Re: The last ancestor of all life


"Seanpit" <seanpitnospam@xxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message news:1156201518.480779.207870@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Perplexed in Peoria wrote:

The individual events are indeed independent events. However,
realizing a specific *sequence* of variables is *dependent* upon the
specific order in which the variables are realized. That's clearly
been my whole position this entire time. I'm talking about the odds
that a specific sequence will exist in a given genome. Those odds are
calculated by multiplying the number of potential variables per
position by the number of positions in the sequence in question - as in
1 in 20^100 for a specific 100-character residue sequence.

I hadn't intended to get dragged into the meat of this
controversy. My original intention was simply to point
out that you were wrong on the technicalities and technical
terminology of probability theory. That is a particularly
dumb thing to do when you do it in a paragraph laden with
sarcastic phrases like "Please do explain ...". You should
avoid doing that. It loses credibility.

But since you seem to want to persist in challenging R Norman
on the substance, let us look into the question as to whether
multiplying probabilities is the way to go here.

Ok, we have a number of proteins involved here, all of which
are complex, all of which do something, and all of which must
work in concert to do some bigger thing? Are we in agreement
so far?

Sure . . .

Call the proteins A, B, C, etc. Call the 'probability of
the proteins' P(A), P(B), P(C), etc. We will get to what
these probabilities actually represent in a moment. What
you want to claim is that the probability of the whole
complex structure should be calculated as P(A) times P(B)
times P(C) etc. What I want to claim is that this algorithm
is wrong and that the real probability is much higher than
the one you are computing. Are we in agreement so far?

I think you are going a little off base here. But, I'll go into that a
bit more later on.

Ok, now what do we mean by these probabilities like P(A).
You want to claim that the right definition for P(A) is
the probability that some random sequence of amino acids
would result in a protein that that does the same thing
that protein A does.

That's true . . .

You will concede that there may
be many amino acid sequences that are functionally
equivalent to the 'true' A sequence. There is a certain
amount of slack allowed. You will concede this because
you know that evolutionists can point out many examples
of proteins where the sequence is slightly different from
one species to another, yet the two distinct proteins do
essentially the same thing. You are a reasonable man.
Yet you will insist that while there is some 'slack' in
a protein's sequence specification, there isn't enough
slack to make much of a dent in your argument. Are we
in agreement so far?

Yep . . .

Ok, since you were so gracious in conceding that, I
will also be gracious. I will accept your definition of
P(A). I will concede that it is a small number. But
I haven't conceded yet that we should just go and
multiply P(A) times P(B) etc. First I want to look
more closely at P(A) in a rather unusual way.

Ok . . .

Ignoring the 'slack' for the time being, you would say
that P(A) is itself a product - it is (1/20)^N where
N is the number of amino acids in the sequence and
the factor (1/20) is the chance that a 'random' amino
acid would be 'right'.

Right . . .

Ok, I concede that, if one
ignores the fact that different amino acids have
different frequencies. It is not exactly right, but
close enough for bio-theology work.

Oh come on now . . . Do you really think this matters much as far as
the main point is concerned?

But I want to break down P(A) into factors in a
different way. I want to define

P(A) = pAf * pAs * pAc
where
pAf is the probability that a random amino acid
sequence of the same length as A will fold
into a (mostly) stable shape. (Most random
sequences don't do that, you know.)
pAs is the probability that a random amino acid
sequence of the same length as A which folds
stably will also do something interesting.
(Most random folded sequences just sit there
you know.)
pAc is the probability that a random amino acid
sequence of the same length as A which both
folds stably and does something interesting
will do the correct thing - the same thing
that A does.

Do you understand and accept these definitions?

I think so . . .

Ok, now to continue. You want to claim that
the right thing to do is to multiply out

P(A) * P(B) * etc.

which is the same thing as

(pAf * pAs * pAc) * (pBf * pBs * pBc) * etc.

Now, you have probably heard before that multiplying
is wrong because natural selection is involved and
that therefore pAc and pBc are not totally independent
probabilities. You may even be willing to concede
this because you think that independence still reigns
at the level of pAf vs pBf and of pAs vs pBs. Well,
I have to disagree. Because all of A, B, C, etc.
are descended from earlier proteins that were selected
by natural selection to fold and to do something
useful.

