Introduction to Information

I agree with you that this is the conflict at hand. Out of your two ways to solve the porblem we've obviously chosen the opposite ones. Why? Just because it reduces disagreement with scientific authority? Why is that necessary or important?I'll tell you what. I chose the first alternative because I'm not familiar enough with Shannon's theories to defend them or counter their misuse. So I chose the first alternative. I know what information is in the informal context. The thing about debates with creationists, is that they're using Shannon's terms in the informal context. They're trying to apply Shannon's (and other's) logic (which is valid for what he called information) to what information is informally.If your argument is that that's incorrect in context, I agree. It's wrong to apply Shannon's work on Shannon-defined information to informally-defined information. You apparently think that whenever biology is discussed, the biological context is assumed. I wish this were the case. What is the case is that creationists talk about biology in the informal context, usually because they're not biologists.By Shannon's definition, DNA has information. But that wasn't the context I was referring to when I made my statement.[This message has been edited by crashfrog, 12-23-2003]

Crashfrog,Assuming this definitional problem with DNA is resolved, was there an assertion that DNAUnion was trying to make regarding information in the genome?Mark

quote:

---

By Shannon's definition, DNA has information.

---

Great! At the risk of misreprenting you, I am going to consider your acceptance of the statement "DNA contains information" in the stricter "Shannon way" as an implicit acceptance also of the much more general biological position that "DNA contains information".That leaves just one hold out now...Peter, who still rejects the simple and self-evident statement that DNA contains information.

Hi DNAunion:I've been following the discussion on this and the other thread with some interest. So far, I have not seen you make any statement that was controversial if taken at face value. However, I'm curious as to your underlying position in this debate - and in fact why you're arguing about it at all. There's one question that hasn't been answered in all this - although holmes (directly) and Peter (indirectly) asked it: When you, personally, refer to information in DNA, are you referring to a physical entity, similar to a particle or waveform that has independent existence? Or are you referring to an abstraction - a measurement tool or theoretical construct - as most of the quotes you've posted several times are explicitly using the term? If the former, then you have a lot of additional explanatory argument to produce. If the latter, then you've made no statement that anyone who is familiar with the subject should disagree with, and hence I don't really understand what the argument is about. Note: simply reiterating your quite uncontroversial statement that "information is contained in the sequence" etc does not address the question.Looking forward to your reply.

quote:

---

Hi DNAunion:
I've been following the discussion on this and the other thread with some interest. So far, I have not seen you make any statement that was controversial if taken at face value. However, I'm curious as to your underlying position in this debate - and in fact why you're arguing about it at all.

---

My position is simply that DNA contains information (and as a second position, that information theory can legitimately be applied to DNA). You're right...that is noncontroversial and there shouldn't be any argument about it at all.Difficulties persist because some people, when addressing DNA, are applying inappropriate definitions of the term information - theirs requires consciousness.

quote:

---

There's one question that hasn't been answered in all this - although holmes (directly) and Peter (indirectly) asked it: When you, personally, refer to information in DNA, are you referring to a physical entity, similar to a particle or waveform that has independent existence?

---

I have answered this, somewhere in the exchanges in this or the other thread. I probably did a better job the first than here, but....A coded symbol sequence, such as this sentence, contains information. But it is not as if I could crack open a letter of the English alphabet, reach in with some tweezers, and pull out a solid nugget of information: and no reasonable person would assume that is what is meant if someone said that these sentences contain information. The same goes for DNA: it stores information in its base sequences and no one should assume I am stating that I could crack open a nucleotide, reach in with some micromanipulator, and yank out a solid nugget of information.[This message has been edited by DNAunion, 12-24-2003]

quote:

---

I am stating that I could crack open a nucleotide, reach in with some micromanipulator, and yank out a solid nugget of information.

---

. Many of us probably DO know that this cant be done but perhaps when you say "I" are you saying that man can not be informed of this possibility?? I would not agree in that chance we (could) know what size a nugget"" is and was impenitrably. There seems to be enough letters and correspondence of scientists to follow thAT Up.

quote:

---

I suggested, several times, that under the 'reduction of
uncertainty' definition DNA still does not contain information,
since no sequence of DNA allows one to predict the next 'character'.

---

quote:

---

Try reading this:

​

http://www.lecb.ncifcrf.gov/...ider1986/latex2001/node4.html

​

If you still don't get it, I suggest you contact the author, T. Schneider, since it is he with whom you really disagree.

