Didn't Lenski just demonstrate this?

[MPM]

http://www.pnas.org/cgi/content/full/105/23/7899

The challenge cites Lenski's earlier 1986 paper, but since then his
lab has shown the evolution of a completely novel phenotype in E. coli
provided only with growth media and time.

[Scott Kerr]

I was literally just about to write the same thing when your message
popped up. Lenski's work easily fulfills every requirement of the
challenge. Just send the $20,000 his way.


Scott

[Deanne Taylor]

I agree...I'm sure Lenski could use the money for more cultures. :)

D

[bkl...@gmail.com]

I discuss Lenski's latest result on my website about panspermia, at --
http://www.panspermia.org/whatsnew.htm#080605

Perhaps "cryptic genes" were activated, as the team acknowledges.
Perhaps contamination played a role, although in a way that the
researchers do not consider. For these reasons, the phenomenon does
not surely demonstrate OEEI-QS. Lenski et al. are admirably restrained
in their own claims for the work. And contamination is apparently more
difficult to prevent than I thought earlier. I am wondering if
computer models are preferable for avoiding both of the possible
problems seen here -- cryptic genes and contamination.

[Scott Kerr]

Assuming "cryptic genes" were activated, how does that fail the
requirements of the challenge. A new trait, one that was not there at
the beginning of the experiment, has arisen with no input other than
food, time, and evolution.

Scott Kerr



On Jun 20, 9:11 am, bkl...@gmail.com wrote:
> I discuss Lenski's latest result on my website about panspermia, at --http://www.panspermia.org/whatsnew.htm#080605

[bkl...@gmail.com]

Let's say we want to know if a computer can write new programs for
itself. Speelcheck, for example. One day well after we begin an
experiment on a given computer, Spellcheck starts to happen. The
"trait was not there at the beginning of the experiment." Did the
computer write it? Or was it preloaded, needing only to be activated?
The former would be extremely interesting. It would demonstrate OEEI-
QS. The latter would be less interesting, and would not demonstrate
OEEI-QS.

[Scott Kerr]

The regulation of genes is just as important as the proteins coded by
the genes. Activation of a gene is not a trivial task. If the
computer was programmed to activate its spell check, then you are
right, it is not interesting. However, if it did not have the ability
to activate the program at the beginning of the experiment, then
programmed itself to activate its spell check that would be very
interesting and seems to fit the criteria of OEEI-QS.


Scott

[bkl...@gmail.com]

I agree that the regulation of genes is important. And I agree, if the
computer "programmed itself to activate" spellcheck or anything else,
it would be interesting. But detecting the difference between
"programmed-itself-to" and "was-already-programmed-to" seems harder
for regulatory functions than for basic applications and subroutines.
How can you tell?

[Bud]

If activating genes already present is assumed to not be OEEI, then
you are precluding any demonstration in a living system. At least to
my knowledge, there is no organism for which the full function of all
parts of all genes is completely understood, so you can always dismiss
any new phenotype as resulting from the activation of previously
inactive genes.

My comment may be stated a different way. Evolution is the change of
organisms in response to their environment, but you are limiting your
definition to a particular type of evolution resulting from the
creation of "new genes" or "new instructions". Yet you do not
rigorously define what you mean by these terms. I think that in both
biology and computer science, there are real difficulties in drawing a
bright line between "new" and "modified".

> > > > Scott Kerr- Hide quoted text -
>
> - Show quoted text -

[Scott Kerr]

Well, going back to Lenski's experiment, the way you can tell is that
for 30,000 generations the bacteria could not use citrate as an energy
source. If they had the ability to activate citrate metabolism they
would have. The early generations could not use a citrate-only media;
they may have had "cryptic genes" but could not activate them to
survive on citrate. At some point the regulatory programming changed
with no input other than media and time.

-Scott

On Jun 23, 9:04 am, bkl...@gmail.com wrote:

[bkl...@gmail.com]

If evolution advances when regulatory changes unleash cryptic
programs, then the range of the process is limited to the original
inventory of cryptic programs. This is not open-ended. Nor could it
account for the history of life on Earth, if the first life was
primitive, as believed.

