I predict that the next goalpost to be moved will be choosing one function while ignoring others. As I noted but Gil doesn’t seem to know, F508P still has the channel function…
1 Like
I don’t understand why you dispute the very simple point that a sequence can have almost all the specific bits necessary to implement a function. Let’s see if I can explain you this simple truth with the example below.
Let’s take the example of a safe lock of the type below, but with 100 wheels instead of 4. And let’s imagine the lock is there to protect a big bundle of money and that only one combination allows to open the safe.
You will agree with me that the function of the winning combination is to open the safe and that the FI associated with this winning combination is equal to -log2 ( 1/10^100 ), that is 332.193 bits.
Now consider the following 2 scenarios.
Scenario 1:
A thief comes to steal the loot but has no information at all about the only combination that can open the safe. Despite this total lack of information, the thief still tries to open the safe and finds the right combination on his first attempt! Now the question is: what is the amount of information associated with this extraordinary event? Quite simple; it is -log2 ( P ), with P, which is the probability to produce the winning combination in a single try, equal to 1/10^100. So the amount of information produced by the thief in this first situation is 332.193 bits, exactly equal to the FI of the winning combination.
Scenario 2:
The thief being a friend of the owner of the loot, he was able to spy on him several times while opening the safe. As a result, he knows all the numbers in the combination, except for the hundredth and last wheel. Armed with this knowledge, he then tries to open the safe. To do this, he first adjusts the first 99 wheels according to the information he has. IOW, he produces a sequence that has all the necessary bits of the winning combination except the 3,322 bits associated with the last wheel. Note here that in doing so, the thief does not produce any new information because it was already in his mind. Then, when he reaches the hundredth and last wheel, he choose a random number out of the 10 possible ones and the safe opens. In this situation, It must be clear for everyone that the amount of information produced by the thief in his attempt to open the safe is equal to 3,322 bits.
Let’s formalize all this.
Let’s call pI the preexisting information and let’s call nI the new information that the thief must produce to open the safe in a single try.
We have FI = pI + nI
So nI = FI - pI
In scenario 1, pI being equal to zero, with have nI=FI=332.193
In scenario 2, pI being equal to 328.87, we have nI = 332.193 - 328.87= 3.322
I hope the above analysis will have shown you that contrary to what you repeatedly and irrationally assert, a sequence can have almost all the specific bits necessary to implement a function, even if it can’t yet perform the function because some few bits are still missing. This is precisely the reason why the chance that the thief can open the safe is incomensurably higher in scenario 2 than in scenario 1.
I don’t understand why you contradict yourself, and now suddenly is using another definition of information instead of FI as it is defined by Hazen 2007. Under that definition, the function of the system is what determines whether it has any FI(not just information, or sequence information, but Functional Information).
I doesn’t matter, using that definition, how close to the functional state the system is, it doesn’t matter how many of the combination lock’s wheels that have already been set in, or how much of the genetic polymer that pre-exists, if that “information”(and we can call it information, just not Functional Information) alone can not perform the function of interest, then it has no FI.
And we know that you agree with this, you said as much:
“It is exactly what I said for I explicitly recognized that a system (or a sequence) unable to perform the function has zero FI.” - You just five days ago.
Is the combination lock with all but one wheel set able to perform the function of opening the safe? Nope. Well then Gilbert Thill thinks “that a system (or a sequence) unable to perform the function has zero FI”. Perhaps you should take it up with him.
We are treading the same territory over and over again. You’re confusing information, or sequence information, with Hazen et al 2007’s definition of Functional Information. You are mixing up different ways to define and calculate information. I think you should decide to just stick to ONE definition.
2 Likes
He is simply making a point outside the strict definition of functional information. In the condition observed which is an open safe with 100 10 unit digit dials to line up and one working combination we are observing over 330 bits of functional information.
The origin of that functional information is another discussion. In the case Gil described about 330 bits came from an intelligent agent and 3 bits came from trial and error.
Yes, so he shouldn’t be saying the last step creates 3.322 bits of FI. He can call it something else, but it’s not FI.
2 Likes
Oh by the way Bill, how would a “mind” create the FI for opening the lock if he wasn’t told by someone else? That’s right, trial and error. He’d have to test the combinations one after another until he found the correct one. Nobody can just sit there and think it up out of nowhere.
2 Likes
I was really hoping 2020 would be the year ID-Creationists finally realize arguments by analogy are useless as actual scientific evidence. But alas, looks like they’re still stuck in their fallacious thinking. 
3 Likes
It happens that I have exactly done that, ie., giving the 3.322 bits of the last step a specific name (nI), precisely to avoid the confusion with FI.
Good, so that last step of setting the 100th wheel adds all the FI under the formalism of FI. It adds ~3332 bits of FI in the case of the 100-wheel combination lock, not just 3.322.
But if you’re using using -log2 of sequence length and you’re only calculating the bit of a single wheel(length 1) instead of the system as a whole(length 100), it adds ~3.322 bits of information(not 3.322 bits of FI however).