That's right . . .

And since gene duplication and divergence is
the way that evolution builds large proteins, by
the time evolution got around to building large
complex structures like flagella and cillia, there
were plenty of sequences around that already folded
stably and did something.

Correct . . .

All that was required was
to duplicate some of them (an easy, almost trivial
kind of mutation) and then to set to work changing
what they did into the new, correct thing to do.

That's right . . .

Your claim of independence fails at all levels.
Are we in agreement so far?

You seem to me, at this point, to be confusing ratio with starting
point. The ratio is still what it is. It seems to me that you are
simply suggesting that the starting points are much closer to the
targets than I seem to realize. Correct? Not nearly as many random
steps need to be taken as I seem to be suggesting - right?

Yep. You got it exactly.

Of course, I've argued that gene duplication does not make novel
starting points. It only makes more of the same starting point. You
are still left with having to cross a gap that includes all potential
sequences - even the ones that won't fold to make much of anything.
This might be taken to mean that the random walk/sampling required is
not significantly reduced.

But there aren't gaps - whole impassable channels - composed of
sequences that won't fold. Sewell Wright has confused generations
with his metaphor of a 'fitness landscape'. Read the recent book
by Sergei Gavrilets and reset your intuitions on this point.

What you seem to be suggesting is that gene duplication does nothing
more than add more paper. The writing on that paper is totally
useless in terms of producing another page for 'the book' by mutation
and natural selection. The paper might as well have been blank to
begin with. Just not so. The page contains whole words which merely
need to be rearranged to form whole new sentences.

Unless, unless of course the genome already has the needed part to make
the next step in the genome pre-formed. Your argument seems to suggest
that the odds that this will in fact be the case are pretty good! -
since many of the non-workable potential sequences are screened out
ahead of time by default. And, this is in fact true at lower levels of
functional complexity. However, at higher and higher levels, the size
of the step to the next novel function grows linearly. It becomes
harder and harder to find a pre-existing sequence that would cross this
step in one fell swoop with one single mutation event - such as a
duplication and insertion of just the right sequence into the correct
spot. The problem is that the odds that the needed sequence that would
fill the gap actually exists preformed in the same genome decrease
exponentially as well with each step up the ladder.

I think that Dawkins's "weasel' game has misled you into how evolution
actually works. Either that, or you have been misled by that "in His
image and likeness" thing. There is no predetermined end point to
be reached. Instead there is a lot of stumbling around as proteins
first take one role all the time (doing X, say), then gradually are
modified to do X only if Y is present in the environment (call it Y -> X)
and then further mutate into something like (X -> ~Y). It is not
a straight line path from something that doesn't work properly to
something that does.

If you want to use a specific example, please do explain a specific
functionally beneficial step in the evolution of the flagellar system
from some proto form.

I'm pretty sure that one has already been sketched - for example,
at the t.o. website.

Or, provide any actual experimental
demonstration of the evolution of any new function that require more
than a few thousand fairly specified bp of DNA.

Hmmm. Let's see. That would require that I bio-engineer something
like (say) 4000 different strains of bacteria - each containing
proto-flagelar proteins doing something useful, with the first not
having anything like a flagellum and the last with a fully functional
flagellum. And when I line up these 4000 petri dishes in a row,
each differs from its predecessor by only a single mutation. Is
that what you are asking for?

Well, I'm an armchair sort of 'scientist'. I don't have the technical
expertise to make that demonstration. But you are welcome to try
the counter-demonstration. As I understand it, that would involve
some 4^4000 petri dishes plus experiments showing that every path
from starting point to end point sinks you deep into a fitness
'channel' at some point.

Not interested in trying that? Well, I guess we will have to
just snipe at each other from armchairs then. But I hope I have
given some sense of why your sniping isn't convincing anyone.

It is true that all the parts needed to form a flagellar system of
motility exist in all bacteria. The question is, how does one get
these parts together in just the right way? The usable chucks just
aren't big enough. That's the problem. I mean, there might be 30aa that
would work in one place and another 42aa that would work in another
place and another 12aa that would work over there etc. How does one
get all of these small chunks together to form the next useful bigger
chunk system?

