle. Maybe I was wrong about that?