Re: The Relationship of Gaps to Thresholds

On Dec 7, 1:21 pm, hersheyh <hershe...@xxxxxxxxx> wrote:
On Dec 6, 6:47 pm, Seanpit <seanpitnos...@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
wrote:

On Dec 6, 9:24 am, hersheyh <hershe...@xxxxxxxxx> wrote:

On Dec 5, 2:07 pm, Seanpit <seanpitnos...@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
wrote:

On Dec 5, 7:23 am, hersheyh <hershe...@xxxxxxxxx> wrote:

[snip]

and never strays far from those pre-existing *functional* sequences
wrt number of mutational steps.

That's not quite right. Random changes/mutations can actually leap
across vast distances in sequence/structure space in a single
bound.

No. Organisms can form hybrid proteins, deletion products,
endoduplications, duplications, and can change sequence within the
constraints of selection.

Not true. Selection doesn't happen until after the mutation occurs.

That is not what I meant. I meant that selection constrains retention
and works against retention of changes that eliminate function. In
particular, your model assumes that there is some evolutionary
mechanism for retaining *proteins* that lack function.

Where did you get that from anything I've said? I never said any such
thing. Natural selection, which is a very real force of nature, tends
to get rid of proteins that lack beneficial function. I've NEVER said
otherwise. Why do you keep coming up with these strawmen
misrepresentations when you know that such statements go completely
counter to what I've pointed out to you many times?

There isn't.
In the absence of functional utility, there is no selective advantage
to making the protein encoded by the sequence. Such sequences tend to
be eliminated or truncated and cease being translated. Thus, if a
pathway exists where the first step involves producing a non-
functional protein, there will generally be a limited number of
further mutational steps.

You mean there will be a limited number of further *retained*
mutational steps. There will NOT be a significant limit when it comes
to non-retained mutational steps however. It is just that the vast
majority of mutational steps are non-beneficial and will be discarded.

So any sequence that requires a long chain
of mutational steps away from all other functional proteins of that
size is 'forbidden'.

That's not true. Any sequences in sequence space can be reached by
random mutations.

< snip more confusion >

That is why mutations can actually jump anywhere in sequence/structure
space. Of course, most of these mutations will end up in places that
are not compatible with any sort of usefulness or even toleration and
are therefore rapidly discarded by function-based selection.

Which means that any sequences that require a long chain of mutational
steps will not be found by any real mutational process.

No sequence necessarily requires a long chain of mutational steps to
find. A far distant sequence could be hit upon in a single or a small
handful of mutational steps. It is just unlikely that such a sequence
would also have a high minimum structural threshold requirement is
all.

However,
this does not mean that a portion of sequence space has not actually
been searched just because it wasn't beneficial and was in fact
detrimental.

If sequence space is as you describe it, where there must be a long
chain of *functionless* intermediates between sequences with any
function, then huge swaths of this *functionless* region can never be
reached because there is no selection for retention of protein
synthesis before deletion or mutation will eliminate the sequence as a
coding sequence.

You forget about multicharacter indel-type mutations and neutral
evolution. Mutations can make large leaps across sequence space in
any direction and into any location starting from any starting
point(s).

But they cannot produce random sequences
that are wildly different from existing sequences. Even the chimeric
products have sequences that are related to *long stretches* of the
proteins they are derived from. But no single step will leap across
vast distances to produce any old random sequence. Rather one gets
modifications of pre-existing sequence or sequences which still have
some of the functionalities of the old sequence.

Every hear of neutral evolution? Long stretches of DNA are thought to
be pretty much free of natural selection in various gene pools. These
stretches can indeed mutate pretty much at random to just about
anything - exploring all over the place in sequence space. I'm afraid
you are mistaken when it comes to your notion of significantly limited
exploring potential.

But, of course, your model is not one which involves or uses "neutral
evolution".

Again, where on Earth did you get that idea? Of course my model uses
neutral evolution! I've specifically said so many times - to you
personally. I've certainly never said otherwise. This is yet another
deliberate mischaracterization on your part.

Neutral evolution occurs when a change in sequence does
not produce any significant change in *function*. In neutral
evolution, every step produce a new sequence and the *same*
function.

That's right - the same function OR - - the same non-function when it
comes to genetic sequences that do not yet produce or significantly
affect protein production.

Your model, OTOH, involves, as a first step, the generation of a
completely *functionless* protein.

What?! Where did I ever say that? I've never said any such thing -
ever! My model starts with fully functional protein systems - just
like your model. I've specifically pointed this out to you many many
times.

