NIH: 100M Years to Change a Binding Site

[Stripe]

I am not conversant with the exact definitions and rules applicable to information theory, so for now I occasionally ask a specific question (like about rabbits) that can probably be considered as an applied case of information theory.

Oh. Well, it's pretty simple.

Information theory can be applied to any data set. All you need is a string of variables that can take a known range of values. Once you have the size of the data set you can calculate ways to forward the same message on using as few bits as possible.

So say you get a whole bunch of phone numbers that you want to transmit by morse code. Instead of assigning all the digits 0-9 an individual code you can analyse the data and assign codes according to certain characteristics. For instance, phone numbers have area codes of certain lengths. Let's say all the area codes are two digits, zero-something. Instead of calling 09 by it's zero-nine Morse equivalent (----- ----) you could call it ---- because the receiver can be told that every first digit is zero.

Or say the area codes are only 06-09 then you could send 09 as -- because you only need two beeps to distinguish between the four possible values (06= .. 07 = .- 08 = -. 09 = --).

Thus if your receiver is informed that the first two digits of each phone number are an area code from 06,07,08,09, and how those four values are represented then you can send two beeps instead of nine.

There are several other techniques that can be applied to further reduce the input required. The greater the reduction in input the less the entropy of the data set. The uncertainty that the receiver starts with minus the uncertainty the receiver has afterward is the information transmitted.

I realise you're not very informed ( :chuckle: ) on this subject, but I'm willing to bet you can quickly understand these ideas and understand how very useful they are.

Now, can you tell us why information theory cannot be applied to the data read from DNA in ATGC form?

For the purpose of living in the forest, I agree the white fur might, in a sense, be considered “corrupted information”.

Actually, it has nothing to do with where the animal lives or the advantage conveyed. It has to do with the intent of the information.

But for the purpose of living in a snowy environment, the white fur is an improvement. As applied to evolution, isn’t this a case where a “corruption” of the information was beneficial?

First of all, I'm not sure your example is very appropriate. White fur in rabbits is a perfectly normal attribute and need not be attributed to any mutation.

An evolutionist cannot speak of information without undermining his own beliefs. No, corruption of information is never beneficial regardless of these sorts of side effects. Sure, white fur might be some advantage. But it can only be considered a gain of any sort if one ignores the fact that the integrity of the information has been compromised.

 

[Tyrathca]

Yes.

Do you have a response to the question?

Can you please show how you calculated the information content of each and then determined that the the content of of the "Daddy rabbit" was greater than the content of the "kid rabbit".

Otherwise it seems you are relying merely on the knowledge that Daddy preceded kid. While it may be useful to assume that any change from the original is a loss in human communications this is merely because the results feed well into quantifying desired/undesired outcomes. It should be obvious that applications to biology would not find this assumption appropriate.

So please show the math showing the quantitative difference between the two (or two real life equivalents).

 

[Jukia]

Can you please show how you calculated the information content of each and then determined that the the content of of the "Daddy rabbit" was greater than the content of the "kid rabbit".

Otherwise it seems you are relying merely on the knowledge that Daddy preceded kid. While it may be useful to assume that any change from the original is a loss in human communications this is merely because the results feed well into quantifying desired/undesired outcomes. It should be obvious that applications to biology would not find this assumption appropriate.

So please show the math showing the quantitative difference between the two (or two real life equivalents).

Asking Stripe to show his math usually results in this answer:
"2+2=4".

 

[Stripe]

Can you please show how you calculated the information content of each and then determined that the the content of of the "Daddy rabbit" was greater than the content of the "kid rabbit".

By accepting as accurate the fact that random changes are always bad for information.

Otherwise it seems you are relying merely on the knowledge that Daddy preceded kid. While it may be useful to assume that any change from the original is a loss in human communications this is merely because the results feed well into quantifying desired/undesired outcomes. It should be obvious that applications to biology would not find this assumption appropriate.