---

No reply from Peter yet...he's probably more tied up with Christmas than I am (until later today when I get my boys).Anyway, I thought others might be wondering about this "problem of no predictability of the next character" since Peter brought it up several times and others have referenced it too.The problem appears to be that Peter is looking at a base sequence on a SINGLE DNA strand for this "predictability" property and it is missing. But T Schneider looks at a SINGLE POSITION across MULTIPLE aligned VERSIONS of a single recognition site for the primary calculation of information: once the individual position's information values are calculated, then they are summed to get the total information for the binding site.First, just to give the overal directionality of the information calculations: Picture a spreadsheet where each row contains a "sentence" (analogous to a single DNA strand's base sequence): a series of 26 English characters, one per spreadsheet cell. Then stack up many "sentences", perhaps 1000, one per row (this is analogous to the many different versions of a given DNA binding site that are aligned). The first information calculation is based on COLUMNS, not ROWS. It would look at all the cells in a single COLUMN and calculate the Shannon information content of that column...imagine adding a =SHANNON(A$2:A$1001) function at the bottom of each column. Then the final information calculation would be a simple horizontal sum of the individual column's =SHANNON(A2:A1001) values: something like =SUM(A1002:Z1002).As a more communication-related explanatory analogy, think of a single modem (analogous to a single position from multiple aligned versions) and a collection of many modems...a modem bank (analogous to a "consensus sequence"). Each modem receives symbols and the information for each modem is calculated independently. For modem 'X' its information calculation allows us to predict what the next symbol received would be, within the degree of uncertainty. There's the predictive nature: it's for each position. Now, since information is additive, the individual modem's information results are simply summed to figure out the total information content of the modem bank. This cumulative information calculation does not allow us to predict what symbols would come from an additional modem were it to be added to the modem bank: the predictive nature is for each modem (position), not the bank of modems (a single DNA strand's sequence).What this all means is that the information calculations for the DNA binding sites DOES allow predictions to be made. If another version of the same binding site were found we could predict what base would be found at position 'x', and position 'y', and position 'z', etc., of that new sequence (the prediction would be constrained by the limits of the already calculated uncertainty associated with that position).

"When you, personally, refer to information in DNA, are you referring to a physical entity, similar to a particle or waveform that has independent existence? Or are you referring to an abstraction - a measurement tool or theoretical construct - as most of the quotes you've posted several times are explicitly using the term?If the former, then you have a lot of additional explanatory argument to produce." I think I know what you're getting at with that last (bolded) statement but I'd like to be sure. Please elaborate.

​

Joralex

Well, it appears that all opposition to my simple statement "DNA contains information" has finally been eliminated (I know, technically that was the other thread, but the two basically became equivalent).Peter was the last holdout and I have eliminated the main counter he offered (that of predictability). We've also seen that Peter was using an inappropriate definition of the term information when countering my statement.

Waht do you meen, mis-speelink????

Hello, hope you had a good christmas.

quote:

---

The problem appears to be that Peter is looking at a base sequence on a SINGLE DNA strand for this "predictability" property and it is missing.

---

So you agree that a single DNA sequence contains no information
according to the definition that you consider most approriate?

quote:

---

But T Schneider looks at a SINGLE POSITION across MULTIPLE aligned VERSIONS of a single recognition site for the primary calculation of information: once the individual position's information values are calculated, then they are summed to get the total information for the binding site.

---

The information above (and on the web-page to which you posted
a link) does not suggest information in DNA.It calculates an H value for a specially aligned set of data.Without the alignment the H value is stated as 2-bits (which for
4 bases represents maximum uncertainty).The conclusion (from an information PoV) is that the information
being measured is the information on the alignment, not of the
DNA.Since the alignment is performed intelligently, for the purpose
of prediction, why should one expect anything else.Finding a pattern in DNA is expected, since all DNA effectively
developed from one or a few original sequences.

Actually, Joralex, DNAunion answered the question quite satisfactorily in message 125. With his response, I no longer have any quibbles with the way he is using the term. He agrees with the majority of those scientists using information in relation to biology, several of whom he quoted: the term represents an abstraction or analytical tool, not a physical entity. The concept allows science to manipulate and analyze genetic "information" in its own right, disassociated from context - i.e., a highly useful analogy.

quote:

---

The problem appears to be that Peter is looking at a base sequence on a SINGLE DNA strand for this "predictability" property and it is missing.

---

quote:

---

So you agree that a single DNA sequence contains no information according to the definition that you consider most approriate?

---

Nope.First, I have not said that there is just one exact definition of information that is applicable to DNA, despite what you imply with your phrase the definition. What I have said is:(1) I see the definitions I have been using, which are all appropriate for the topic at hand, as being different sides of the same coin (a reduction in uncertainty, in one form or another).(2) YOUR definition, which requires a conscious agent for there to be information, IS inappropriate when discussing cellular DNA.Point (2) is key. It shows us why your position that DNA does not contain information is, and has been, invalid.

quote:

---

But T Schneider looks at a SINGLE POSITION across MULTIPLE aligned VERSIONS of a single recognition site for the primary calculation of information: once the individual position's information values are calculated, then they are summed to get the total information for the binding site.

---

quote:

---

The information above (and on the web-page to which you posted a link) does not suggest information in DNA.

---

LOL! You’re funny!

quote:

---

It calculates an H value for a specially aligned set of data.

