🚀 go-pugleaf

baez@guitar.ucr.edu (john baez) wrote:

>In article <4fd9gs$ueu@ds2.acs.ucalgary.ca> kmuldrew@acs.ucalgary.ca (Ken Muldrew) writes:
>>baez@guitar.ucr.edu (john baez) wrote:

>>>I'm mainly interested in how many bits of information needed to create
>>>a new organism are contained in the phenotype and NOT in the DNA.  This
>>>seems to be a much trickier and less-studied question.    

>>I'm still not entirely sure what you're asking, here. Perhaps if I
>>phrase it experimentally, then you could elaborate. Let's say we take
>>a two-cell embryo and split it. Then take the DNA from one cell and
>>insert this into a one-cell embryo from an outbred line (ie. from a
>>different mother). Wait for these "twins" to mature and then compare
>>phenotypes. Maybe you could try using zenogeneic (from a different
>>species) hosts for the DNA as well. 

>>Are these the sort of questions you're getting at? 

>The experiment you propose might help answer my question, but I think my
>question is very difficult and possibly not very precise as it stands.
>I don't really expect anyone to be able to answer it anytime soon, but I
>find it interesting.  

>Here's what I mean in concrete terms.  In the movie Jurassic Park the
>idea is that scientists find the DNA for dinosaurs and use it to clone
>dinosaurs.  The details are omitted.  Viewers ignorant of biology might
>think that you just hand a biologist the genome of a dinosaur and then
>they mix up a bunch of chemicals and it turns into a baby dinosaur. 
>Of course it isn't quite like that.  There is no stage in the life
>cycle of a dinosaur when all its information is in the DNA.

>So what I'm wondering is this: given the complete genome of an organism, how
>much *additional information* would you need to give someone for them to
>be able to build a living copy of that organism?  

>Here we disallow "cheating" by relying on existing organisms.  For
>example, while I'm not sure, I suppose we could create a Siamese
>cat from its genome with the aid of a one-cell embryo of a Persian cat
>and a mother cat.  (Is that actually true?  Can we remove the
>nuclear DNA of the one-cell embryo and substitute some other DNA without
>ruining things?  Are the mitochondria of Siamese cats pretty much the
>same as those of Persians?  Anyway, in my ignorance I imagine it's possible in
>principle.)  Perhaps we could even create a zebra with the aid of a
>one-cell embryo of a horse and a mother horse.  But here we are relying
>on a lot of pre-existing infrastructure... or in abstract terms, a lot
>of pre-existing *information* about how to turn a particular genetic
>code into a living organism of that type.  

Mitochondria are pretty well independent organelles. The mitochondria
of a persian cat would probably fully suffice in our own cells.
They're thought to have originated as bacteria (hence the different
genetic code) which entered into a symbiotic relationship with
eukaryotic cells. The bigger problem with mammals is that you need a
functioning mother to produce an offspring! So not only is the DNA
inadequate, but so is the the one-cell embryo.

>In other words: if we wanted to transmit instructions by radio to a
>highly intelligent alien civilization, which would enable them to create
>a human being, how many bits would the message have to be?   (Assume we
>have already set up lines of communication and can understand each
>other... actually this point is a bit subtle.)  The human genome has about
>3 billion base pairs, each of which contains 2 bits before we apply
>any data compression, so transmitting the genome takes only 6 gigabytes.
>If we omitted the "junk DNA" (but see below!) we'd only need 600 megabytes.
>But the information in the genome is not enough for the aliens to 
>build a person!  I think it's possible *in principle* to figure out the
>needed extra information someday and transmit it.  But I have little idea 
>how much more information it is.

There are those who would argue that you only need to transmit the
laws of physics, which contain the laws of chemistry, which inexorably
lead to an autocatalytic set of organic compounds given the right
conditions (liquid water, energy source, stuff like that) which will
then evolve on its own to produce life forms which, while not human,
contain much of the complexity that exists in a human (thus saving
some bandwidth). Of course if you're talking to a highly intelligent
alien civilization, then they already know most of this stuff, so what
you really need to tell them is how to make a human from what they've
already got. (I seem to be suffering now from the same sort of
confusion that I get when I think about 'communication' theory and
'information' theory--I suppose what I really need is to learn more
about both rather than dwell on my confusion about the difference).