Whyever not? Perhaps you can answer the question.

Why can information theory not be applied to the data read from DNA?

 

[Jukia]

By accepting as accurate the fact that random changes are always bad for information.

Whyever not? Perhaps you can answer the question.

Why can information theory not be applied to the data read from DNA?

And notice the math Stripe used.

 

[Stripe]

:mock: Jukia

 

[DavisBJ]

Oh. Well, it's pretty simple.

Information theory can be applied to any data set. All you need is a string of variables that can take a known range of values. Once you have the size of the data set you can calculate ways to forward the same message on using as few bits as possible.

So say you get a whole bunch of phone numbers that you want to transmit by morse code. Instead of assigning all the digits 0-9 an individual code you can analyse the data and assign codes according to certain characteristics. For instance, phone numbers have area codes of certain lengths. Let's say all the area codes are two digits, zero-something. Instead of calling 09 by it's zero-nine Morse equivalent (----- ----) you could call it ---- because the receiver can be told that every first digit is zero.

Or say the area codes are only 06-09 then you could send 09 as -- because you only need two beeps to distinguish between the four possible values (06= .. 07 = .- 08 = -. 09 = --).

Thus if your receiver is informed that the first two digits of each phone number are an area code from 06,07,08,09, and how those four values are represented then you can send two beeps instead of nine.

There are several other techniques that can be applied to further reduce the input required. The greater the reduction in input the less the entropy of the data set. The uncertainty that the receiver starts with minus the uncertainty the receiver has afterward is the information transmitted.

I realise you're not very informed ( :chuckle: ) on this subject, but I'm willing to bet you can quickly understand these ideas and understand how very useful they are.

Now, can you tell us why information theory cannot be applied to the data read from DNA in ATGC form?

See below

Actually, it has nothing to do with where the animal lives or the advantage conveyed. It has to do with the intent of the information.

First of all, I'm not sure your example is very appropriate. White fur in rabbits is a perfectly normal attribute and need not be attributed to any mutation.

My example was for illustrative purposes. There are documented cases where a novel capability was introduced into an organism, similar to previously all-brown rabbits having a white descendant.

An evolutionist cannot speak of information without undermining his own beliefs. No, corruption of information is never beneficial regardless of these sorts of side effects. Sure, white fur might be some advantage. But it can only be considered a gain of any sort if one ignores the fact that the integrity of the information has been compromised.

I agree that changes in the DNA in the brown rabbits would likely be detrimental – for brown rabbits. But that is precisely the point of evolution. Evolution is not primarily concerned with brown rabbits having brown furred kids. Evolution is asking what happens when brown rabbits live next to a snow field, where white fur would be better. A corruption in DNA for a rabbit that is going to stay in the forest might be a corruption that would be beneficial for the snow. A faithful reproduction of the brown DNA message is not in the best interest of the snow-bound rabbits.

To expand a bit – you said that going from the brown fur to the white was a loss of information. Let me denote this as DNA “W” (white fur) having less information than DNA “B” (brown fur). But what I did not tell you is that a couple thousand years earlier that entire rabbit biome was coming out of an ice age, and was all covered in snow. Back then, all rabbits had white fur, and it was a brown DNA mutation that allowed them to move into the forest. Eventually all the rabbits with the original white fur gene died off. By your rules at that time, going from the original DNA (white fur) to the corrupted DNA (brown fur) was a loss of information.

Now, as the weather brings snow back, that original white fur gene would come in really handy. But for reasons I cannot fathom a mutation which goes back to the white fur gene is, under your rules, another loss of information. If that ice field spreads and recedes a few dozen times, and the rabbits adapt accordingly, they must lose so much information that they not only don’t have any DNA information at all, but they are sucking it out of the information debtors bank. Yet looking at the DNA itself, as far as the while and brown fur genes go, they have just come and gone several times over.

Strange world you live in, where going from W to B to W to B to W means the last W has far less info than its carbon-copy original W.

