A prominent ID supporter at UD, gpuccio, has this to say:
My simple point is: reasoning in terms of design, intention and plans is a true science promoter which can help give new perspective to our approach to biology. Questions simply change. The question is no more:
how did this sequence evolve by some non existent neo darwinian mechanism giving reproductive advantage?
but rather:
why was this functional information introduced at this stage? what is the plan? what functions (even completely unrelated to sheer survival and reproduction) are being engineered here?
Gpuccio references actual biology in his writings and is one of the few at UD that do, and as such I’m prepared to take him at his word that the ID project is now ready to move from simply determining design to answering the questions he posed:
- why was this functional information introduced at this stage?
- what is the plan?
- what functions (even completely unrelated to sheer survival and reproduction) are being engineered here?
If any ID supporter would like to provide a specific example with answers for those 3 points for discussion that would be perfect.
Gpuccio’s OP concludes:
The transition to vertebrates was a highly engineered process. The necessary functional information was added by design.
In response I simply repeat back the question what is the plan?
That’s not a story, that’s a fact. They ARE similar. Even within the flagellum, some of the proteins are similar to other flagellum proteins.
Yeah pretty much.
Why does it have to pass through the entire space? If protein A, is highly similar to protein B, surely it doesn’t have to pass through every possible permutation to get from A to B. Right?
Maybe there IS a selective advantage to a reduced structure? Have you thought of that? Have you thought about what a precursor structure to a flagellum could be? Why does it have to go from completely nonfunctional to a complete modern flagellum, why can’t there be functional ancestral stages that are simpler?
I dare say you are working very hard to avoid thinking about it.
Imagine, in a primitive system, only 2 available amino acids, 1 polar 1 hydrophobic. The space for 20 amino acids in strings of 100 is still 100^20, but we cannot explore most of it, since we don’t even have 18 of the letters. So, is the absolute size or density of a 20-acid space relevant here? Of course not, we don’t even have 20 acids to play with. But maybe, one day, we will …
So imagine adding one amino acid to the library. We can now explore a fair chunk more of the 100^20 space. Do we expect our chance of finding a functional peptide from where we are to go down? Why would it? We have more opportunity for subtle variation, and configurations not available to 2. And, we are starting from functional peptides anyway. If we hit a detrimentally nonfunctional 3-acid peptide, the lineage simply dies out, leaving the world to the ones that don’t. 2-acid space would have to be completely surrounded by nonfunctional 3-acid space for this wider region not to be explorable. [eta – and if that were the case, the third acid would not be added].
Keep adding 1 acid to the library until you get 20, each time you further expand the accessible portion of the wider space. But, all of a sudden, when you get to 20, the chances of evolutionary exploration have become an astronomical squiddillions to one against … or have they?
Let’s examine A and B. B is identical to A apart from B has some random bit sticking out due to a mutation. Under certain circumstances B will flex the random bit sticking out just as part of living. Flexing that bit moves B. Only a tiny tiny bit. The width of a molecule, no more.
But B is now infinitely more motile then A, despite the fact it’s just a random protuberance that only occasionally provides movement as an indirect by-product of some other proces. B’s descendants also have the bit sticking out and A’s do not.
Now, there is a food source. It’s out of reach of A but B happens to reach it as it now can ‘move’ about. It’s random and sporadic, but it’s movement.
Is there a selective advantage for B over A? But it’s just a rudder/oar rather then a motor and it’s not under direct control.
Sure, this is just what you’d call a ‘just-so’ story but you need to understand that half an eye is infinitely better than no eye at all! A light sensing patch with no directional ability is infinitely better then no eye at all! A bit that sticks out and twitches occasionally is infinitely better then no ‘motor’ at all.
This is all far better explained in the resources you’ve already been given so I don’t know why I’m bothering tbh.
colewd,
At this point your agenda is showing. You might want to pull your trousers up.