If the regulatory apparatus can swap and choose programs without
crashing, that's impressive, no doubt. In fact, regulatory-system-
dominated method could account for the history of life on Earth, if
you allow more cryptic programs to be introduced as needed -- if the
system is unquarantined. But we are exploring the range of quarantined
systems.

I am finding your questions very helpful. I hope my replies are
helpful also.

[mike markey]

From your blog:
"... the capability of darwinian evolution to write new programs in a
closed system has not been clearly demonstrated."

Really, I think your dismissal of this work is unfair. If there were
contamination in the form of plasmids contributed from other bacteria,
there would be plasmids in the citrate-using strain. There isn't.
They can show that the new bacteria is the ancestor of the original
strain. "Activation of cryptic genes" in this context is more similar
to what I would describe as cooptation, which in itself is no small
feat. That is, the genes for some existing protein or series of
proteins were gradually altered in such a way that they were put to
work at a second task to what they normally do, namely digesting
citrate. But what if the changes were in non-coding regions of DNA,
or pseudogenes? Would that now qualify as OEEI? I'm concerned that
what your asking for is a demonstration of ex nihilo creation of a
gene, genetic material and all. Evolution requires some heritable
material to act upon, either from gene duplication, abiogenic methods,
or other means; nothing poofs into existence.

I appreciate the question you're trying to ask, but ultimately isn't
the fact that we're here, living with millions of other species
filling every conceivable niche, demonstrate that evolution can
innovate? If the answer is "no, genetic material could have been
delivered from space", then lets just zoom out our frame of reference
from the earth to the solar system, or the universe.

--
Mike Markey
http://www.geocities.com/jfsebastion/

[bkl...@gmail.com]

Plasmids -- The citrate-using strain has the genetic program that has
been shown in other cases to be delivered by plasmids. Plasmids can
become incorporated into the main chromosome and disappear as separate
entities. Even if the subject strain has no plasmids now, the above
scenario is not ruled out. Future analysis by Lenski et al. may be
illuminating here.

Cryptic Genes -- My computer came with software that can find wireless
networks. If I never knew it had that function, it is effectively
cryptic. If I eventually need it, and with tinkering I can find it and
get it to work, good, but no OEEI. But if my computer composes the
software itself, using available unrelated bits, bytes and files, then
yes, spectacularly, OEEI. The same logic applies in biology.
(Available unrelated DNA is never lacking.)

Zoom out -- Theories of the origin of the universe are explicitly
excluded from consideration in this project. The big bang is too
remote and too fluid to close the subject. In my opinion, the lack of
direct evidence has at least equal importance, and the question is
open.

Nothing poofs into existence -- This is what I'm wondering! But let's
try to find out -- can new genetic meaning come into existence? Is
OEEI-QS possible? The question is not an easy one. Thanks for thinking
about it.

[engr.student]

Maximum entropy is a measure of the information capacity of the
system.
For any closed system there is a maximum level of entropy. This means
that there is a maximum level of information.

Neither disorder alone nor order alone are measures of information.
Information is a mixture of order and disorder. (I know that is like
saying music is assonance and dissonance.)
For any closed system the entropy rate is greater than or equal to
zero. This means that there is a tendency away from some intermediary
state toward an information-poorer pole... the system tends toward
dissonance. In general information tends to be lost, or stay
constant. This is a macro-scale property.

You are asking about the micro scale. Is it possible in the disorder-
generating process of information transfer, coupled with Darwinian
evolution to concentrate information? Yes. That is inarguable.

What you are looking for is the limits on the process.

Darwinian evolution can be fully modeled as an ensemble kalman
particle filter, a mathematical model that has nearly nothing to do
with particles, or filtering them, but has everything to do with
optimal state estimation, and propagation. If the "ideal" information
peaks are highly localized (think dirac-delta functions) then the
method has problems finding them. If the peaks are too large and
shallow the method also has problems. It should be demonstrable that
if the "evolution rate" and "particle population" are properly scaled
then the system tends to converge to the "evolutionary ideal".

