Re: The Relationship of Gaps to Thresholds

On Dec 5, 1:32 pm, Seanpit <seanpitnos...@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
wrote:
On Dec 4, 9:34 am, hersheyh <hershe...@xxxxxxxxx> wrote:



It is your position that everything evolved. The argument is over
that position. You cannot assume the truth of the very position that
is being questioned. You have to come up with some sort of evidence to
support this position. You can't just assume it as a given in this
particular discussion.

Yes. *If* my mechanism is correct and new functions arise by
modification of old structures, I would expect, especially for
recently evolved 'new' functions, that there should be substantial
sequence similarity between the old structure from whence it sprang
and the new 'function' that it is derived from.

No one is saying that evolution must work otherwise than by modifying
pre-existing or "old" structures.

Your MATH explicitly rejects this idea.

Also, no one is suggesting that if
evolution occurred that the newly evolved systems would show anything
other than a fair degree of sequence similarity.

Your MATH explicitly says that the size of the gap needed to be
crossed by completely random processes (selection only happens after
the keystone falls into place by chance alone) to produce a function
is roughly 30% of the total size of the end protein.

Your MATH belies your verbiage.

That is certainly
reasonable and downright obvious. What is in question here is the
notion that sequence similarity alone is enough to conclude the
activity of random mutation and function-based selection in all
cases. For low-level systems, this is a fine conclusion because it
can actually be demonstrated in real time and because it can be shown
that the gaps between old and new are quite small indeed. However,
when it comes to those systems that have greater minimum structural
threshold requirements, the case is not so clear.

By this, presumably, you are saying that some functions that *did*
appear had to have done so by a process that involves your
mathematical model involving gap sizes that are roughly 30% of total
size. And that there are no functional intermediate stages during the
changes that randomly produce the right combination of aa's in all the
30% of total size.

The old and the new
are demonstrably more widely separated in sequence/structure space.

No. They *diverge* in sequence over time. Initially by quickly
optimizing for new function. More generally by random drift. There
is no correlation between degree of sequence similarity and function
for sequences that have the same function that is not time-related.

And, the degree of this separation has a linear relationship to the
minimum structural threshold requirements.

No it doesn't. I have repeatedly asked you to demonstrate that there
is a correlation of degree of divergence and size that is not time-
related (that is, reflects time since divergence).

That poses a real problem
for the proposed mechanism of random mutation and function-based
selection - a exponentially expanding problem.

So you keep asserting without evidence.

But, importantly, for
recently evolved 'new' functions, there would be greater than
necessary similarity. That is, like hemoglobin's globin genes, there
will be similarity in features that are not *required* for function,
such as the placement of introns. And, to the extent that selection
is conservative, there will be specific retention of selectively
relevant sites. Enough so that genes can be grouped into families of
proteins. Now, we can actually check if this, in fact, is what we
see.

The location of introns is not functionally neutral as you suggest.
Neither is the sequence of introns functionally neutral.

*If* Sean's model of evolution were correct, each new function must
start at some average distance away (that distance being a function of
the size of the end product) and must *independently* evolve by a
random walk.

Again, that is a deliberate strawman mischaracterization of my
position - i.e., a lie.

That is what your MATH says, Sean. Your MATH says that evolution
works by starting with some protein that, on average, is x aa away
from having a new function, with x being the gap size (which you
assume is nothing but a linear extrapolation from cytochrome c). You
have stated that gap size is about 30% of total size. And your MATH
assumes that the starting point for evolution is some protein x aa
away from your stated functional sequence *and* that no intermediate
between the two can be functional. That the gap must be crossed by a
random walk. That is exactly what a calculation of 20^x generates.

Novel potentially beneficial systems that are
actually found are not the average distance, but the minimum
distance.

Which you call the gap size *by your MATH*. If that is not what you
meant, then change either your math or your words. Otherwise you come
across as incompetent and unable to understand what your math is
saying.

It is just that the minimum likely distance is affected by
the average distance. To use your own example of Starbucks Coffee, if
the average distance between you and a Starbucks is 50 miles, does
that mean that a random walk would have to cover a linear distance of
50 miles before you would find a Starbucks? Not at all. With an
average distance of, say, 20 miles, the likely minimum distance would
be quite a bit shorter than that - perhaps as short as 1 mile? - - or
even the minimum possible distance of zero miles? But, given a
particular person, what are the odds that they will be within, say,
100 yards of a Starbucks given an average distance of 20 miles?

SFW? Evolution does not have to happen independently in each
organism. Average distance is completely irrelevant. As long as
there is one person close to a Starbucks, someone will get coffee.
And unlike a business like Starbucks, functions only have to evolve
once to be seen. If thousands of species have a particular function,
they almost always have acquired it by vertical inheritance from
common ancestors, not by evolving it independently in each species.

Now, lets consider a situation were the average distance between
starting and target locations was a trillion miles. What are the odds
that anyone of the old "starting" locations would be within 1 mile of
any "target" location? You see, average distance has a clear effect
on the likely minimum distance between existing starting points and
potential targets.