---

You seem to be confused.It calculates an INFORMATION value for a binding site. As a part of doing that, it calculates INFORMATION values for each position in that binding site. As part of doing that, it calculates/uses TWO uncertainty values (before and after) for each position in that binding site. Did you realize all of that?

quote:

---

Without the alignment the H value is stated as 2-bits (which for 4 bases represents maximum uncertainty).

​

The conclusion (from an information PoV) is that the information being measured is the information on the alignment, not of the DNA.

---

The (total) information being measured is for DNA. The information of a single position is not enough to allow recognition, as Dr. Schneider explained. It takes a certain amount of information in order to be able to accurately and consistently select "1" out of an enormous number of possibilities, and one nucleotide position alone doesn't have enough information to do that. It requires the cumulative information, contained in a sequence of DNA bases, for binding to occur as needed.

quote:

---

Since the alignment is performed intelligently, for the purpose of prediction, why should one expect anything else.

---

Because the aligning doesn’t involved any illegitimate procedures, despite your insinuation.Why are those particular base sequences compared and not others? Because they are all recognized by a specific non-conscious, cellular machine as having the same meaning, such as, figuratively speaking, ribosome, bind here. The information is not generated or infused by humans: it’s there, being used by cells, whether or not any humans are around measuring it.Humans then collect all known sequences that CELLS have selected as having that same meaning, line them up so that position 0 within sequence 1 is associated with position 0 of sequence 2, both of which are associated with position 0 of sequence 3, and so on, then measure the information associated with position 0 across the collection of sequences, allowing one to predict that were another binding site found what base would be at its position 0 (within the limits of the uncertainty).But that single position, though giving us predictability, does not contain enough information for a recognizer to select its specific target from all of the various sequences it encounters. It takes the cumulative information of a sequence of bases to do that.

quote:

---

Finding a pattern in DNA is expected, since all DNA effectively developed from one or a few original sequences.

---

Out of curiosity, do you have any idea what functional constraint means? Do you know what happens to sequences if there are no functional constraints on them?PS: Do you yet agree that YOUR definition of information, which required a conscious agent for there to be information, was not appropriate to the discussion of celluar processes involving DNA?[This message has been edited by DNAunion, 01-07-2004]

quote:

---

PS: Do you yet agree that YOUR definition of information, which required a conscious agent for there to be information, was not appropriate to the discussion of celluar processes involving DNA?

---

Well ... duh!! Since I was saying that, according to the definition
of information that I used, there was no information in DNA
I would have thought that was self evident.

quote:

---

First, I have not said that there is just one exact definition of information that is applicable to DNA, despite what you imply with your phrase the definition. What I have said is:

​

(1) I see the definitions I have been using, which are all appropriate for the topic at hand, as being different sides of the same coin (a reduction in uncertainty, in one form or another).

---

So you, effectively, have only one definition (otherwise it
wuldn't be very definitive), and that is 'reduction in uncertainty'.
If all other definitions that you use are 'flip-sides' then
you really only have ONE definition.

quote:

---

The problem appears to be that Peter is looking at a base sequence on a SINGLE DNA strand for this "predictability" property and it is missing.

---

So were you saying in the above that the property of predictability
is missing, or that I claim it to be missing?

quote:

---

The (total) information being measured is for DNA. ....

---

No, it's not.I wasn't suggesting any under-handedness in the alignment, nor
that there is no usefullness in that process. What I am saying
is that the information measurement performed does not actually
measure information in DNA.It, basically, is being used as a filtering technique to
get rid of the noise on a signal to find the most likely
signal.It is suggesting that all E.Coli share ancestry.It is NOT measuring information content of DNA ... that measurement
provides a value of maximum uncertainty i.e. no information in
the Shannon sense.[This message has been edited by Peter, 01-08-2004]

quote:

---

First, I have not said that there is just one exact definition of information that is applicable to DNA, despite what you imply with your phrase the definition. What I have said is:

​

(1) I see the definitions I have been using, which are all appropriate for the topic at hand, as being different sides of the same coin (a reduction in uncertainty, in one form or another).

---

quote:

---

So you, effectively, have only one definition (otherwise it
wuldn't be very definitive), and that is 'reduction in uncertainty'.
If all other definitions that you use are 'flip-sides' then
you really only have ONE definition.

---

If you want to believe that fine, but I didn't say it...you did. Keep that in mind.

quote:

---

The problem appears to be that Peter is looking at a base sequence on a SINGLE DNA strand for this "predictability" property and it is missing.

---

quote:

---

So were you saying in the above that the property of predictability is missing, or that I claim it to be missing?

---

Your statement is sloppy and/or loaded.Predictability IS there, just not where you were looking for it.

quote:

---

What I am saying is that the information measurement performed does not actually measure information in DNA.

---

And you're wrong.

quote:

---

It is NOT measuring information content of DNA ... that measurement provides a value of maximum uncertainty i.e. no information in the Shannon sense.

---

Again, sloppy or loaded. Could you please learn to communicate effectively?