-mike

[bkl...@gmail.com]

Mike, thanks for your thoughts. (Let's keep the original thread topic,
so everyone will recognize it.)

Just this comment: Almost any process generates "information." I am
asking about genetic programs. These are a very specific kind of
information. Whether darwinian evolution creates these has not been
established.

[Keith C]

bkl,
You wrote, "If the regulatory apparatus can swap and choose programs

without
crashing, that's impressive, no doubt. In fact, regulatory-system-
dominated method could account for the history of life on Earth, if
you allow more cryptic programs to be introduced as needed -- if the
system is unquarantined. But we are exploring the range of quarantined
systems. "

I think Avida has already met your challenge, whether you want to
admit or not.
There was no question of swapping and choosing programs or of cryptic
programs being introduced as needed. Avida satisfies your quarantine
specification.

What does seem to be important for evolution of novelties is the
presence of some redundant or surplus 'gene' sequences, usually formed
by gene duplication. Such sequences are free to change by random
mutation with little constraint, but which may eventually performing a
new function. A key is that the organism or program must be rewarded
if the function is performed. That requires that the ability is
tested for.
As I have said before, the only sense in which Avida is not open-ended
is that there are only a finite number of activities which are
rewarded.

[bkl...@gmail.com]

"Avida satisfies your quarantine specification" -- agreed.

"A key is that the organism or program must be rewarded if the
function is performed. That requires that the ability is tested for."

-- This is the manmade teleology that amounts to cheating.

"the only sense in which Avida is not open-ended is that there are

only a finite number of activities which are rewarded." -- Exactly.
The only thing rewarded by darwinian evolution is survival. The system
must produce Innovations that are not explicitly rewarded in order to
have the potential for Open-Endedness.

[Bud]

I think that I have to disagree with one point. You write "the only

sense in which Avida is not open-ended
> is that there are only a finite number of activities which are
> rewarded"

It seems to me that the number of activities which are rewarded is not
limited.
The competition between organisms in Avida is complex. It depends on
the rate of reproduction and the probability that copies of the
organism will be viable. This latter means that the "progarm" (DNA)
be resistant to damage from mutation, and also that it be efficient.
The limited activities which are tested for are the analogs of
metabolism of various food sources. This is only a limited part of
what determines the ability of an organism to compete successfully.

Consider the number of types of metabolism found in nature. There are
quite a few, but the number is definitely finite. This fact has not
precluded open ended evolution!

The Avida model is restrictive in that all of the organisms must
compete within the same space. There is at present no analog of
isolated ecological niches, but there is no reason this feature could
not be added.

When you watch the education version of Avida run, what you can easily
see is the evolution of the abilities to metabolize the various food
sources. These abilities are monitored, but they are not the only
capabilities which evolve. The others are harder to see, but only
because they are not tabulated in real time while the program runs.

Further, you write "A key is that the organism or program must be

rewarded
if the function is performed. That requires that the ability is

tested for." In Avida, a program or organism can be rewarded (by
increased probability of survival) without testing for a particular
ability. An example is resistance to lethal mutations. This is not
tested for, but is clearly rewarded by survival. I think that the
fact that Avida reports which food sources an organism has developed
the ability to utilize has misled people to think that these are the
only ways that an organism can improve its chances for survival. They
are just the most visible, and they appear rather quickly.

I can see no basis for stating that evolution as modeled by Avita is
not open-ended.

> > > > On Jun 25, 8:25 am, "mike markey" <mikepmar...@gmail.com> wrote:- Hide quoted text -

[Keith C]

bkl,
In responce to my statement, "the only sense in which Avida is not

open-ended is that there are
only a finite number of activities which are rewarded."

you wrote

" -- Exactly.
The only thing rewarded by darwinian evolution is survival. The system
must produce Innovations that are not explicitly rewarded in order to
have the potential for Open-Endedness. "

I get my information about Avida from here:-
http://myxo.css.msu.edu/papers/nature2003/
and I suspect you should re-read all of this very carefully.