No it doesn't. The *minimum* distance (gap size) is *always* one
mutational step regardless of size of the end result. Minimum
distance is an absolute. You can't be a little bit minimum. You do
understand what the word "minimum" means. You can talk about the
probability of any particular evolutionary event being at the
minimum. You can calculate the *averge* gap size based on whatever
assumptions you want. But the *minimum* gap size is still always
one. And those systems that *did* evolve were ones in which, in some
ancestral organism, the gap size was close to the minimum according
to *real* evolutionary models (but not your strawman).

< snip >

*If* my mechanism of evolution is correct, then the entirety of known
protein structures will be 'clustered' rather than being found in
widely separated randomly placed 'islands' (precisely because not all
of protein structure space has been explored). The structure of the
space bounded by and including the functional sequences that *do*
exist will be organic, with flows outward and along useful axes as
nearby structure space is gradually explored.

All language/information systems show this sort of clustering.

Understandably so, because language systems *evolve* and change over
time.

The
problem is that all language/information systems show a linear
separation of the clustered islands with increasing structural/
sequence threshold requirements. This feature is exactly what is
illustrated in the work of Choi and Kim (see link below).

Can you point out the place where Choi and Kim state that they
analysed their data to show that there is, specifically, a linear
increase in distance between larger proteins and the proteins from
which they are derived? You and I both know that there was no such
analysis and that all you are doing is presenting an eyeball "This is
what I would like to believe that I see" conclusion based on what you
think a particular figure shows. Instead, you are intentionally
misreading a Figure that does not show what you claim. That includes
an inability to understand the difference between a figure that shows
the population of all protein structures that have ever existed (not
shown) and a sampling of all the protein structures that are currently
known. Again, I attribute this to your incompetence and inability to
understand the math of any argument you have ever made.

For your
position to actually work, the clustering would have to stay clustered
to the same degree with increasing structural/sequence threshold
requirements. This is clearly not the case. The beneficial islands
become less and less tightly arranged with increasing size and/or
specificity requirements. They are not isolated in one tiny corner of
sequence space as you imagine, but do in fact traverse the entire
distance across sequence space. The "axes" you speak of actually
extend over the entire distance of sequence space and show a linear
increase in the *minimum* distance between protein-based systems with
increasing threshold requirements. This is a fact that you have yet to
grasp.

http://www.pnas.org/cgi/content/full/103/38/14056

Except that *your* interpretation is not supported by the authors.
And apparently you took their figure as representing something it does
not. What it shows is that each type of increase in size is *from*
(by smallish modification from) some close existing structure. In
some cases this involves chimera formations. In others it involves
endoduplications. That is an evolutionary pattern involving repeated
exploration of nearby sequence space, not a pattern that involves
random searches throughout total sequence space. That the functional
structures that *have been* found by exploring structure space are not
randomly scattered in total seqeunce space is the important point that
differs from your model of the structure of structure space.

*If* Sean's model of evolution is correct, then I would expect to see
a random pattern representing the randomly placed functional
structures in total sequence space.

That's nonsense. As already noted many times for you, existing
beneficial sequences in all language/information systems show this
clustering effect - even those that were designed (like computer codes
for example).

Again, I say that because that is what your MATH says. What you say
contradicts the math you say supports your ideas.

The reason for this clustering effect is due to a
feature known as "conservation of design". There simply is no need to
reinvent the wheel every time a wheel might be needed for a new type
of function. Simply use the old wheel with modifications to work in a
different application/system. There is no reason, then, to expect a
random distribution of existing beneficial systems in sequence/
structure space, even at higher threshold levels, if they were in fact
designed.

Your MATH is based on assumptions of randomness only in constructing
new function. Now you are saying that whoever designed these systems
is using "old wheels with modifications to work in a different
application/system". But that would imply that your numerology that
involves gap size is irrelevant. If a designer can borrow features
from pre-existing systems and modify them to work in a different
application/system, what exactly does gap size tell us about
evolvability. After all, if a designer can re-use the heme-binding
ability of cytochrome c1 in making cytochrome c, that means that
almost all of cytochrome c function pre-exists and only a few steps
are needed to convert a cyt c1 into a cytochrome c.

Given my model of sequence space and the distribution of
existing and potentially beneficial sequences/structures, existing
systems should always have a clustered appearance.

Bull ***. Your model does not predict a clustered appearance at
all. At least that is what the MATH says. Your lips say something
else, but I try to follow the MATH for the same reasons that people
follow the money rather than rely on the words of a corrupt
politician.

It is just that
this clustering will also show an increase in the minimum distance
between existing systems with increasing structural threshold
requirements, which is exactly what we see in the available data for
every language/information system - to include biosystems (see link to
Choi and Kim above or do a BLAST search to check for yourself).

Apparently you still do not know what a BLAST search does. I have
read Choi and Kim and would appreciate it if you would point out where
they did the analysis you claim they did. Don't just point me to the
figure. Your eyeball interpretation of that figure is not a
statistical analysis of the data.

< snip >

If the size of the minimum gap distance is in fact related to the
minimum structural threshold requirements for a system, then the
answer to this question for higher level systems is no. It is not at
all likely or reasonable to suggest that the gap distance would ever
have been small enough to cross - - even given trillions upon
trillions of years of time.

