PDA

View Full Version : Abiogenesis - The Information Model



Selfsim
2012-Dec-13, 05:34 AM
The latest Paul Davies/Sara Walker thinking on an informational model type approach on abiogenesis has turned up in 'Interface'. As follows:

The algorithmic origins of life. (http://rsif.royalsocietypublishing.org/content/10/79/20120869.full)

I haven't had a chance to read through it all just yet, but there's an easier-to-read Physorg article about it also:

Researchers propose new way to look at the dawn of life (http://phys.org/news/2012-12-dawn-life.html)


"When we describe biological processes we typically use informational narratives – cells send out signals, developmental programs are run, coded instructions are read, genomic data are transmitted between generations and so forth," Walker said. "So identifying life's origin in the way information is processed and managed can open up new avenues for research."

"We propose that the transition from non-life to life is unique and definable," added Davies. "We suggest that life may be characterized by its distinctive and active use of information, thus providing a roadmap to identify rigorous criteria for the emergence of life. This is in sharp contrast to a century of thought in which the transition to life has been cast as a problem of chemistry, with the goal of identifying a plausible reaction pathway from chemical mixtures to a living entity." Focusing on informational development helps move away from some of the inherent disadvantages of trying to pin down the beginnings of chemical life. "Chemical based approaches," Walker said, "have stalled at a very early stage of chemical complexity – very far from anything we would consider 'alive.' More seriously they suffer from conceptual shortcomings in that they fail to distinguish between chemistry and biology."The underlined bits above should be of interest to some folk 'round here.

More later.

Comments on the paper, the model, and the new approach are welcome ... (it looks interesting).

Cheers

Selfsim
2012-Dec-13, 06:20 AM
Here's an interesting snippet (from the paper itself).

Section 4.2 Algorithmic Takeover:


We point out a curious philosophical implication of the algorithmic perspective: if the origin of life is identified with the transition from trivial to non-trivial information processing—e.g. from something akin to a Turing machine capable of a single (or limited set of) computation(s) to a universal Turing machine capable of constructing any computable object (within a universality class)—then a precise point of transition from non-life to life may actually be undecidable in the logical sense. This would probably have very important philosophical implications, particularly in our interpretation of life as a predictable outcome of physical law.
Hmm ...

Paul Wally
2012-Dec-13, 11:23 PM
The latest Paul Davies/Sara Walker thinking on an informational model type approach on abiogenesis has turned up in 'Interface'. As follows:

The algorithmic origins of life. (http://rsif.royalsocietypublishing.org/content/10/79/20120869.full)



Thanks for that. There are a few things I would like to comment on from the article.


However, this very generality is also the greatest weakness of the paradigm as applied to the origin of life: it provides no means for distinguishing complex from simple, let alone life from non-life. This may explain Darwin's own reluctance to speculate on the subject, ‘One might as well speculate about the origin of matter’, he quipped.

On the question of ordinary matter: Someone once asked in some thread whether rust (iron oxide) is alive, because it tends to grow just like real life. In a sense one could say iron oxide replicates itself as a 'chemical species'. So what's the difference between this and genetic replication? My answer to this problem is that iron oxide is not unique. The very same species will pop up in different places independent of each other, and there is no single gene that is spread from some place of origin to other places. I think there should be some concept in information or complexity theory for this kind of uniqueness of information.


In our view, an explanation of life's origin is fundamentally incomplete in the absence of an account of how the unique causal role played by information in living systems first emerged. In other words, we need to explain the origin of both the hardware and software aspects of life, or the job is only half finished.

I see there's heavy reliance on computer analogies like 'hardware', 'software', 'information flow', 'bits', 'algorithms' , 'endcoding', 'decoding', 'storage' etc. In my opinion, I believe it's a mistake to draw such artificial distinctions. I would prefer a more holistic approach, where information is defined as what makes something as a whole different or unique from other things. Can information perhaps be defined and quantified as degree of uniqueness?