However, I have a problem with this definition of information too, because it seems to imply recombination creates no new information. Which I explained in my very first post in this thread.
Under your definition of information, I can write a book with all the words rearranged and recombined into a new one with a totally different story and meaning, and you’d be saying there’s no new information in the new book, because all the sequences used in the new book are found in the old book.
Here it is again:
You’re confusing information with sequence.
You can always break up any sequence of symbols into smaller fragments and rearrange them, producing a new sequence. But then someone like you can come along and declare that the sequence isn’t new, it’s just rearranged fragments of already existing sequence
I could rearrange
CACACAGAGAGA
into
GAGAGACACACA
And you’d say there’s no new information, because it’s just the first sequence broken in half and the latter half put before the former.
But we could do that again, beak it up into smaller bits like CA and GA, and rearrange them.
GACAGACAGACA
And you could still say no new information, because it’s still just rearranged already existing seqeunce. All the CAs and GAs were all there to begin with.
And we could do it again, break it up into individual letters A, G, and C.
CCCAAAGGGAAA
And you could still say no new information, because it’s still just rearranged already existing sequence. All those As, Gs, and Cs were there to begin wtih.
Which reveals the absurdity of what you’re saying. So no, rearrangement really is new information. That is the only sensible position to take.
The problem is that you haven’t defined what constitutes “new” information to begin with, that means you can just make up the rules as you go along and keep denying that any example we come up with counts as new information. And the one you seem to be working from here is absurd.
2 Likes
This has always been the ID-Creationist strategy. Refuse to define their terms so they can always have wiggle room. Keep in mind IDC was never about trying to convince scientists. IDC pushers are trying to convince scientifically illiterate laymen who don’t know (or don’t care) how disingenuous the IDC arguments are.
5 Likes
The problem here for you is that my example of the safe lock is not at all an analogy with regard to the issue of FI. Both safe locks and proteins implement functions to which FIs can be attributed. So, as far as FI is concerned, reasoning on the safe lock example is a valid way to understand what’s going on with proteins.
Blatantly false. We can only calculate the FI for a lock exactly because we know beforehand the number of all possible combinations that meet the minimal threshold for function: It’s one.
We don’t know that with proteins. The combination lock analogy, which it of course still is to biology(a combination lock is not an organism, nor a biological polymer), fails for another reason too, which is that other combinations on the lock don’t have other functions. There is no possibility of overlap between different functions. The system just has one function by definition: To open the lock. You can’t spin the wheels and get a different function. But you really can mutate a protein to make it able to perform another function.
With proteins sequences they can have functions that overlap in sequence space, and different combinations of sequences can create new functions. That’s the whole basis for the adaptive immune system, the fact that VDJ recombination and somatic hypermutation occurs is to find those other functions that overlap with the existing sequence fragments. The adaptive immune system only works because there are other functions and they overlap in sequence space.
5 Likes
I see Rumraket has already kicked to the curb this false claim by you so there’s no need for me to pile on.
When are you ever going to provide your definition of “new information” and provide a genomic example of “new information” you accept? The second Tuesday of next week? 
3 Likes
Well, we do know with Gil’s failed attempt to cherry-pick that:
Single mutant F508P: less conservation, less function
Double mutant F508P+G550E (or any of three others): even less sequence conservation, much more function:
http://www.jbc.org/content/285/46/35825.full.pdf
specifically, Fig. 1.
So Gil, in the very example he chose, falsified the hypothesis that sequence conservation correlates with functional information. It also shows that Gil’s estimates of function in sequence space are far too low.
No, it’s not an example, it’s an analogy. That’s analogie en francais, so you have no linguistic excuse for your misrepresentation.
I think that if we are testing the hypothesis that protein sequence conservation correlates with functional information, we will learn and agree on much more by looking at proteins instead of locks. That seems obvious to me. Does it to you?
1 Like
Gil, you’ve made it perfectly clear that you don’t know very much about the relationship between protein sequence and function, so you have no basis for making that claim.
3 Likes
As there are different safes with different combinations. You can also change safe combinations. The analogy although not perfect is useful.
That has zero relevance. Getting the right combination for another safe, while trying to open this one, is useless as it won’t open this safe. It also won’t open the other safe. You simply have no idea whether the combination you are trying is even a successful combination for some other safe.
2 Likes
Wrong again, for the simple reason that I have never said that sequence conservation correlates with FI. My take is that sequence conservation through deep indicates high FI, but the reverse is false; you can have high FI without sequence conservation.
So how do you determine that it doesn’t simply indicate a local optimum on a more rugged fitness landscape?
1 Like
It’s worthless. An useful analogy would be a single safe with multiple combinations, and you have no clue how many there are.
You and Gil can’t even cherry-pick competently.
Again for CFTR:
F508P: reduced function
F508P with any of 4 suppressors: reduced homology, increased function.
Homology doesn’t correlate with function at all, Bill.
2 Likes