That’s not how FI is calculated. FI is calculated as you described earlier:
Where did you do this for the sequence under consideration? Do we see any conservation through deep time for the segments that end up in the variable region of the final antibody protein?
The real question at this point is if evolution needs to produce FI, as you define it, in order to produce the biodiversity we see today.
If we take @gpuccio most famous examples of FI, did he look at the genomes just prior to the emergence of these proteins and see if the information was already in place in the genome? Nope. All he has done is look at the conservation of a sequence once it acquired function. He has never looked at genomes prior to the emergence of these proteins and calculated FI. So how is it that pre-existing sequence is being cited in the example of antibodies when this is never done in other examples? I have also not seen any comparison of sequences across deep time as is done in @gpuccio’s other examples.
Please note that the quote you refer to is not about FI but about the amont of information (I) produced by the system. And I made clear in my post that the two notions should not be conflated.
Are you saying that @gpuccio’s method can not measure functional information?
Also, you have yet to describe how you calculated all of the possible amino acid combinations that could function as an antibody to a specific antigen.
Next, you have yet to demonstrate that this amino acid sequence existed in immature B-cells prior to V(D)J recombination.
That’s absolutely false, Gil.
This topic originated because you offered a testable hypothesis, misrepresented as a fact. You continue to misrepresent hypotheses as facts.
Let’s stay on track here. Your hypothesis is:
Why are you afraid of testing this? Why do you repeatedly misrepresent mere assertions as facts?
Credit where credit is due - Bill is much better than some others I can think of, and for this we should be thankful!
Interesting yes, but not relevant under the definition of FI. A very small change could cause the function threshold to be exceed. A very large change might lead to a great increase in practical function, but it doesn’t count because the threshold remains the same.
If I understand correctly a mutation/deletion could achieve the same function with a shorter sequence and it would count as decreased FI (someone correct me if this is wrong).
There are limits to any measure where we try to boil all the information content down to a single number. FI has a very specific usage and meaning, and (IMO) doesn’t extend to the interpretation you would like it to have.
Which only correlates with functional information in his cherry-picked cases.
Likely because it’s becoming obvious to @Giltil, and maybe even to @colewd, that FI is an extremely poor proxy for functional information.
Given that FI stands for functional information, what you are saying here is that functional information is an extremely poor proxy for functional information! Quite a strange statement, isn’t?
Since you haven’t applied @gpuccio’s measurements of FI to the question at hand, would it be safe to say that @gpuccio’s measurement is not a good proxy?
I think it would be more correct to say that FI as defined by Hazen et al 2007 is not a useful measure of FI in practice, given the difficulty of establishing with any appreciable confidence, the true FI of the system in question.
Hazen et al seems to have recognized as much in their 2007 paper:
We conclude that rigorous analysis of the functional information of a system with respect to a specified function x requires knowledge of two attributes: (i) all possible configurations of the
system (e.g., all possible sequences of a given length in the case of letters or RNA nucleotides) and (ii) the degree of function x for every configuration.
These two requirements are difficult to meet in many systems. In the case of letter sequences, for example, the sequence is obvious, but it is difficult to determine quantitatively the degree of function of many sequences. By contrast, it is relatively straightforward to determine the degree of function (for example, the ligand affinity) of any given RNA sequence, but impossible with present technology to measure all sequences in a large population, e.g., 10^14 randomly generated 100-mers as used in some aptamer evolution studies (although single-molecule methods may ultimately provide a technical solution to this challenge).
Note that the sequence space for all possible 100-mer RNAs is roughly 1.6×10^60, so in a typical evolution study they’re still off by 46 orders of magnitude.
At best, Hazen et al. can describe a single peak in a fitness landscape. They can’t calculate the total number of functional sequences, but they can find important residues for a specific and well conserved protein. As others have stated, Hazen et al. are measuring mutual information instead of functional information.
There are also layers of contingency. At the molecular level, there are inter- and intramolecular contigencies due to the interactions between and within molecules. This takes the form of Muller’s Two Step for protein-protein interactions and epistatic effects between residues in the same protein. At the organismal level there is the contingency of fitness where there are many possible adaptations, but when one is found it can be frozen in place and built on (i.e. finding a fitness peak on the fitness landscape). Out of all of this, you can’t simply look at the one protein that did evolve and pretend as if it is the only protein that could have evolved, or the only function that could have evolved. Mutual information can never tell you about total functional information.
No, I’m saying that it was presumptively and incorrectly named. If I named my son Donald Trump, would that make him president?
As with your VDJ hypothesis that you stubbornly misrepresent as fact, @gpuccio’s notion is merely a hypothesis, and an overtly false one at that.
Why do you so explicitly reject the scientific method, Gil? Do you not understand the essence of science, or do you just have too little confidence in your hypotheses to test them?
Would using French help?
No, it doesn’t mean this at all.
To see this, please consider my example of the mutant CFTR gene at 86. Would you say that by reversing the Pro into a Phe at position 508 the physicians would have generate an amount of information equal to the FI of the functional CFTR?
What would you say? What methods and equations would you use to quantitate FI before and after the mutation? If a single mutation changed the function of a protein, would that be new functional information?
If you randomly cut up 3 sections of the CFTR gene, each about 30 base pairs, and spliced them together, would you expect to have the same function?
I’m considering this case in real life and I only see a blatant contradiction of “FI” as a proxy for functional information.
In the real world, F508P and F508del function can be restored by introducing G550E, R553M, R553Q, or R555K to restore both proper glycosylation and function:
http://www.jbc.org/content/285/46/35825.full.pdf
This single experiment alone shows that @gpuccio’s “FI” does not measure functional information. If you disagree, show the math.
It’s truly amazing that both you and @colewd have freely chosen cases that demolish @gpuccio’s hypothesis.
Since you are claiming that “FI” works across all biology, cherry-picking a tiny number of cases and misrepresenting them as “examples,” while refusing to do the calculation for other systems is just plain unethical and unscientific.
Real scientists do much, much more real work than you’re willing to do before even tentatively suggesting that they’ve found something that applies universally.
Everyone makes mistakes. Ethical people admit them.
And the JBC paper cited above is a fine example of the latter. Moreover, the function of mutant CFTR proteins can be restored by making them fold differently to restore their trafficking and posttranslational modification, which would add precisely zero of @gpuccio’s “FI.”
F508P, as well as the most common allele F508del, are both functional as ion channels. They cause cystic fibrosis when homozygous because they are neither trafficked within the cell nor posttranslationally modified in the same way as the wild-type protein, so @gpuccio’s calculation fails miserably as a proxy for function.
Yes it does, Faizal is absolutely correct. Now you’re once again contradicting yourself when you wrote that:
“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.
Yes, I would say that. For reasons that I explained in this post of mine you seem to have skipped over. It logically follows from your acknowledgement that “a system (or a sequence) unable to perform the function has zero FI”. So it must logically go from having zero FI, to having all the FI of the sequence able to perform the function. So any single mutation that changes the sequence from nonfunctional to functional, gives it the full amount of FI. Not a fraction of the total FI, all of it. In one mutation.
You can’t keep making up your own new rules for defining and calculating FI as you situationally see fit. Or well you can, but we’re going to keep calling you out on it especially when you contradict yourself.