The questions you are ignoring are more telling than the answers you are writing. You have an opportunity to learn from some people who actually know what they are talking about but you insist on things that are simply not true being true and resist correction. Soon nobody will bother with you as what’s the point?
Maybe you should try to think about what you wrote and what I wrote in response. I was using the reference to support the general picture of protein evolution as being actually much more tolerant to mutations than what you seem to think.
The question this reference should lead you to ask is: If protein evolution is really so constrained as I think, how come it was possible to evolve a population of E coli to replace all of the roughly 21.000 tryptophan codons in it’s genome into a different amino acid?
I’m not sure you understand the weight of this evidence or what it even means.
Every protein in the E coli genome that uses a tryptophan residue somwhere, have had that amino acid replaced with a different but chemically related one. Are we now to believe that every single one of those ~21.000 tryptophan residues had unimportant functions as part of the proteins they make up?
Actually there is no such range. The real number is probably even smaller than 1 in every 10^11 proteins, since only one function was tested for. But 1 in 10^11 will suffice here.
The numbers you cite (such as 1 in 10^64) are either from studies that address a different question (how many sequences correspond to a certain fold), or misleading creationist experiments (Axe et al 2004 and others) that draw conclusions not supported by their data (and I have shown this before, to you, on this site and on Sandwalk). I want you to stop using the 1 in 10^64 number, since it is demontrably wrong. That number is NOT an estimate of the density of all possible functions in protein sequence space. Don’t ever reference it again, you will be referencing a proven falsehood.
Prove it. Don’t just mindlessly assert it, prove it.
Why do you believe that? Bring evidence to support that belief.
Your opinion is simultaneously anything but humble, and without merit. Stop merely stating your opinion, it has no effect, start referencing evidence to support your beliefs.
Those are various estimates of the density of function in protein sequence space and the density of sequences that correspond to some specific functional fold. They’re two different numbers. The relevant one is the first one (the density of function), the second one is an estimate of particular fold-adopting sequences.
The first number is 1 in every 10^11 randomly generated protein sequences 80 amino acids in length, will have [The Function In Question]. It’s based on this paper: Functional proteins from a
random-sequence library by Keefe & Szostak.
The second number is a creationist number, IIRC spuriously calculated by Douglas Axe back in some work he did in 2004. Can’t be bothered to dig it up now, but if I’m not totally demented yet, I think what he did was take some versions of extant Beta-Lactamase enzymes, deliberately insert mutations into them and then see how many it took to “break” the particular enzymatic function of the protein into undetectable levels. He then does something like calculate the number of functional Beta-lactamases folds out of the total space of possible sequences and then conclude that must be the density of all functional proteins in all of protein sequence space.
If the problems with this method isn’t immediately obvious to the reader, then they’re probably too uneducated to be bothering discussing this subject in the first place. Quick hint to readers: The method doen’t probe the density of all possible functions, it merely estimates the density of sequences that correspond to the fold.
Actually I bothered to look it up on Panda’s Thumb: Axe (2004) and the evolution of enzyme function.
Axe is wrong by approximately FIFTY ORDERS OF MAGNITUDE. Creationists never bothered to understand what Axe’s work was about and are now citing his work as if it addresses a different question than what it actually does. Axe is of course all too happy with creationists operating on this misconception and have been working tirelessly with Ann Gauger to artificially inflate the number further. Last I checked he’s now claiming the density of functional proteins is somewhere around 1 in 10^84 sequences (since, apparently, 1 in 10^64 or 10^77 isn’t big enough).
Rumraket:
Don’t be so picky, Rumraket. That’s a mere 0.77 dembskis of error according to Wes Elsberry’s calculator.
Well within standard ID tolerances.
Another thing here that comes to mind, there is such a thing as detection limits. Testing for enzyme function for example will often involve an assay that has to be read by a machine that returns an aborbance at some wavelength. Weak enzyme catalytic rates might have lower substrate turnover than can normally be detected after a few minutes of incubation, yet still be visible to selection in very large populations or over longer timescales.