That would generally produce a
protein that is selected against (if the unmutated protein has a
beneficial function in the environment that the organism is in), not
one that is selectively neutral.

Of course . . . Again, I've never said otherwise.

Lets say you flip a coin a trillion times and get heads only twice.
How many flips, on average, would you suggest it would take to get
heads again?

That's just it. You claim that is what happens, despite the fact that
it is impossible to *actually* flip a coin a trillion times.

It is not impossible for a trillion members of a population all
flipping a hundreds of coins ever 20 or 30 minutes. That's what
happens with genetic evolution you know.

Sean! Can't you keep your argument straight? In the rest of your
argument, you claim that it is impossible for the random process you
describe as the mechanism (the one that is equivalent to randomly
synthesizing a protein of 30% of the total length of any functional
protein) if you include all the organisms on the planet and trillions
of years of trials. That implies that you think that it is impossible
to search total sequence space, even for proteins of a 4-5 dozen aa
length. That clearly means that you think total sequence space has
NOT been searched and CANNOT be searched in the time available. Now
you are saying that not only can it be searched, it has been. What
gives?

How many flips of the coin do you think a colony of a trillion
bacteria can achieve in 4 or 5 billion years? Hmmmm? The answer is
not nearly enough to cross the gap between anything that exists in the
gene pool to any potentially beneficial system in sequence/structure
space that requires a minimum of over 1000 fairly specified amino acid
residues. Do the math for once and see. It is impossible for a
colony of even a trillion individuals to search even a tiny fraction
of the sequence space of 50-60 residue changes in trillions of years.

BTW, this is a real conundrum for you. If it is, in fact,
impossible for all the organisms that have ever lived to *actually*
search total sequence space, then your calculation of the ratio of
cytochrome c sequences to total sequence space is as meaningful and
useful as the calculation of the ratio of the area covered by
Starbucks outlets divided by the surface area of all the planets in
our solar system. Probability is calculated as the ratio of observed
events to the actual number of trials, not the ratio of observed
events to all the possible trials that could possibly exist.

You have this strange notion that potentially beneficial targets are
all clustered together into one tiny corner of sequence space - like
Starbucks clustered on one tiny planet when the rest of the universe
has no Starbucks (yet). That notion is simply ludicrous when it comes
to language/information systems. Potentially beneficial targets are
NOT clustered together into one tiny corner of sequence space. For
each level of sequence space, they stretch in loosely clustered
islands from one side to the other. At higher levels, the loosely
clustered islands become less and less clustered. This drifting apart
creates a linearly expanding gap between those islands that have
actually been discovered and those that might be beneficial if they
were ever discovered. The MINIMUM gap distance also increases in a
linear manner with the increase in the average gap distance.

If, as you claim (and I don't doubt it), it would be impossible for
all of life to search total sequence space in the time span that the
universe existed (or in trillions of years), that means only that the
*actual* number of sequences searched is NOT the "total sequence
space" number you use.

I never said that all of sequence space had to be searched to find a
higher-level system. All that has to be searched is the space between
what already exists and the next closest higher-level system before it
is likely to be found.

Again, that is not what your math is based on. Your math that you
used to calculate the "average gap size" is based on the ratio of
cytochrome c-related sequences to total sequence space. Are you now
saying that your "average gap space" calculation is a meaningless
calculation?

Not at all. The minimum gap distance is very much related to the
average gap distance. As the average gap distance increases, the
likely minimum gap distance also increases - along a Poisson
distribution.

And that the *real* determinant of how a function arises
at any level system is whether or not there is a sequence that can be
modified in a few steps to produce that new function. That, however,
means that "average gap size" is irrelevant.

You wouldn't say this if you knew anything about statistical
averages. This all important minimum distance of a "few steps" you
keep talking about is very much related to the average distance. As
the average distance increases exponentially, the likely minimum
distance does not stay the same - i.e., it does not stay at a "few
steps" anymore. It also increases as the average distance increases.
Again, the minimum possible distance isn't the same thing as the
likely minimum.

< snip >

Searching all of sequence space is not
required here. That's not at all correct. The problem is that the
smaller space between what exists and the next closest potentially
beneficial target that might exist is huge.

So you assert. But if that smaller space of sequences *actually*
searched is the real number, how do you calculate it and why do you
use a number that is not relevant?