>For example, say we tried to transmit intructions on how to create a
>one-cell embryo, together with instructions on how this one-cell embryo
>was supposed to be dealt with in order to get a baby organism.  Then
>we'd have to give a lot of information about what the cell is like,
>enough to be able to build a cell *from scratch*.  For all I know the
>amount of information required for this is vastly more than the
>information in the nuclear DNA.  After all, the nuclear DNA never builds
>a new cell from scratch; only with the help of an existing cell.

This will be a lot easier if we restrict ourselves to simple bacteria.
There's no organelles, no non-coding DNA, and sometimes you even have
overlapping genes (built in data compression).

>>>It is a subject of debate as to whether these are really useless "selfish
>>>DNA" whose only job is to reproduce itself, or whether it has an
>>>undiscovered function in the organism.  I think people have estimates at
>>>to what percent of DNA is this "junk DNA".

>>It's pretty highly conserved, so calling it "junk DNA" is probably
>>just due to the hubris of biologists, IMHO. 

>I'd love to know more about that.  What are some hypotheses for what its
>role might be?

The name 'junk DNA' probably comes from the hypothesis that it is
parasitic--that it simply got there due to duplication accidents and
then continues to be replicated with the rest. If you think of genes
as 'selfish' (ala Richard Dawkins), then you might even think that
these sequences evolve ever more sophisticated mechanisms by which
they can duplicate themselves and assure their survival.

As for hypotheses for what the role of non-coding DNA might be, I'm a
little short on knowledge here. I'll just briefly give an overview of
some of the ideas that I'm aware of and hope that some experts jump
in.

First is the structural component. The human genome is *enormous*.
There is some sophisticated house-keeping required just to pack it
into the nucleus of a cell, let alone maintain it in perfect
condition. There are a bunch of proteins involved in supercoiling the
DNA called histones. It's thought that some of the non-coding DNA is
used to interact with the histones for efficient packing.

There's also the matter of regulatory sequences--specific sequences
which are recognized by the proteins which turn gene transcription on
and off. Of course, this is a very small portion of the non-coding
DNA.

Somatic cell mutation usually uses a strategy in which there are
multiple copies of an exon in between the neighboring exons. OK,
that's probably a little too much jargon, let me try to explain it in
simpler terms. Eukaryotic genes are not continuous on the DNA, there
are coding parts (the exons) and non-coding parts (the introns) which
are all transcribed (copied to RNA) whence the introns are then
cleaved out. Let's say you want to make antibody molecules with about
a billion different recognition sites. It's going to be a bit much to
stick a billion genes into your genome, so what you do is cut the gene
into pieces, make several copies of each piece which are all slightly
different. Let's say you have three pieces (A, B & C), each of which
has two copies. You transcribe the whole business to RNA and then cut
it down to make a gene with only one A, one B and one C. Depending on
which copy of each piece you cut out, you can make 8 different genes.
This is commonly done in eukaryotes. The information as to where to
cut and splice these sequences has to be coded somewhere and the
introns just happen to be at the right place at the right time. Since
RNA is single stranded, it can (and probably does) fold in on itself
to form hydrogen bonds. The introns probably contain sequences which
allow for specific tertiary structures to form which are then
recognized by splicing proteins (look at a picture of a tRNA to get an
idea of what I'm talking about here).

Another idea is that the DNA replication machinery was built to handle
genomes which are much smaller than in eukaryotic organisms. Thus the
error rate which was built into the system (it's the same machinery
for all living things) was designed to introduce novelty at a rate
which was consistent with a smaller genome. There will be a certain
critical size for the genome, beyond which, the errors introduced by
copying will corrupt the genome too quickly (ie. there will be too
many errors introduced into the genome to make a viable offspring).
Thus the non-coding DNA is there to absorb mutations. If a point
mutation occurs every 10^9 bases (I think the copying errors are about
1 in 10^6, but there are proof-reading enzymes which improve this),
then you spread your gene out so that maybe you only get half the
mutation rate within the coding region.

There are also huge chunks of DNA which have repeated sequences of a
short oligonucleotide (maybe 4 or 5 base pairs, repeated over and
over). These sequences undergo duplications and deletions are a much
higher rate than aperiodic DNA (due to the way duplication occurs).
It's thought that these sequences might be there specifically to
introduce duplication, especially of coding regions. This is one way
to develop redundancy, which is necessary for evolution.

I could speculate more, but I've probably dug myself in too deep
already. I'm interested in this stuff, so any corrections or
information based on more current knowledge of this subject would be
welcomed.

Ken Muldrew
kmuldrew@acs.ucalgary.ca