🚀 go-pugleaf

swarat_c@lycos.com (Swarat Chaudhuri) wrote in message news:<51c62199.0112171741.565cea72@posting.google.com>...
> May some kindred spirit shed some light on the following!
> 
> Has there been any effort to mathematically describe the minimum  
> information content of a DNA string? For instance, in the lines of 
> Kolmogorov complexity? Note that DNA strings are not completely
> random:
> there is a certain amount of structure. Chaitin's 1977 article on 
> algorithmic information theory points to some possible (IMHO, 
> rather ambitious) directions in which information theory may be
> applied to
> biological systems. He talks about giving a quantitative
> characterization
> of the structural information contained in a living organism. 
> 
> My question is: has there any further (concrete) work in this
> direction? If yes, can anyone supply some pointers? I checked out the
> webpage of a lab at NCI which works on applying notions of entropy to
> biosystems, but this doesn't seem to be remotely close to what
> Chaitin discusses.
> 
> Disclaimer: I am more or less illiterate so far as biology literature 
> goes, so don't flame me if this turns out to be a perfectly standard
> and
> well-studied issue in theoretical biology.  
> 
> Cheers,
> Swarat



Well, I don't think you would call it "perfectly standard", but there
are a number of people working in this area.

Try Tom Schneider's page for starters (and lots of links):
http://www.lecb.ncicrf.gov/~toms/

Cheers,

Matt

Swarat Chaudhuri <swarat_c@lycos.com> wrote:
> May some kindred spirit shed some light on the following!

> Has there been any effort to mathematically describe the minimum  
> information content of a DNA string? For instance, in the lines of 
> Kolmogorov complexity? Note that DNA strings are not completely
> random:
> there is a certain amount of structure. Chaitin's 1977 article on 
> algorithmic information theory points to some possible (IMHO, 
> rather ambitious) directions in which information theory may be
> applied to
> biological systems. He talks about giving a quantitative
> characterization
> of the structural information contained in a living organism. 

> My question is: has there any further (concrete) work in this
> direction? If yes, can anyone supply some pointers? I checked out the
> webpage of a lab at NCI which works on applying notions of entropy to
> biosystems, but this doesn't seem to be remotely close to what
> Chaitin discusses.

Don't know about "concrete work", but a serious theoretical issue is
this: genes are not DNA sequences, but are encoded in such sequences.
The decoding mechanism is the cellular machinery that interacts with
the DNA.  Anyone who doubts that needs to read Darwin's Dangerous Idea
by Dan Dennett.  And it's the gene, actually an abstraction, that carries
the information.

-- 
Robin Faichney 
"It is tempting to suppose that some concept of information could serve
eventually to unify mind, matter, and meaning in a single theory," say
Daniel Dennett and John Haugeland. The theory is here: http://www.ii01.org/

In article <j9gtv9.tg.ln@linuxsys.ii01.org>,
Robin Faichney  <robin@ii01.org> wrote:
>
>Don't know about "concrete work", but a serious theoretical issue is
>this: genes are not DNA sequences, but are encoded in such sequences.
>The decoding mechanism is the cellular machinery that interacts with
>the DNA.  Anyone who doubts that needs to read Darwin's Dangerous Idea
>by Dan Dennett.  And it's the gene, actually an abstraction, that carries
>the information.

The same is true of computer programs!  Remember the old amusement
of writing a program that would be valid in N languages and would
print the name of the language it was compiled as when run?

Quite a few mavericks (yes, I admit to being one) have been saying
for some time that the engineering models of computer programming
match what is actually going on less well than the biological ones
of breeding and so on.  That is the converse of the question that
was being asked, but has its relevance.

One of the consequences is that, as with English text, what you
regard as the information content depends very much on at what
level you are analysing it.  This is very true of DNA, but we do 
not yet know enough about how it works to know what levels of    
analysis are meaningful.

So I don't think that there is a simple answer, and all answers
will be pretty fair hand waving for a long time yet - probably
half a century or so, though progress may speed up.


Regards,
Nick Maclaren,
University of Cambridge Computing Service,
New Museums Site, Pembroke Street, Cambridge CB2 3QH, England.
Email:  nmm1@cam.ac.uk
Tel.:  +44 1223 334761    Fax:  +44 1223 334679

(Sorry, I missed the original poster's article & now it seems to be Gone.)

In article <b86268d4.0112201150.52a5833e@posting.google.com>,
Matt Jones <jonesmat@physiology.wisc.edu> wrote:
>swarat_c@lycos.com (Swarat Chaudhuri) wrote in message news:<51c62199.0112171741.565cea72@posting.google.com>...

>> Has there been any effort to mathematically describe the minimum  
>> information content of a DNA string? For instance, in the lines of 
>> Kolmogorov complexity? Note that DNA strings are not completely
>> random:
>> there is a certain amount of structure.

Well, yes, but you might have guessed this from the fact that you and
every other DNA-based organism exists ;).

>>Chaitin's 1977 article on 
>> algorithmic information theory points to some possible (IMHO, 
>> rather ambitious) directions in which information theory may be
>> applied to
>> biological systems. He talks about giving a quantitative
>> characterization
>> of the structural information contained in a living organism. 