The big question here is "if". If the smaller space actually is the
real gap distance, like your often suggested minimum possible distance
of one, then it will be crossed in short order. The question is, what
are the odds that this tiny distance is actually the real distance?
You simply assume that this small distance was there for everything
that currently exists in all biosystems. You do this without any
sequence comparisons that demonstrate this small gap distance beyond
the 1000 minimum structure threshold requirements. You also have no
other genetic evidence or mathematical evidence. You certainly don't
have any real life experimental evidence beyond starting with a fully
functional system, messing it up a bit, and then getting the same type
of function back again. You think this is the same as demonstrating
how evolution could have happened when the function in question was
not yet in existence? Think again.

Searching just this
fraction of sequence space is what would take trillions upon trillions
of years of time beyond the 1000aa level. The 1000aa level is not the
gap distance.

I never claimed it did.

You have claimed this so many times as a strawman I've lost count.
Have you finally given up on at least this particular strawman
mischaracterization?

I kept asking you for the number that is
relevant: "average gap size". That number, for your 1000 aa level
(keeping in mind that most functional proteins are about 300 aa in
length, which poses some interesting comparisons) determined there to
be an "average gap size" of 300 aa's.

We aren't talking about most functional single-proteins. We are
talking about protein-based systems that do happen to have minimum
structural threshold requirements greater than 1000 fairly specified
amino acid residues. Many protein-based systems have much lower
threshold requirements. You do understand this concept - right? It
is just that those that do in fact have higher minimum threshold
requirements are much harder to evolve. That's the whole point.

Beyond this, those rare systems that do have fairly specified 1000aa
sequences (with a 30% specificity) are indeed separated from each
other by an average of 300aa residue differences. That average gap
distance translates into a likely minimum distance, even relative to a
large gene pool, of dozens of residue differences.

And your math is identical to
what one would expect if one had to create a 300 aa protein of a fully
specified sequence (only one specific aa at each site) with all other
sequences being *functionless*. Given that most known functional
proteins are 300 aa's in length, it would think it difficult not to
produce at least a few other intermediate functional sequences during
this process.

You would produce intermediate functional systems - just not ones that
have the higher 1000 fairly specified residue threshold minimum
requirements. Coming up with lower-level functions as part of larger
sequences doesn't really help one find novel higher-level targets.
There is simply no significant statistical advantage to finding lower-
level steppingstones when it comes to helping a population find higher-
level steppingstones.

What it does do is produce a minimum gap distance of
dozens of needed changes. This minimum gap distance of dozens of
changes has its own space that is actually enormous. This gap space
is what needs to be searched. A search of such a gap space would
indeed require trillions upon trillions of years for even a population
the size of all the bacteria on Earth (i.e., about 1e35 organisms).

This is the key concept in this discussion and you didn't even
acknowledge it?

I just did. By pointing out that if that is true, then the amount of
total sequence space that has been *actually* searched is much smaller
than the number you used. And that means that the "average gap size"
calculation is as meaningful as a calculation of "average distance"
between any person and a Starbucks that was based on calculating the
area of all the Starbucks as a fraction of the area of all the planets
in our solar system.

Your notion of the nature of sequence space would only be correct if
potentially beneficial targets where in fact as clustered as your
Starbucks scenario into one tiny corner of sequence/structure space.
Unfortunately for you, that notion doesn't represent what is actually
known about reality.

Moreover, it does not come close to meeting the real objection to your
numerology. Which is that those features that *have* evolved or
appeared represent those features that *can* be reached by a stepwise
series of short random walks between useful structures from pre-
existing structures that exist in some organism.

Actually no. Those systems that do exist which require more than 1000
fairly specified amino acids at minimum cannot be reached by anything
else that exists in the gene pool. The distance between a higher-
level system and the next closest homologous system is dozens of amino
acid changes away.

That is your assertion. You only back it up by a calculation of
"average gap size" that appears to be irrelevant.

It is your bald assertion that for existing systems the likely minimum
distance was only one or two character changes or mutations wide. You
make this assertion with absolutely no genetic-based evidence
whatsoever. It is complete hand waving and smoke blowing, but nothing
else. The actual evidence available supports my position that the
MINIMUM gap distance does in fact increase, in a linear manner, with
each increase in the minimum structural threshold requirements.

Your claim to the contrary is built on fantasy.

Let's see. I give examples of proteins that *did* evolve and in each
case

You mean you've given me examples of systems that you assert evolved.
None of your examples beyond the 1000 threshold level have actually
been shown to evolve in real time when it comes to producing a truly
novel functional system that wasn't already there to begin with.

there is a clearly reasonable pathway to evolve the *function*
from sequences that served different *functions* and could reasonably
be present (as ancestral proteins) in some ancestral real organism.