 

[Stripe]

My example was for illustrative purposes.

I know. It wasn't a very good example for the reasons given.

There are documented cases where a novel capability was introduced into an organism, similar to previously all-brown rabbits having a white descendant.

White isn't novel to rabbits. :nono:

I agree that changes in the DNA in the brown rabbits would likely be detrimental – for brown rabbits. But that is precisely the point of evolution. Evolution is not primarily concerned with brown rabbits having brown furred kids. Evolution is asking what happens when brown rabbits live next to a snow field, where white fur would be better. A corruption in DNA for a rabbit that is going to stay in the forest might be a corruption that would be beneficial for the snow. A faithful reproduction of the brown DNA message is not in the best interest of the snow-bound rabbits.

This is entirely irrelevant.

Evolutionists must, of course, claim that random changes can be beneficial. But they can only do so while ignoring the facts of information theory.

To expand a bit – you said that going from the brown fur to the white was a loss of information. Let me denote this as DNA “W” (white fur) having less information than DNA “B” (brown fur). But what I did not tell you is that a couple thousand years earlier that entire rabbit biome was coming out of an ice age, and was all covered in snow. Back then, all rabbits had white fur, and it was a brown DNA mutation that allowed them to move into the forest. Eventually all the rabbits with the original white fur gene died off. By your rules at that time, going from the original DNA (white fur) to the corrupted DNA (brown fur) was a loss of information. Now, as the weather brings snow back, that original white fur gene would come in really handy. But for reasons I cannot fathom a mutation which goes back to the white fur gene is, under your rules, another loss of information. If that ice field spreads and recedes a few dozen times, and the rabbits adapt accordingly, they must lose so much information that they not only don’t have any DNA information at all, but they are sucking it out of the information debtors bank. Yet looking at the DNA itself, as far as the while and brown fur genes go, they have just come and gone several times over.

Correct. Both changes (assuming that random mutations were responsible for them - which is highly unlikely) resulted in a degradation of the information.

And there's a prediction for you. A population that is known to adopt different traits according to an environmental change will progressively lose that adaptive ability as successive generations are switched between the two environments. So take the famous nylon digesting bacteria that evolutionists love. Have a stock of original bacteria moved into nylon and watch them adapt to digesting nylon. Move that population back to the normal environment and observe how poorly they readjust. Continue moving successive descendants of the same original population between the two environments and eventually all adaptive ability will vanish.

Strange world you live in, where going from W to B to W to B to W means the last W has far less info than its carbon-copy original W.

Your example is ignoring the fact that there is much more to DNA than the colour of a rabbit's fur. The must be a cost for any changes that result from random mutations (fur colour probably isn't from mutation). That cost must exist even though you cannot see it.

 

[Sherman]

:mock: Jukia

Her last few posts were pretty snarky, adding nothing to the discussion. Lets hope she's better behaved when she gets back.

My thought on mutations: Have we seen any that were actually beneficial over the long term? When an animal or a person is subjected to radiation and they have offspring, are the mutated offspring better off? Not. I've seen pictures of five legged cows, a snake with three heads etc. These creatures are not better off.

Another human mutation is albinism. An albino is not better off either. It is asociated with the number of vision defects, photo-phobia, nystagmus and astigmatism. The person is more susceptible to sunburn and skin cancers.

 

[DavisBJ]

I know. It wasn't a very good example for the reasons given.

It is a fine example, if your focus is on the underlying concepts it illustrates.

White isn't novel to rabbits. :nono:

Nor did I say it was. I only said that there are clearly documented cases of mutations bestowing a new capability on an organism. That too hard for you to get your mind around?

This is entirely irrelevant.

It is not irrelevant, it is the crux of our disagreement.

Evolutionists must, of course, claim that random changes can be beneficial. But they can only do so while ignoring the facts of information theory.

