A Ubiquitin Response to Gpuccio

Yes. I’ve partly explained it here. Swamidass: Computing the Functional Information in Cancer . They have to take into account (1) the ability of common decent to create MI, and (2) alternate ways of arriving at the same function, and (3) the effect of purifying selection, and so on. Everyone of these factors biases their results upwards.

How do you establish functional information is increasing in cancer or is it assumed?

How do you establish that common descent can create FI or are you assuming this?


No, Bill. This is exactly the calculation they aren’t making and cannot make.

1 Like

Bear in mind that Joshua, like Joe Felsenstein, is being generous. They are accepting the concept of “Functional Information” at face value. The allowed assumption is that genotype maps to phenotype. This is too generous for me.


If this is false then what good does evolutionary claim have.

Evolution is not a claim about abstract sequences of symbols. That is to say, it is not a claim about information.

That the ID folk continue to construe evolution as a claim about information, only demonstrates the weakness of their program.

1 Like

The good of being a reasonable working hypothesis based on available evidence.


To elaborate a little, mathematics is a modelling tool. But to come to some conclusion about reality, your mathematical model has to match reality to some extent, otherwise you need to modify or discard your model. You can’t expect reality to conform to your model.


[quote=“nwrickert, post:111, topic:1608”]

That the ID folk continue to construe evolution as a claim about information, only demonstrates the weakness of their program.

And if you take a number and change it to “negative log to the base 2” it tells you something more than the original number.

If you want to make a criticism of a mathematical model you have to directly access that model and make specific comments. This is what Josh is doing with Eric H. Both models have strengths and weaknesses and are the beginning of measuring functional information.

I’ve done that on Durston’s work @colewd. Did you not know?

1 Like

Yet to see any strengths of Eric H. model. Josh’s simulations crush the hypothesis.

Apples and oranges. Duston’s and Puccio’s models are based on empirical measurement and not just a math proof. It is also too early to tell on Eric’s model.

No. Can you cite?

I’ve engaged at length with Durston. Also Eric Holloway correctly explains that Durston is applying the exact same method he has put forward. Both Durston and @EricMH claim that FSC (that he calculates) is just an example of ASC. Look what Eric says about ASC (which FSC from Durston is an example):

That is exactly right. I entirely agree. However, Durston does not even try to formulate a P that takes into account common descent. That is his error. That means his estimate of is going to be much higher than it is supposed to be. He did not even test this with simulation. If he had, he would have found out that the vast majority of what he is measuring is the mutual information that arises from common descent.

Until Durston can produce a valid P, the FSC method fails. Even then, it still fails, as I have covered elsewhere. However just taking into account common descent is enough to explain the vast majority of information he is measuring.

1 Like

How did you validate that common descent generates FI? Copied information is not new information.

For FSC, copied information registers as new information. It is mutual information after all. If you think that is wrong because copied information isn’t new information, you are making my point for me. Go explain it to Durston. That is what his algorithm is measuring.

1 Like

Take a look a Szostak and Hazens paper. What you are claiming is that functional information was added when you came from your parents. This is about the ratio for function to non function in sequence space as illustrated by S and H.

There is no mechanism involved in common descent that has been shown to increase FI except possibly recombination in plants which is more likely simply re arranging it.

We also need to come to consensus on the definition of what common descent is really claiming.

If you think that is wrong because copied information isn’t new information, you are making my point for me.

I am happy to make a point for you. BTW very nice post re VJT’s article at TSZ.:smiley:


@colewd please stop asking me to read sostack. I’ve read it and am telling you it has nothing to do with FSC. FI is not merely copied information, but FSC (Durston’s measure) counts copied information as if it is functional information. Therefore FSC is not measuring FI.

1 Like

I don’t see this as what he is doing. He is measuring FI by taking total sequence space and reducing it by the substitutability of AA’s. Theoretically if a sequence was indifferent to the AA at any sequence FI would equal zero. He is measure a single sequence and not a copied version.

You are right he is using protein families but this is just to estimate substitutability.