Not even remotely true beyond the 1000aa threshold. Your fantasy
scenarios are based on the notion that what is possible is also
likely. There is in fact a huge difference between what is possible
and what is at all likely. Your scenarios, while certainly possible,
aren't even remotely close to being likely this side of trillions upon
trillions of years of time. Because of this, they are basically
nothing more relevant than fantasy - than extraordinarily wishful
thinking equivalent to blind faith of the most fundamental degree.

That the *modern* proteins may differ at dozens of sites is
irrelevant, since most of these differences are due to neutral drift
that did not affect *function*.

Yet another bald completely baseless assertion. If more than a
handful of differences are truly neutral in the modern comparisons, it
should be a piece of cake to get the modern equivalent to evolve into
the modern high-level systems. This has not been done. Why not?

Some of the sequence changes are
*functionally relevant*, but only in the quantitative sense of
optimizing a function that already exists.

This is complete nonsense. While some features are most certainly
quantitative in nature, the minimum differences needed still leave
dozens of qualitative residue differences between what exists and the
next closest homologous system with a different type of function. That
is why Kenneth Miller loves to use the TTSS system. What is
interesting about the TTSS system, vs. the flagellum, is that the TTSS
system requires only about 10 proteins. Compare this to the flagellar
motility system and its minimum of about 40 proteins. That's a
gigantic gap. Of course, those like Matzke try to do a bit better,
but no one has gotten remotely close to demonstrating your desired
minimum qualitative gap distance of one or two differences. That's
completely counter to the available evidence.

What is important is
whether or not, in some *ancestral* sequence, there be either the
modified *function* already present as a minor or secondary *function*
(duplication and subsequent specialization or divergence is a *major*
evolutionary mechanism). What is important is whether pre-existing
*functions* can be linked into some chimeric or multimeric structure
that now, by virtue of the pre-existing *functions* being in a
different relationship. None of this is a function of the size of the
end product. None of these methods involve *average gap size* at all.

What involves the "average gap size" is the likelihood that these
imaginary proto-systems of yours would actually exist in the form in
which you need them to exist for your theory to work. That is where
the average gap distance comes into play suggesting that your needed
minimum gap distance isn't remotely likely to have ever actually
existed at such levels of functional complexity.

< snip >

You misquote me. The ratio is not the minimum gap size. If you will
note, I specifically pointed out that the 308 residue differences was
the average gap distance for the type of threshold used in my example
(i.e., 1000 fairly specified residues). This average gap distance is
NOT the gap distance of importance. The important gap distance is the
minimum gap distance. I specifically suggested using a Poisson
distribution to determine the likely minimum gap distance given the
stated ratio. I specifically suggested that the likely *minimum* gap
distance would be no less than 50 in this case. You did read this,
but you choose to leave this part out and misrepresent me instead I
suppose.

So now the "average gap distance" is NOT the gap distance of
importance? The gap distance of importance, then, has nothing to do
with the "average". It has to do with what pre-existing ancestral
systems existed. Then why bother calculating it at all?

The average gap distance, while not itself the important gap distance,
has a great deal to do with the important gap distance - i.e., the
minimum gap distance. The greater the average gap distance, the less
the likelihood that the minimum gap distance will not have increased
as well. This is a relatively simple mathematical concept. Go look
it up.

Again, Sean, the *minimum* gap distance is ALWAYS one.

LOL - The minimum *possible* gap distance is always one. Again, this
doesn't mean that the minimum likely gap distance is always one as
well. That definitely isn't true.

Any other
number is merely a number you pulled out arbitrarily between the
minimum and the "average gap distance" (although I would consider your
"average" more of a *maximum* gap distance).

The maximum possible gap distance is always the size of the protein in
question. However, the maximum possible gap distance is also not
likely to be the actual minimum gap distance. The same thing is true
of the average gap distance. The actual minimum gap distance is
likely to be less than both the maximum possible and the average gap
distances while also being greater than the minimum possible gap
distance - along a Poisson distribution. I know this is higher math
for you, but think about it for a while.


The lactase system, by comparison, requires a threshold minimum of
about 380 fairly loosely specified residues. The average gap distance
in this case would be less than 100

By your math it would be closer to 114.

Not for a system as loosely specified as the lactase enzyme is.
Lactase doesn't require the same degree of specificity as does
CytoC.

and the likely minimum gap
distance, relative to a sizable bacterial gene pool, would be one or
two amino acid residue changes wide.