But it appears that you are defining DNA information theory as dealing with the accurate transmission of the DNA information from generation to generation. By definition, “beneficial mutations” are DNA errors which are favorable, and so at the outset they lie outside the realm of perfectly accurate DNA message replication. That means your use of information theory is defined to be something other than what evolution depends on. You are applying a round tool to a square problem, and then blaming the problem.

Correct. Both changes … resulted in a degradation of the information.

There was a degradation of the information only from the view of the parent population. For the children rabbits who were thereby able to move into a new niche the modified information was more valuable to them.

And there's a prediction for you. A population that is known to adopt different traits according to an environmental change will progressively lose that adaptive ability as successive generations are switched between the two environments. So take the famous nylon digesting bacteria that evolutionists love. Have a stock of original bacteria moved into nylon and watch them adapt to digesting nylon. Move that population back to the normal environment and observe how poorly they readjust. Continue moving successive descendants of the same original population between the two environments and eventually all adaptive ability will vanish.

That is not part of info theory, is it?

Your example is ignoring the fact that there is much more to DNA than the colour of a rabbit's fur.

Sure. The original snow DNA message told the genome to have sharp eyes, good hearing, fast legs, and camouflage white coloring. Then a kid was born with a corrupted DNA message that said have sharp eyes, good hearing, fast legs, and camouflage brown coloring.

The must be a cost for any changes that result from random mutations (fur colour probably isn't from mutation). That cost must exist even though you cannot see it.

There is not an a-priori reason why there has to be a cost. And in those cases where there is a cost, it must be weighed against the benefit the favorable mutation brings. If the mutated rabbit can now access a significant source of previously inaccessible food, it may well be worth it.

 

[DavisBJ]

My thought on mutations: Have we seen any that were actually beneficial over the long term? When an animal or a person is subjected to radiation and they have offspring, are the mutated offspring better off? Not. I've seen pictures of five legged cows, a snake with three heads etc. These creatures are not better off.

Another human mutation is albinism. An albino is not better off either. It is asociated with the number of vision defects, photo-phobia, nystagmus and astigmatism. The person is more susceptible to sunburn and skin cancers.

Most of the mutations you list are of dramatic in-your-face changes, which are probably very rarely good. From Darwin to today it has been recognized that most beneficial evolutionary changes are going to be small, and often would go unnoticed in the short term.

 

[Stripe]

It is a fine example, if your focus is on the underlying concepts it illustrates.

You're assuming the truth of evolutionary theory. I'm assuming the truth of information theory.

Which underlying concept should we assume as true?

Nor did I say it was. I only said that there are clearly documented cases of mutations bestowing a new capability on an organism. That too hard for you to get your mind around?

And this is not necessarily one of those cases. In fact it almost certainly isn't. If whiteness was already possible then there is no need for a random change to make the population white again.

It is not irrelevant, it is the crux of our disagreement.

It's necessary as a crux for your explanation. It's utterly irrelevant for mine.

But it appears that you are defining DNA information theory as dealing with the accurate transmission of the DNA information from generation to generation.

Close enough.

By definition, “beneficial mutations” are DNA errors which are favorable

And by definition a random change comes at a cost to information. Every time. Thus a beneficial mutation must come at a net cost even if you do not know what that cost is.

There was a degradation of the information only from the view of the parent population. For the children rabbits who were thereby able to move into a new niche the modified information was more valuable to them.

Possibly. But there is still a net cost.

That is not part of info theory, is it?

It's a prediction based upon my understanding of biology and information theory.

Sure. The original snow DNA message told the genome to have sharp eyes, good hearing, fast legs, and camouflage white coloring. Then a kid was born with a corrupted DNA message that said have sharp eyes, good hearing, fast legs, and camouflage brown coloring.

That's a simplistic way of looking at it. DNA is information on how to build a rabbit body. It doesn't contain information for "fast legs", it just knows how to build a rabbit that will be able to run fast.

And it is most likely that rabbits were originally created with the capacity to adapt their colour according to the environment.