In considering this problem, one must also consider that multicharacter
mutations are much less common than point mutations or mutations
involving relatively short sequences. Getting a specific mutation of,
say, exactly 50 characters to be inserted into a specific spot is
extraordinarily rare regardless of if no specific 50 character sequence
is required - just 50 characters in any order.

Also, let's say a particular step requires a specific insertion into a
specific spot of a series of codons that would code for a fairly
specified 100aa sequence - with a ratio of about 1e-40? (see
calculations by Yockey and Sauer) What are the odds that this specific
sequence exists, preformed, somewhere in a genome of only 5-10 million
bp? Then multiply this times the odds of this specific sequence
getting copied out of its current location and pasted into the specific
needed location. How much time does this require - on average? What if
the gap is 100aa?

Now, I suspect that you may be tempted to bring
up abiogenesis at this point. You might wish
to claim that my arguments here don't apply to
abiogenesis. And, you know what? You would be
right. But my arguments do apply to cilia and
flagella and clotting cascades and those kind of
things which no one thinks were invented yet at
the time of abiogenesis.

Forget about abiogenesis. I'm only interested in evolutionary
potential given the starting point of a living creature that can
reproduce itself. I really don't think you've solved the problem. All
your theory does is create little islands that happen to link up with
each other like a sticky bubble gum. This does make evolution quite a
bit easier at very low levels - where pre-formed chucks of this can
combine with pre-formed chunks of that to make a novel functional
system. However, at higher and higher levels of minimum size and
specificity requirements, the odds that the right pre-formed chuck will
actually exist anywhere in the genome to get you to the next step in
the pathway toward any higher level system drop off exponentially - and
you are back to where you were to start with.

And I think that you are trying to scale "methinks it is a weasel" up
to a monkey typing "Hamlet". Evolution just doesn't work that way.

We can talk about abiogenesis some other day. You
can even talk some more about cilia and flagella.
I certainly haven't completely explained them here.
All I have done is to show that simple multiplication
of probabilities is not the way to approach these
kinds of evolved 'IC'.

I really don't think you've done what you think you've done at all.
The islands of potentially beneficial sequences simply become more and
more stretched out at higher and higher levels of functional complexity
until the bridges between them simply snap apart and they become truly
isolated in a very remote way. Now, no single multicharacter
insertion/deletion/translocation etc is going to get you there.
Multiple mutations involving multiple characters each are required.
And, multiplication of the odds is back . . . and at very low levels.

It must be very frustrating for you when knowlegeable people don't
share your intuitions. Hell, I know it is. It is frustrating to
me in the abiogenesis field where everyone else seems to be stuck
on "First you create the building blocks, then you join the blocks
together".

.

Relevant Pages

  • Re: Rational numbers, irrational numbers: each dense in real numbers
    ... that it is extremely unlikely to have the sequence ... Such a probability distribution, ... probabilities of all but a finite or countably-infinite ... Any uniform nested-finite-distribution system which is ...
    (sci.math)
  • Re: Howard theory - "cannot be computed"
    ... in fact require some sort of probability calculation. ... the expectation of chance alone. ... but some of the differences may arise rapidly by selection for changes ... that the current sequence retains or changed. ...
    (talk.origins)
  • Re: The last ancestor of all life
    ... that a specific sequence will exist in a given genome. ... terminology of probability theory. ... multiplying probabilities is the way to go here. ... However, at higher and higher levels, the size ...
    (talk.origins)
  • Re: The last ancestor of all life
    ... what is the probability of flipping a quarter and landing ... What is the probability of flipping a quarter having heads on the ... Any specific sequence of heads and tails will have the same odds. ... multiplying the odds of the individual *dependent* events. ...
    (talk.origins)
  • Re: Linfords Fallacy
    ... it is not true that the probability of having all ... heads is the same as the probability as having any mixture of heads and ... or tails than there are series of coin flips with only heads. ... It is true that the probabiliy of any sequence is the same ...
    (sci.math)