Neither Avida nor 'Darwinian evolution' rewards survival. Survival or
reproduction is a consequence of demonstrating fitness relative to the
competition.
In the case of Avida, fitness is measured by performing one or more of
the nine 2 input logical operations which the program checked. Reward
was SIP's dependent on the difficulty of the operations performed.
SIPs were needed for reproduction.

This system is completely incapable of evolving a program to perform a
logical operation which is not checked for and adequately rewarded,
and it is this restriction which limits the range of novelties which
can be produced ie stops the program from being 'open-ended' in your
sense.

[Patrik]

Ruling out Lensky's work on the basis that it was merely reactivation
of cryptic genes is a cop-out, in my opinion. Evolution *always* works
with what is already there. Those cryptic genes may stem from a long-
lost ability to metabolize citrate, or they may have had a different
role altogether.

Perhaps a clearer demonstration would be to show evolution of a
xenobiotic metabolic function. There exist a good number of non-
natural amino acids, for example. There's all the D-forms of the
standard L-amino acids, to start with, but there are also a number of
novel manmade amino acids. There are all fairly simple chemicals,
which should only require a few enzymatic steps to metabolize, yet
they have never been seen in nature in a sufficient quantity to assume
that E. coli should have had any prior adaptation to utilizing them.

Figure out which unnatural amino acids E. coli cannot already
metabolize due to some fluke of chance. Then concoct a "rich broth"
mixture, combining as many of these as possible without affecting
growth rate, to maximize the number of ways a mutated strain could
gain an edge over wild type. Dump into a bioreactor, and sit back and
wait...

If a new strain evolves to metabolize one of the unnatural amino
acids, it may still be due to activation of some cryptic genes, but it
would necessarily be using those genes to perform an overall metabolic
function which it has never done before.

On Jun 24, 8:42 am, bkl...@gmail.com wrote:
> If evolution advances when regulatory changes unleash cryptic
> programs, then the range of the process is limited to the original
> inventory of cryptic programs. This is not open-ended. Nor could it
> account for the history of life on Earth, if the first life was
> primitive, as believed.

>

[Patrik]

PS: Unnatural amino acids is just the first thing that popped into my
head. Of course, this is merely one example of a more general idea. In
fact, focusing on amino acids may not be the best starting point,
because I would assume that a good number of unnatural amino acids can
already be degraded by existing metabolic pathways, even though the
amino acids themselves could not be directly incorporated into
proteins.

PPS: If we select unnatural compounds which cannot already be
metabolized by *anything* in your lab - not just E. coli - this would
also greatly reduces the chances that the novel function could have
been brought in by contamination. So leave some plates out on a shelf,
throw a few on the ground, stick one in the garbage bin, and see if
anything grows on it...

[zeamays]

I agree that Lenski has, to a first approximation, complied with the
requirements of this Challenge. He is a conscientious scientist who
stated in his article that he is currently doing the necessary
controls to refute some of the objections that were made above,
particularly the concern for contamination.

However, I think that for a stronger case for an experimental proof of
OEEI-QS, one needs to consider that the forms of evolution that are of
most interest to the public are those in sexual eukaryotes, not in
clonal prokaryotes. However, here too, there is plenty of evidence
already available for how complex evolution has occurred in the past
from DNA sequencing studies. Some very dramatic developmental changes
are known to occur in plants, animals and fungi with very small
changes in homeobox DNA sequences, while other dramatic changes are
known to be associated with gene and chromosomal duplications and
other chromosomal changes that occur in sexual recombination. By
their eukaryotic and sexual nature those systems can't be demonstrated
in a closed system with a clonal bacteria line. They might be
demonstrated in a closed experiment with a sexually reproducing
eukaryote, but the time limitation (two years) of this InnoCentive
competition would most likely preclude such a study because of the
longer generation times.