I'd suggest looking up Chris Adami's book (_Intro to Artificial Life_)
and any articles he's published on information theory recently.  You
could also contact him directly by going to his Web site

	http://www.krl.caltech.edu/~adami/

cheers,
--titus

In article <j9gtv9.tg.ln@linuxsys.ii01.org>,
Robin Faichney  <robin@ii01.org> wrote:
>
>Don't know about "concrete work", but a serious theoretical issue is
>this: genes are not DNA sequences, but are encoded in such sequences.
>The decoding mechanism is the cellular machinery that interacts with
>the DNA.  Anyone who doubts that needs to read Darwin's Dangerous Idea
>by Dan Dennett.  And it's the gene, actually an abstraction, that carries
>the information.

Actually, the concept of gene itself is a bit tough to pin down.  Most
molecular biologists tend to just define the gene as the protein-coding
region, although regulatory regions -- which can be just as complicated
as the gene -- tend to get called "genes" by geneticists, who w/o the
sequence usually can't tell if their loci are in protein coding regions
or not.

I'm also not sure what you mean by abstraction: the amino acid sequence is
very strongly tied to the genomic sequence; heck, in prokaryotes, where you
don't have to worry about differential splicing and mRNA processing,
it's absolutely tied to the genomic sequence.

I wouldn't recommend Dennett for this basic stuff, if only because it's
apt to be a simplification.  We know an awful lot about genes, proteins,
and gene regulation; any recent molecular biology textbook would be a good
source.

--titus

Nick Maclaren <nmm1@cus.cam.ac.uk> wrote:
> In article <j9gtv9.tg.ln@linuxsys.ii01.org>,
> Robin Faichney  <robin@ii01.org> wrote:
>>
>>Don't know about "concrete work", but a serious theoretical issue is
>>this: genes are not DNA sequences, but are encoded in such sequences.
>>The decoding mechanism is the cellular machinery that interacts with
>>the DNA.  Anyone who doubts that needs to read Darwin's Dangerous Idea
>>by Dan Dennett.  And it's the gene, actually an abstraction, that carries
>>the information.

> The same is true of computer programs!  Remember the old amusement
> of writing a program that would be valid in N languages and would
> print the name of the language it was compiled as when run?

I don't actually "remember" that, but I take the point.

> Quite a few mavericks (yes, I admit to being one) have been saying
> for some time that the engineering models of computer programming
> match what is actually going on less well than the biological ones
> of breeding and so on.  That is the converse of the question that
> was being asked, but has its relevance.

Indeed.

> One of the consequences is that, as with English text, what you
> regard as the information content depends very much on at what
> level you are analysing it.  This is very true of DNA, but we do
> not yet know enough about how it works to know what levels of
> analysis are meaningful.

> So I don't think that there is a simple answer, and all answers
> will be pretty fair hand waving for a long time yet - probably
> half a century or so, though progress may speed up.

It's a deep issue.  There was a discussion of it in one of the philosophy
newsgroups a few months ago, a few of the contributions to which were
quite interesting.

-- 
"The concept of information is the key that decodes mind, matter,
meaning, consciousness..."
Robin Faichney -- inside information -- http://www.ii01.org/

Titus Brown <t@chabry.caltech.edu> wrote:
> In article <j9gtv9.tg.ln@linuxsys.ii01.org>,
> Robin Faichney  <robin@ii01.org> wrote:
>>
>>Don't know about "concrete work", but a serious theoretical issue is
>>this: genes are not DNA sequences, but are encoded in such sequences.
>>The decoding mechanism is the cellular machinery that interacts with
>>the DNA.  Anyone who doubts that needs to read Darwin's Dangerous Idea
>>by Dan Dennett.  And it's the gene, actually an abstraction, that carries
>>the information.

> Actually, the concept of gene itself is a bit tough to pin down.  Most
> molecular biologists tend to just define the gene as the protein-coding
> region, although regulatory regions -- which can be just as complicated
> as the gene -- tend to get called "genes" by geneticists, who w/o the
> sequence usually can't tell if their loci are in protein coding regions
> or not.

> I'm also not sure what you mean by abstraction: the amino acid sequence is
> very strongly tied to the genomic sequence; heck, in prokaryotes, where you
> don't have to worry about differential splicing and mRNA processing,
> it's absolutely tied to the genomic sequence.

> I wouldn't recommend Dennett for this basic stuff, if only because it's
> apt to be a simplification.  We know an awful lot about genes, proteins,
> and gene regulation; any recent molecular biology textbook would be a good
> source.

You miss the point.  This is a conceptual truth, with which philosophers
like Dennett (and myself, as an amateur) deal.  Any secondary school level
text provides all the biology we need to understand this.  DNA on its
own is not enough.  It requires the machinery of the cell to interpret
its code.  Thus, genes are abstractions, being encoded in the DNA,
as decoded by the cellular machinery.  Discussion of different kinds
of sequences is irrelevant, in this particular context -- that's just
arguing about the form of encoding, when that point is that genetic
information is encoded -- anyhow.

-- 
"The distinction between mind and matter is in the mind, not in matter."
Robin Faichney -- inside information -- http://www.ii01.org/