In the Keefe&Szostak experiment, the detection limit is the column elution efficiency. There’s probably even more ATP binding proteins in their random library sample that are washed away because the binding was too weak to withstand washing, but might still be strong enough in a biological context to be visible to selection.
But really, the main issue is alternative functions besides the one tested for. Only one was tested for, it is entirely possible and indeed likely that many other different functions exist within that initial sample. So the 1 in 10^11 frequency can only be a minimum, there has to be more.
Allan Miller,
Like that!
I have a large unruly tool box. If I need to fix something in a hurry, I grab the first tool I come across that might do the job. I balance the time taken to find exactly the right tool against the time taken to complete the job less efficiently.
@ colewd. This is the crux of arguments over functionality. Gpuccio talks about superfamilies of related functional proteins as if the distribution of functional proteins varies between proteins that are known and putative proteins that have not been already exploited by some organism. Why should the density vary across sequence space? Why should not the occurrence of functional proteins turn out to be near constant across the whole space of all possible protein sequences?
So, in the interest of making progress in the discussion, do you accept the evidence provided as answering your question regarding a significant subset of de novo genes?
Excellent! In what way does his explanation of the connectedness of functional space not address your “needle in a haystack” issue?
Rumraket,
Can you explain why his work is 50 orders of magnitude off? Art Hunts paper did not make this claim.
Alan Fox,
I will ask gpuccio at UD and get his opinion.
He doesn’t explicitly state it, but if you read it, it is actually in Hunts essay. Notice there are two different concepts talked about:
A. The number of sequences of a specific length, that fold into the same fold (The Beta-Lactamase fold), and retains the known function of that fold (Beta-lactamase enzymatic activity), out of the total number of sequences of that length.
B. The total number of functional sequences (whatever fold or function), of a certain length, out of the total number of sequences of that length.
Those are two very very different concepts, and an estimation of one has almost nothing to do with an estimation of the other.
Axe et al 2004 attempts to address A (the one about number of functional folds corresponding to a specific fold), yet the number that matters for evolution’s ability to “create new genes with new functions” is B.
Creationists are citing Axe as if he addressed B, and using the number he came up with for B, as if it is the number for A. Axe doesn’t correct them. You have been using his number before many times, as if it was an estimation of B.
The number for A is off the number for the concept of B by approximately fifty orders of magnitude (10^63 – 10^11) = 10^52.
Ian Musgrave states as much in one of the comments, regarding possible alternative functions:
Hunt himself also mentions it, yet in passing:
So, interestingly, Hunt even knows of functionally related proteins that were never tested for in Axe’s work. His mutationally inactivated Beta-Lactamases could be functional DD-peptidases, or even something else.
Rumraket,
Thanks. Your position is now clear.
Do you think evolution only refers to case B?
In the context of our discussion, the important number is B. After all, we want to know whether it is plausible for evolution to hit upon a novel function for a new gene by an accumulation of mutations within a viable timeframe. To do that, we need some rough idea about B.
If we were concerned about how tolerant some particular fold was to change, we’d be trying to estimate A. A might be an relevant number for evolutions capacity to retain a specific function in a broad number of environmental circumstances. If A is too small, it might be the case that some enzymes can’t be adapted to, for example, high temperature environments.
It is entirely possible that there exist enzymes that could never be made to function inside bacteria living at a hydrothermal vent, for example. Or at least, that evolution couldn’t “find” the solution even if it existed from another place in sequence space even with the same overallf old, because only an exceedlingly small number of particular sequences that fold into the right structure would retain function at (say) 110 degrees C.
So it’s not that I’m saying A is never a relevant or interesting number, it just isn’t for the question we have been discussing.
Rumraket,
So A is a protein having to bind to something specific.
B is a protein just having to bind to something.
What about a protein needing to bing to multiple entities? Do you think this would make protein folding rarer then the above case if:
A has to bind to multiple specific proteins.