Why would the size of the gene pool matter?

The greater the number of starting points, the greater the odds of
having a closer sequence to a potential target. For example, the odds
of one person being within 100 meters of a Starbucks aren't as good as
the odds of at least one person being within 100 meters of a Starbucks
given a population of a few million.

Yet, for a protein of 1000, where the average gap size would be about
300, you claim that the minimum [sic] gap size would be 4-5 dozen?
For an average gap size of 100, the minimum gap is 1-2, yet for an
average gap size only three times larger, the minimum [sic] gap size
is now 50-65? Can you explain?

Sure. Lactase has much lower specificity requirements.

Of course, if all you are doing is
pulling these numbers out of yer arse, that certainly could explain
it.

The difference in specificity between systems like CytoC and lactase
are well-known.

This in fact turns out to be the
case. Such a small minimum gap distance can easily be crossed very
quickly in less than a handful of generations.

Whose generations? The bacteria's or yours?

Come on now - we've been talking about bacterial generations all along
here.

Take a look at the *function* of rotary motility (bacterial
flagella). When I look at the bacterial flagella, I see it as
composed of the linkage between two subsystems, both of which have
independent functional utility but no rotary motility function.
Specifically, both types of rotary flagella in bacteria represent the
linkage of a stator subsystem (the motor) and a pore designed to
export proteins from inside the bacteria to outside the bacteria.
Both these subsystems are functionally useful and exist even in
bacteria without rotary motility. Moreover, in the case of the
eubacterial flagella, the two subsystems are linked by a single
protein, FliG, which (among other functions) has an amino end that
binds tightly to the pore subsystem of the eubacterial flagella *and*
has a carboxy end that interacts with the motor subsystem. I have
pointed out that generating the new *function* of rotary motility (by
a 'novel' mechanism that the real eubacterial flagella does not use)
from a model system only requires a deletion that forms a hybrid
protein. Thus the *mutational gap* between a particular *functional*
state, with two systems having the functions of exporting proteins and
causing motion as a result of a proton gradient and a *functional*
state in which these two subsystems are linked in a way that causes
the action of the motor to rotate the pore (rotary motility) is one
mutational step.

Pity this experiment has never been done aside from using parts that
already existed as parts of a fully formed flagellar motility system.

Why does that matter?

It matters a great deal if you want to suggest that your fantasy world
might actually represent true reality.

After all, the idea for evolution is that the
ancestral sequence is not some random sequence, but one which can be
converted to the new function. Now you whine that the sequence is too
close to one that *functions* as a flagella?

It isn't just close, it was originally part of the fully operational
flagellum. Hello! If you want to support your contention that the
proto-forms were likely as close, why not find some subsystem that
actually exists that is remotely close enough to get you across any of
your proposed steppingstones in your proposed pathway toward flagellar
evolution? It is because such systems do not exist that you resort to
using parts that actually came from the fully formed flagellar system
itself to evolve this same system back again. That's just classic.

< snip rest of replacing reality with fantasy - suggesting that what
is theoretically possible is actually likely this side of trillions of
years of time >

Sean Pitman
www.DetectingDesign.com

.

Relevant Pages

  • Re: Experimental basis for the Non-Beneficial Gap Problem
    ... unless you think evolution starts from some random sequence maximally ... It is never at the maximum possible distance - ... greater, on average, compared to a collection of smaller proteins." ... If an amino acid is entirely free to vary, it doesn't contribute to any gap size. ...
    (talk.origins)
  • Re: Richard says Howard Hershey is wrong
    ... I didn't say that the sequence specificity was the same. ... *all* such proteins. ... average distance between a given sequence in sequence space and CytoC ...
    (talk.origins)
  • Re: Felsenstein v. Dembski
    ... the Hamming distance between: ... How do you define a "valid" vs. an "invalid" sequence? ... proceeded if the lengths of all proteins had to remain constant. ... matters is the location of potential target sequences (i.e., ...
    (talk.origins)
  • Re: Pitman numerology
    ... sequence space to end up in any corner of it in one shot. ... You don't seem to understand the concept of Euclidean distance. ... The total ratio between useful proteins between all possible proteins ... close (in mutation distance) to the current available genotypes. ...
    (talk.origins)
  • Re: Experimental basis for the Non-Beneficial Gap Problem
    ... unless you think evolution starts from some random sequence maximally ... distance is always smaller than the minimum structural threshold ... greater, on average, compared to a collection of smaller proteins." ... The maximum gap size for a 100aa system is 100aa differences. ...
    (talk.origins)