It isn't an assumption; the null hypothesis here is that that doesn't happen. Genes coding for cryoprotective proteins have indeed, as you suggest, evolved independently among various species. The resulting genes, despite all producing proteins similar enough to do the necessary job, are "radically different" and "highly diverse." [1]
What's different in this case is that, in three otherwise very distantly related species of fish, we find their antifreeze proteins are coded for by the same genes:
> But, the isolated occurrence of three very similar type II AFPs in three distantly related species (herring, smelt and sea raven) cannot be explained by this mechanism. These globular, lectin-like AFPs have a unique disulfide-bonding pattern, and share up to 85% identity in their amino acid sequences, with regions of even higher identity in their genes. A thorough search of current databases failed to find a homolog in any other species with greater than 40% amino acid sequence identity. [1]
In light of the fact that all other genes known to code for these proteins are very distinct both from this one and from one another, that three species should have a near-identical sequence coding for a near-identical protein suggests rather strongly that this version of the gene arose in one species and was then acquired by the other two, i.e., that horizontal gene transfer has occurred among these vertebrates.
> that three species should have a near-identical sequence coding for a near-identical protein suggests rather strongly that this version of the gene arose in one species and was then acquired by the other two
We'd strongly expect the amino acid sequence to be similar both by "convergent evolution" (each case evolved independently with the same motivation) and "lateral transfer" (one case evolved and then shared DNA across species), so this wouldn't typically distinguish those two cases.
The sibling answer about structure of introns and exons is a more convincing answer, in my opinion. I don't think we would expect to see that in convergent evolution, but we would in a copy-paste job.
On what basis do you hold any such expectation? The paper explicitly contrasts its subject with several examples of convergent evolution producing functionally equivalent, but proteomically and genomically highly distinct, outcomes - which is typical of convergent evolution in general.
That said, I agree that the similarity of adjacent noncoding sequence is also a strong indicator that convergent evolution isn't causative here.
> On what basis do you hold any such expectation?... The paper explicitly contrasts its subject with several examples of convergent evolution producing functionally equivalent, but proteomically and genomically highly distinct, outcomes
On the basis that the protein is the function here. (antifreeze protein). There might only be one good, or best local maximum, solution for this problem at the protein level. So, we would expect natural selection might converge on that one solution. And, the results of two runs would not be nearly as different as they are in cases where natural selection is optimizing for a system process.
Obligatory coding comparison:
If I asked two programmers to code a webshop, I would expect the underlying code to look substantially different - if the code looked the same, I'd take it as evidence of copying.
If I asked two programmers to code "If A then B", I would expect the underlying code to look substantially the same, whether or not they copied.
A specific antifreeze protein is the second case: both the code and the outcome. It's not part of a system which would have more freedom of variation in its solutions.
Preventing crystallization of water is the function. And again, on what basis so presume? Trivial literature review would have sufficed to reveal that there is a whole, mostly very nonhomologous, class of these proteins, not just the one. [1] It is precisely for this reason that near identity observed in the proteins used by these three unrelated fish species is surprising.
As I have already noted this morning, it is at best pointless to attempt to reason out genomics based on first principles drawn from computing. Thank you for taking the time to demonstrate the kind of error that invariably results!
Even with all this 'trivial literature review', there still remains the possibility three fish might have randomly walked [or non-randomly walked] into the same solution with the same local maximum, which couldn't be distinguished from lateral transfer just by looking at the protein structure.
"A doesn't always happen this way" isn't evidence, at all, for B happening. Your logic is faulty.
Thank you for appreciating my sense of humour. As someone who has worked in a genomics lab, I think coding analogies are perfectly fine. The analogy is not in error.
Happily, the paper does not only do that! Too, there are several comments peripheral to this thread which discuss the paper's findings outside the proteome.
Far be it from me to suggest that anyone in a Hacker News thread has failed to do even the most basic of reading in a field outside their own, but I will say that the paper is linked in one of my earlier comments, should you perhaps like to renew your acquaintance with its contents.
Yes, happily! Since, as I was saying in my first comment: I didn't agree with this part of the paper's abstract being relevant evidence, or your take on it; but I agreed with it in other aspects.
Yes, and your disagreement appears to proceed from an attempt to reason purely from first principles, with no sign of apprehending either the clear evidence that convergent evolution on proteins which prevent water from crystallizing into ice in no other case has produced anything like such genomic or proteomic similarity as in the case under discussion, or the infinitesimal probability of that happening by coincidence.
I'm not averse to the idea that I may be wrong on any of those points, but thus far I'm not seeing anything substantive to suspect I am likely to be so. These are just assertions that you're making, and while your reasoning itself is not unsound, the premises from which it follows as yet lack anything resembling substantiation, which is sorely needed given that those premises so contradict all available evidence.
...and, in response to your prior edit, this is coming from someone who has also worked in a genomics lab. Even if I hadn't, what point to claiming authority on that basis?
I apprehended it perfectly well; I'm still in disagreement, since my argument is unaffected.
> so contradict all available evidence
It doesn't, and that's what you have missed. What I said is logically harmonious with all available evidence.
By observing three fish with the same solution for antifreeze, we know that three fish have the same solution for antifreeze. This immediately contradicts any claim that all unrelated species have different solutions for antifreeze, which makes them worthy of study. It's a "black swan".
As such, whatever mechanism has caused this has not been seen to work this way elsewhere. Therefore, saying "this mechanism is not seen to work this way elsewhere" is not remarkable as evidence.
It's now a neutral statement which matches our expectation, and can't therefore be evidence against the mechanism. It's certainly not evidence for another mechanism.
I could just as well say "I have only observed horizontal transfer in N other cases, and this is not one of those N cases, therefore it is not horizontal transfer". That would be wrong, but has equal logical merit as your claim.
All of which still ignores how wildly unlikely it is that such a high degree of similarity occurs by chance.
The paper doesn't claim causality either, but only argues, in my view pretty convincingly, that lateral gene transfer is a likelier explanation for the observed similarity than any other including convergent evolution. You haven't argued otherwise, but only that convergent evolution in this case is not implausible - which is true, but answers no claim that anyone is actually making.
There's no point in that that I can see, so if you want to keep on doing it, I'm afraid you'll need to do so in the absence of an interlocutor, or at least of an interlocutor who is me.
> All of which still ignores how wildly unlikely it is that such a high degree of similarity occurs by chance.
It is wildly unlikely that I should exist through the process of evolution, to waste my afternoon on this argument, and yet: here I am :) Have a nice day.
> Even if I hadn't, what point to claiming authority on that basis?
Oh, this was a direct response to the fact that you repeatedly implied that I was ignorant and hadn't done basic reading in the field. You were wrong about that as well.
Someone disagreeing with you is not always a sign of ignorance.
What's different in this case is that, in three otherwise very distantly related species of fish, we find their antifreeze proteins are coded for by the same genes:
> But, the isolated occurrence of three very similar type II AFPs in three distantly related species (herring, smelt and sea raven) cannot be explained by this mechanism. These globular, lectin-like AFPs have a unique disulfide-bonding pattern, and share up to 85% identity in their amino acid sequences, with regions of even higher identity in their genes. A thorough search of current databases failed to find a homolog in any other species with greater than 40% amino acid sequence identity. [1]
In light of the fact that all other genes known to code for these proteins are very distinct both from this one and from one another, that three species should have a near-identical sequence coding for a near-identical protein suggests rather strongly that this version of the gene arose in one species and was then acquired by the other two, i.e., that horizontal gene transfer has occurred among these vertebrates.
[1] https://pubmed.ncbi.nlm.nih.gov/18612417/