B is able to bind to several different proteins i.e. multiple active sites.
Alan Fox,
Here is Gpuccio’s response.
Please extend an invitation for him to join us here, along with anyone else from UD who would like to participate in the conversation. This is an open forum where no one will be arbitrarily banned nor will anyone’s comments be deleted.
What do the pros here think about gpuccio’s metric for “functional information?
I don’t understand what you’re trying to say there. What are you asking?
The posterior probability fallacy is the quintessential fallacy of creationism.
Moving from any point A to any point B, if crossing the same total distance, is equally improbable. Yet by random accumulation of change, a point B will be reached, and that point B would have been just as improbable as any other point B. Yet it happened anyway.
It’s like rolling 100 dice and then calculating the odds of that particular arrangement coming up. It will be 1 in 6^100. So an event with a probabiliy of 6^-100 just happened. So every roll of 100 dice is a miracle, or what? There really is no fucking point doing those kinds of post-fact probability calculations, they will all appear to be practically impossible.
Will creationists ever stop committing this idiotic fallacy?
gpuccio@UD
It is clearly an observed fact that many folds are not obviously related to other folds within extant life. This, however, does not prove them isolated islands of function in itself. Because if evolution and common descent were true, one would expect loss of relationship information over time. In an evolutionary scenario, many configurations are passed through then lost. It is wrong – the same mistake as Durston’s – to regard extant life as a representative sample of all sequence space. It is also wrong to regard the lack of detectable ‘lateral’ connection to be conclusive of a lack of coalescent connection when adding the dimension of lineage. It seems to be a molecular version of the ‘dogs giving birth to cats’ trope – you can’t get to modern sequence A from modern sequence B.
Things seem to be processing much as expected:
gpuccio
Somehow I doubt that if gpuccio was asked to peer review a similar paper he’d respond with “it’s all lies!”.
gpuccio, can you be specific? Who is lying and about what?
They seem to be swarming over Keefe & Szostak (2001!!! Ancient!!! points out Dionisio) as if it were the only paper ever to look at random peptide libraries. And claim that they used ‘intelligent selection’. So, are we sayng that the only legitimate way to experimentally examine function and selection in random libraries is to perform no assay for it?
Perhaps they could suggest an improved experimental protocol that would address their own objection to the one presented?
gpuccio begins by lamenting the lack of input from objectors. Then, some time later, notes with approval that this thread is longer than that one (even more so when you take out the stuff about cats!). Attention! Even attention from liars. It’s still attention.
Unless someone quotes him, I don’t read anything he writes, because I don’t read UD. He’s welcome to drop by if he wants.
I’d be interested in hearing about the lies about Keefe and Szostak 2001 too. Perhaps we could discuss Hayashi et al 2006?
Experimental rugged fitness landscape in protein sequence space.
Rumraket,
I think they’d play the ‘intelligent selection’ card again. Although it would be missing the point by some way. Selection, intelligent or otherwise, is not going to help navigation between isolated islands of function, if that is how protein space is structured.
Of course they would, it’s emergency fallback rhethoric. As if humans doing the selecting makes functional variants miraculously appear in a sequence space creationists insist is largely empty of function.
Anyone remember Kirk Durston saying it wasn’t somehow enough to bind ATP (as Keefe&Szostak showed)?
Well, it seems somebody went ahead and took the experiment a step further:
ATP selection in a random peptide library consisting of prebiotic amino acids.
I’m sorry for all these literature dumps, but really the most effective way to show that the creationist assumptions about protein sequence space are false, is with concrete empirical experiments.
So the peptide, made of a reduced alphabet of amino acids more similar to what would have been available if made from some nonbiological process, could not only bind ATP, but actually use it. Two very important functions, in the same sequence, near each other in protein sequence space. What are the odds? Is protein sequence space now a miracle?
gpuccio responds:
Poor gpuccio.