[bkl...@gmail.com]

After some of the preceding discussion, I have had the opposite
thought -- the best system for a proof might be even simpler. In very
complex systems, like eukaryotes where maybe 99% of DNA does not code
for protein, it is very hard to know exactly what the initial system
contains. For example, in a recently reported analysis of fruitflies,
nine "de novo" genes are said to have emerged, all longer than 300
nucleotides, already properly sequenced. These would have to be
"cryptic genes."

And in biological systems, contamination is virtually unavoidable.

So I am, for the moment, inclined to favor computer simulations as the
best medium for a demonstration. The starting inventory can be known
completely, and contamnation can be easily prevented. But I am open-
minded.

PS -- the link given in June for an article on panspermia.org about
Lenski has changed. The permnanent link is http://www.panspermia.org/whatsne51.htm#080605

[zeamays]

I'm sorry, but I don't agreed.

It really doesn't matter whether a biological system has been fully
defined in advance. If don't see any reason that needs to be a
precondition to demonstrating evolution.

The argument that contamination (and similar artifacts) might be
responsible for the observations of Lenski for citrate metabolism in
coli can be made for any experimental system, but should be subjected
to Occam's razor. They aren't the most simple, straightforward
solution, and although Lenski is in the process of excluding it, such
complex explanations should not be assumed in the absence of contrary
evidence.

As for computer programs, I am sorry, but they are not really
experiments. The result of the program will be determined by the
programmer, who will either program the computer to make evolution
possible, impossible, or a matter of chance. If the result of the
program is not known in advance then (unless chance is programmed into
the solution) it is just a matter of the programmer not being smart
enough too know the result in advance. A program just follows the
logic assigned by the programmer.

On Aug 29, 3:10 pm, bkl...@gmail.com wrote:
> After some of the preceding discussion, I have had the opposite
> thought -- the best system for a proof might be even simpler. In very
> complex systems, like eukaryotes where maybe 99% of DNA does not code
> for protein, it is very hard to know exactly what the initial system
> contains. For example, in a recently reported analysis of fruitflies,
> nine "de novo" genes are said to have emerged, all longer than 300
> nucleotides, already properly sequenced. These would have to be
> "cryptic genes."
>
> And in biological systems, contamination is virtually unavoidable.
>
> So I am, for the moment, inclined to favor computer simulations as the
> best medium for a demonstration. The starting inventory can be known
> completely, and contamnation can be easily prevented. But I am open-
> minded.
>
> PS -- the link given in June for an article on panspermia.org about

> Lenski has changed. The permnanent link ishttp://www.panspermia.org/whatsne51.htm#080605

[Keith C]

Zeamays,
I think your comments on computer programs are wrong.
Suppose the 'problem' is to convert an English word into another by
changing one letter at a time, with the constraint that each of the
intermediates must be in a dictionary or a derivative of such a word.

It is easy to find pairs of words for which I do not know whether the
conversion is possible.
Start with a list which includes only the first word of the pair.
Have the program program to choose a letter at random in the first
word, make a random change, then look up the dictionary to see if that
word is legal. If it is legal, add that word to the list and make
another random change to it.
If the word is not legal, select a word from the list of known words
and make a random change.
Continue until the target word is reached, or until some extremely
large number of changes have been tried.

In one sense such a program is following the logic programmed into
it. However I am not determining its 'solution' in any way, or even
determining whether there is a solution for the two words used.

[zeamays]

You've answered your own question.

I'm sorry, but in your example the problem is not, "Can evolution
occur in a general sense?" (the problem that was posed), but "Can it
occur between two specific states, a specified initial state and a
specified final state?." The program design you described makes the
evolution possible in the general sense, since I can easily name at
least one initial and one final state for which your program will
work.

I can, however, postulate there are programs or states where the
programmer is just not smart enough to know if the object can be
achieved or where randomness is assigned to whether the objective is
possible. An example of the former is where the complexity of a
successful route, particularly due to the number of options generated
along the pathway, precludes knowing if it will lead to a solution.

Best wishes.

> ...
>
> read more »