There is not an a-priori reason why there has to be a cost.

Yes, there is.

Random changes mean the receiver cannot predict the next result as accurately. Thus information decreases. Take the telephone number example. The data compression protocol might be based on a relationship between subsequent numbers. So the receiver knows that (say) after a 5 as the third digit there is always a digit within 0-5 as the fourth. If a random change could make the fourth digit a 7 then the uncertainty goes up and information is lost.

And in those cases where there is a cost, it must be weighed against the benefit the favorable mutation brings. If the mutated rabbit can now access a significant source of previously inaccessible food, it may well be worth it.

The cost can only be measured by looking at the overall capacity for the DNA to build a rabbit. You are ignoring the cost by looking at one or two expressions of the information.

 

[Frayed Knot]

Most of the mutations you list are of dramatic in-your-face changes, which are probably very rarely good.

The ones he listed were not even genetic mutations - they were developmental disorders. A five-legged cow or a three-headed snake have perfectly good DNA, but something in the development went wrong.

Back to the information in DNA, I know a lot more about the subject that Stripe and I've posted earlier in this thread exactly how he's wrong. He just puts his fingers in his ears and says "nuh-uh." If someone wants to ask a question, I'll be more than happy to answer it, but I have no interest in dealing with dishonesty so won't be responding to Stripe.

 

[Alate_One]

Another human mutation is albinism. An albino is not better off either. It is asociated with the number of vision defects, photo-phobia, nystagmus and astigmatism. The person is more susceptible to sunburn and skin cancers.

How about a human that is essentially immune to AIDS and smallpox? Or a human with a much higher chance of living to 100 and staying healthy while doing it? How about a human that doesn't get heart disease nearly as easily? All of these are known and documented mutations. I'm sure your creationist magazines don't want to talk about those . . . .


Frayed already mentioned that three legged and two headed animals are often not due to mutations but developmental problems.

 

[Stripe]

The ones he listed were not even genetic mutations - they were developmental disorders. A five-legged cow or a three-headed snake have perfectly good DNA, but something in the development went wrong.

How did something go wrong in the development without something going wrong in transmission and reception of the DNA?

Back to the information in DNA, I know a lot more about the subject that Stripe and I've posted earlier in this thread exactly how he's wrong. He just puts his fingers in his ears and says "nuh-uh." If someone wants to ask a question, I'll be more than happy to answer it, but I have no interest in dealing with dishonesty so won't be responding to Stripe.

:rotfl:

:loser:

 

[Lovejoy]

The ones he listed were not even genetic mutations - they were developmental disorders. A five-legged cow or a three-headed snake have perfectly good DNA, but something in the development went wrong.

Back to the information in DNA, I know a lot more about the subject that Stripe and I've posted earlier in this thread exactly how he's wrong. He just puts his fingers in his ears and says "nuh-uh." If someone wants to ask a question, I'll be more than happy to answer it, but I have no interest in dealing with dishonesty so won't be responding to Stripe.

A two headed snake is, in essence, a form of a conjoined twin, which is why it is not a mutation. However, a five-legged cow would be similar to polydactyly, which in humans is usually a mutation in GLI3. I think development problems resulting in extra limbs/digits are rare.

 

[Tyrathca]

By accepting as accurate the fact that random changes are always bad for information.

I notice you didn't show any of the maths you used to quantify the information and then demonstrate this principle. I could have sworn that I asked for that rather than a mere statement...

Whyever not? Perhaps you can answer the question.

Why can information theory not be applied to the data read from DNA?

Because the assumption relies on desired outcomes, of which there are none in biology. Since there is no preference for the original data, merely one which survives, it makes no sense to use such an assumption in your calculations.

My problem is with your application not with information theory itself. Information theory is a broad subject with multiple ways to quantify information. All I'm asking for is for you to show that the information drops (preferably for all types), rather than just claim it. Thus far your particular usage seems to assume a condition for application (that only the original information is wanted and thus any change is loss) as demonstrated to be be universal rather than a convenient assumption for certain scenarios (i.e. practical applications in human technology).