He’s fighting a battle against reality, and reality is winning.
At least he’s whining, though.
Glen Davidson
But given how big the entire space is finding any at all with even weak bindings to ATP would be unexpected according to ID and it’s islands of functionality idea, right gpuccio?
What proportion of total protein configuration space was represented by that ‘big enough random library’ used in the paper gpuccio?
No need to apologise. I think “islands of function” has quite a grip on the collective ID consciousness. That there is plenty of evidence to show it’s an unjustified claim is a most valuable point and worth highlighting.
Why does he talk about lies but not show anyone lying? What has been claimed about what the paper means and how is it a lie?
Yes, in a library of approximately 10^11, there is at least one with a detectable level of binding affinity, for some arbitrarily picked function.
Shouldn’t that be exceedingly improbable if ID-creationism was true, to generate a library of random sequences and then by sheer chance, have the function you are looking for turn up in that library already to begin with? Shouldn’t it be more like 1 in 10^77 as Douglas Axe claims?
What the hell are you talking about “non-naturally selectable” and “like it’s precursors in the original library” ? Is he trying to say that ATP binding is not a biologically relevant function?
Hey Gpuccio, how does ATP synthase function in reverse?
This is nothing but empty rethoric. Because humans set up the experiment, call it engineering and pretend it supports design.
The key points are:
1. There is biologocially relevant function in the original pool of random peptides.
2. They can be randomly mutated, and among those mutants are functional versions with even better function, clearly visible to a selection process.
Having a human person stand next to the test tube isn’t going to make function magically more prevalent in amino acid sequence space. There does not emanate magical function-creating beams from human bodies into proteins such that they can suddenly do things they are not supposed to.
If Gpuccio wants to try to dismiss it on the basis of ATP binding not being “naturally selectable”, he can do a simple literature search for countless other similar experiments that find functional variants among pools of random polymers for many other functions than ATP binding at the same rough frequency. Phage infectivity, for example. Or how about ATP binding and hydrolysis in the same molecule? (So it can bind and USE ATP). There are many other experiments that also probe random libraries for arbitrarily picked enzymatic functions, and find them.
These results should not be possible of amino acid sequence space is mostly a nonfunctional desert that random mutations could never succeed at finding function in. But these results are routinely achieved, so now mr. Gpuccio is reduced to trying to dismiss the result because of some strange idea about whether ATP binding is biologically selectable. Even if it isn’t (for the sake of argument), why does the function even exist at such an easily found frequency?
Almost everything is deleterious in the right environment. Having a thick coat of thermoprotective fur on the north pole may be highly advantageous on the north pole, but pretty shit if you’re living in the Sahara desert. Function is context-specific both at the micro and macroscopic level, so that kind of mindless dismissal cannot be a successful argument against protein sequence space being mutation+selectively navigational.
What the fuck relevance does folding even have here? There are functional unfolding proteins known in life. Some of them sit in the ribosome, stabilizing the surrounding structures.
That’s just more empty red herrings by Gpuccio.
Except that it DOES tell us something about the occurrence of naturally selectable protein sequence in random libraries, because the function “BIND ATP” is an important biological function. It is one among a host of important functions for thousands of enzymes in life. The ATP synthase binds and hydrolyses ATP to ADP when it is running in reverse (and pumping protons back out across the membrane).
ATP binding can be part of allosteric regulation mechanisms causing conformational change in enzymes or even regulatory elements that bind DNA, reducing their activity or inhibiting binding.
I’m afraid mr. Gpuccio just doesn’t know what he’s talking about, is looking for excuses to dismiss the research being referenced, so he should crack open a biochemistry textbook at some point.
Now, those lies again. Where are they?
Apologies to gpuccio for not having responded to this earlier.
He asks:
Were there no evidence regarding the numbers of possibly biologically active proteins that might exist among all theoretically possible (I may have misused “putative” in this context) sequences, then the only reasonable thing to say is we would have no idea of the density of occurrence of such proteins. You could claim they are extremely rare and I could claim they were extremely common and we’d have to agree to disagree.