And this is an assertion based on numerology alone. If you really
want to demonstrate that there is a correlation between gap size and
total size, you need to present *data*, not numerology. Otherwise it
is merely an assertion based on an assumption.

I have presented the data to you many times. Do a BLAST search or
look at Choi and Kim's paper again. The minimum distance between one
protein and the next does in fact increase in relationship to the
structural threshold requirements.

Bull ***. The degree of sequence similarity/dissimilarity between aa
sequences for a given function is largely a function of time since
divergence and the fraction of the particular protein that is under
selective constraint. That is what the data say. That is what path
analysis says. This also holds for sequence similarity/dissimilarity
between aa sequences for proteins that are derived from each other but
have different functions, although there *are* more differences
because of selection quickly optimizing for difference in function.
Take for example myoglobin and the hemoglobins or alphaglobin of
hemoglobin and betaglobin or the different embryonic, fetal, and adult
hemoglobins. Although there are many similarities, because there are
no differences in size, we can clearly observe that the differences
are a function of time since divergence (producing those differences
that are selectively neutral), constraint in those (small) parts of
the sequence that are involved in the parts of function that are
similar, and differences in sequence that affect the specific
differences in function (both the latter being of the more slowly, for
the conserved, or more rapidly, for the ones optimizing difference in
function, changes due to selection). Overall, most of the differences
between these proteins are probably due to neutral drift since the
functions diverged.

There is also the evidence for the evolution of the aldosterone
receptor from a cortisol receptor, with an ancestral receptor having
both capacities. That is, duplication and divergence allowed
specialization. This dual ability was a consequence of change in two
aa's.

This is a demonstrable fact. There
really is no argument here for anyone who actually looks at the data.

That obviously excludes you.

What do I mean by a higher-level system? I mean a system that has a
minimum structural threshold requirement of more than 1,000
specifically arranged amino acid residues and/or codons of genetic
real estate. Is this threshold requirement the gap distance as you
keep claiming? No. Absolutely not.

Dammit. I do not say that the 1000 aa is the gap size, Sean. I
never have. In fact, every time I say, specifically, that it is NOT
the gap size. I do say that, because that is the only number you give
(you do not even give the 30% value) until I point out that 1000 is
not the gap size, you seem to be intentionally or incompetently trying
to mislead people into thinking it is the gap size.

As I've noted for you more times
than I can count now, the gap distance is always smaller than the
threshold size.

BUT, as I keep pointing out, you do not mention gap distance UNTIL I
remind you that that is the number you need. I am merely trying to
get you to *honestly* present *your* argument rather than hide *your*
argument by only using the size number. I am NOT lying about *your*
argument when I point out that the number you need is "gap size" and
not "total size". I am trying to get you to *honestly* present *your*
argument.

You don't have to lie and deliberately misrepresent my position to
make this point clear to your satisfaction. All you have to point
out, whenever you feel the need, is that the structural threshold
level is not the minimum gap distance - that the minimum gap distance
is always smaller than the minimum structural threshold requirements.

But I cannot also point out that the gap size you do present is far
too large and irrelevant to real evolutionary mechanisms if all I do
is vaguely say that the minimum gap size (again the *minimum* gap size
is always one) is "smaller" than 1000, when what *you* mean is that
the minimum (or average or maximum) gap size is 300 without
misrepresenting you? Just saying that the minimum gap distance is
"smaller" is not enough when you are claiming to mathematically
demonstrate the impossibility of evolution because of the "gap size".

Pointing that out would not be a misrepresentation. And, it would be
something new for you besides your usual resort to strawman building
and deliberate lying misrepresentations.

So, the *minimum* gap size is what? The *average* gap size is what?
Why do you think that my saying the vague word "smaller" would not be
misrepresenting you when, in fact, your math (which says that for a
total size of 1000, the gap size is about 300 and that this
relationship is linear both up and down) declares either that the
*minimum* size (again, in my book *minimum* size is always one) or
*average* gap size or *maximal* gap size is 300? Just saying "gap
size" doesn't tell me which term applies to your *number* of 300. And
it matters whether you regard 300 as the *minimum gap size* (which is
often what you talk about), the *average gap size* or the *maximal gap
size*. I don't regard *your* number of 300 (for a total size of 1000)
as being either the *minimum* size (nor even the average gap size; it
is the *maximal* gap size -- the gap between a current function and
some protein that has not even a hint of that function). 300 is not
even significantly "smaller" than 1000 IMHO.

They are not the same thing. Is there a relationship
between the two? There most certainly is and this relationship is
linear.

Except that you have presented no actual data to support this.

Yes, I have. Choi and Kim clearly illustrate an increase in the
distance between protein-based systems with increasing structural
requirements. A simple BLAST search will show you the same thing.

Bull***. You are eyeballing a figure and misrepresenting it.

You
are merely *assuming* (actually, repeatedly asserting) that the
relationship is linear. I see 'new' modifications of old proteins
(such as resistance to antibiotics) that are not a function of the
size of the old proteins.

Resistance to antibiotics generally doesn't ...

.

Loading