 

[Stripe]

I notice you didn't show any of the maths you used to quantify the information and then demonstrate this principle. I could have sworn that I asked for that rather than a mere statement.

The statement fully addresses the issue and it is currently unknown exactly what DNA codes for all aspects of a trait.

So you're asking for something nobody can answer and rejecting the answer that is universally accepted.

Because the assumption relies on desired outcomes

What assumption relies on desired outcomes? Mine? My assumption is that it is true that random changes are detrimental to information. How does that rely on "desired outcomes"?

Since there is no preference for the original data, merely one which survives, it makes no sense to use such an assumption in your calculations.

Which is simply assuming the truth of evolutionary theory and rejecting information theory again. Can you tell us why information theory cannot be applied to evolution?

My problem is with your application not with information theory itself.

Yet you're trying to answer the question. :think:

Information theory is a broad subject with multiple ways to quantify information.

So?

All I'm asking for is for you to show that the information drops (preferably for all types), rather than just claim it.

How about you show us how information theory cannot be applied to data from DNA? It's just data. You just put numbers on it. Why can't we do this with data from DNA?

Thus far your particular usage seems to assume a condition for application (that only the original information is wanted and thus any change is loss) as demonstrated to be be universal rather than a convenient assumption for certain scenarios (i.e. practical applications in human technology).

It's not something one applies selectively. Every time you make random changes to numbers, any information that was put into them is degraded.

 

[Stripe]

Here's another example of how information works. I give you data of which you have no indication how it is generated.

05175753095160065199041151069923580357

In order to try and figure out the English message contained within these numbers you can apply information theory (it's not important that you figure out the message, it is important that you realise that random changes can only make your task more difficult).

So to start with you might notice that there are five nines in the code and four are found in two somewhat evenly spaced pairs. Even with this small sample it should be clear those four nines mean something useful and with a larger piece of code their use would be immediately obvious.

But imagine this was the code you got (one error) and you did not have the clues about the nines:

05175753095160065199041151069023580357

Now you only have four nines and no immediately obvious pattern. Has the information available to you been affected?

Now the question. Why can this sort of work not be done on data from DNA?

 

[DavisBJ]

You're assuming the truth of evolutionary theory. I'm assuming the truth of information theory.

Which underlying concept should we assume as true?
I am confident in the correctness of evolution. And I see no reason to doubt information theory, when it is correctly applied.

And this is not necessarily one of those cases. In fact it almost certainly isn't. If whiteness was already possible then there is no need for a random change to make the population white again.

…

And by definition a random change comes at a cost to information. … And it is most likely that rabbits were originally created with the capacity to adapt their colour according to the environment. … The cost can only be measured by looking at the overall capacity for the DNA to build a rabbit. You are ignoring the cost by looking at one or two expressions of the information.

You keep focusing on the rabbits rather than the underlying concepts. Forget rabbits. Instead, I am going to speak of a new organism called “boders”. Their DNA does not retain memory of past fur colors, since in boders the gene that controls fur color has no provisions for alternative colors. If some mutation causes it to change fur color, that mutation replaces the original DNA coloring mechanism. Once all of the members of the group that have the old color are dead, there is no “memory” in the DNA of that old color.

Now I am going to address, with a specific example, your claim that any corruption of the DNA message entails an information loss. Daddy boder has brown fur, kid boder has white fur. The kid leaves the forest and moves into the snowy environ (the boders have long overtaxed the forest food supply anyway).

Question 1) From the viewpoint of a snow-bound kid boder – which DNA has the most “information” – that coding for brown fur like its dad, or its own DNA which gives it white fur?

Question 2) From the viewpoint of a forest living parent boder – which DNA has the most “information” – that coding for brown fur like it has, or the DNA which gives its kid white fur?