But Keefe and Szostak showed ATP binding is common and Rumraket above points out other research that supports the idea that functional proteins are not islands in a vast sea of non-functionality.
I wonder what gpuccio thinks actually happened to vertebrates. He accepts common descent, but thinks there’s some impossibility in the transition. Since he continually mocks the existence of intermediates, does he think God quantum interfaced to inject all the “6000 bits” of information to some tunicate or lancelet looking organism (well, at least a couple of them, male and female) and let natural selection do it’s thing? Or maybe pre-vertebrates were suddenly giving birth to all sort of different vertebrates?
Patrick,
Hi Patrick
He is managing his very active op right now which is consuming his spare time. Hopefully he can join us sometime in the future. If you have commented with a minority view on a blog you know how time consuming it is.
I think so, yes:
gpuccio
And also:
I think it’s fairly wrong to include those others as an attempt to legitimise those views gpuccio. I know Penrose thinks quantum effects are part of consciousness but I don’t recall him saying that’s how the Intelligent Designer did it!
Huh? His is the majority view at UD and they were complaining that there were few arguments from the opposing side. So how does any of that make sense?
Rumraket,
What is the chance of the same number coming up 3 times in a row?
What is the chance that the result is one of 10 pre selected numbers?
Your point is valid for some of biology but is it for all we are observing?
Especially when you lack a case for your claims.
Glen Davidson
OMagain,
Yeah, I know he mentioned “quantum interfacing”, but I’m curious about the biological details of the effects of such QI on vertebrates and life in general: if RV+NS can’t explain the transition, how did the designer introduce the new “information” in the vertebrates lineage? Apparently the only logical conclusion from his premises is that pre-vertebrates were giving birth to full blown vertebrates
Much, much lower than 1 in 6^100, which is already an absurdly low chance. It’s 6^-100 x 6^-100 x 6^-100 = 6^-300
10 in 6^100
Which is still an absurdly low number, so low that intuitively it seems impossible.
Yes.
Again, you specify a sequence of 100 numbers from 1 to 6, then you calculate that the odds of rolling that sequence on 100 dice is 6^-100. An absurdly low number.
You roll the dice and of course a different sequence comes up. Then you ask yourself; But what would have been the odds of that sequence before I rolled the dice?
Well, it would also have been 1 in 6^100.
So an event with a probability of 1 in 6^100 just happened. So this immediately raises the question; What the hell use is it to calculate these kinds of conditional probabilities when all the events that can happen are equiprobable and happen anyway? They don’t tell us anything about what will happen or should happen or can happen. ANYTHING will look fantastically improbable in the same way, if we try to calculate the odds before it happens.
Even if we take som random sequence of 100 amino acids, randomly mutate it in 20 positions and it turns out that sequence is nonfunctional, what are the odds of that particular string of 20 mutations out of all the possible mutations that could have happened instead? Well, it is just as low as if 20 mutations had resulted in something functional.
Whether the mutations lead to function or not, if you specify 20 beforehand and try to calculate the odds you’re going to get a fantastically low number. Yet 20 mutations will accumulate regardless, so an event with that fantastically low probability will still happen.
So it’s an utterly worthless type of argument to make, to ask what the probability of it happening, or having happened, is. It tells you nothing about how the process came to be. Conditional probabilities neither establishes nor refutes evolution or design.
colewd,
The probability of you existing today as the person you are is (approx) one in 400 trillion.
http://blogs.harvard.edu/abinazir/2011/06/15/what-are-chances-you-would-be-born/
And yet the world is full of people!
Rumraket,
If we roll the dice and a very improbable outcome came up 10 times in a row, at the time we observed it the probability of the out come is indeed 100%.
The next question might be, what is the chance the result we are observing is the result of a stochastic process?