Later on their thinking does seem to move in the direction of a holistic approach, and they relax a bit on the computer metaphor:


The algorithm itself is therefore highly delocalized, distributed inextricably throughout the very physical system whose dynamics it encodes. Moreover, although the ribosome provides a rough approximation for an UC (see endnote 5), universal construction in living cells requires a host of distributed mechanisms for reproducing an entire cell. Clearly, in an organism the algorithm cannot be decomposed and stored in simple sequential digital form to be read-out by an appropriate machine in the manner envisioned by Turing and von Neumann for their devices.

I think perhaps Tononi's 'integrated information' concept is on the right track.


It therefore provides a measure of the distributed information generated by the network as a whole as a result of its causal architecture.


One last comment on the following:

Purely analogue life forms could have existed in the past but are not likely to survive over geological timescales without acquiring explicitly digitized informational protocols. Therefore, life forms that ‘go digital’ may be the only systems that survive in the long-run and are thus the only remaining product of the processes that led to life.

I have some issue with this. If we interpret life through a digital paradigm I don't think it makes sense to ask why is life digital and not analogue. Is life digital?

Colin Robinson
2012-Dec-14, 01:45 AM
I have some issue with this. If we interpret life through a digital paradigm I don't think it makes sense to ask why is life digital and not analogue. Is life digital?

Not sure about the statement "life is digital"... But I think it's reasonable to say that the genetic code is digital, rather than analogue.

The thing is, digital doesn't necessarily have anything to do with electronics or the base 2 system. An abacus is a digital device, so is the sort of mechanical desk calculator which was in use for much of the 20th century.

Digital basically means that information is stored as a string of digits — items that have a strictly limited number of states — just two in the case of a modern computer, but ten in the case of the old mechanical desk calculator... That is how the genetic code works – it uses just four different nucleotides strung in groups of three to get 64 possible codons.

By contrast, an analogue system uses one continuously-variable quantity to represent another. E.g. A simple analogue representation is a temperature graph, where distance above the horizontal axis represents temperature. And that is clearly not how the genetic code works, although it possible to imagine a simple self-propagating system storing its information in such a way.

Paul Wally
2012-Dec-14, 02:50 AM
Not sure about the statement "life is digital"... But I think it's reasonable to say that the genetic code is digital, rather than analogue.

The thing is, digital doesn't necessarily have anything to do with electronics or the base 2 system. An abacus is a digital device, so is the sort of mechanical desk calculator which was in use for much of the 20th century.

Digital basically means that information is stored as a string of digits — items that have a strictly limited number of states — just two in the case of a modern computer, but ten in the case of the old mechanical desk calculator... That is how the genetic code works – it uses just four different nucleotides strung in groups of three to get 64 possible codons.

By contrast, an analogue system uses one continuously-variable quantity to represent another. E.g. A simple analogue representation is a temperature graph, where distance above the horizontal axis represents temperature. And that is clearly not how the genetic code works, although it possible to imagine a simple self-propagating system storing its information in such a way.

I get that. But in the paper it appears that they question DNA as the sole storage of the genetic code (and information) which means the information is distributed through the organism. It becomes then unclear whether we are still dealing with an actual digital information system or whether they're just applying it as a model.

Colin Robinson
2012-Dec-14, 05:47 AM
I get that. But in the paper it appears that they question DNA as the sole storage of the genetic code (and information) which means the information is distributed through the organism. It becomes then unclear whether we are still dealing with an actual digital information system or whether they're just applying it as a model.

I don't think they're saying that life is a purely digital system. On the contrary, they say that life processes information in both digital and analogue ways (section 2, first paragraph):

"In informational language, genetics and metabolism may be unified under a common conceptual framework by regarding metabolism as a form of analogue information processing... to be contrasted with the digital information of genetics."

In section 2.2 they say that "modern life" is a "hybrid" of digital and analogue...

Selfsim
2012-Dec-14, 06:19 AM
On the question of ordinary matter: Someone once asked in some thread whether rust (iron oxide) is alive, because it tends to grow just like real life. In a sense one could say iron oxide replicates itself as a 'chemical species'. So what's the difference between this and genetic replication? My answer to this problem is that iron oxide is not unique. The very same species will pop up in different places independent of each other, and there is no single gene that is spread from some place of origin to other places. I think there should be some concept in information or complexity theory for this kind of uniqueness of information.
I see there's heavy reliance on computer analogies like 'hardware', 'software', 'information flow', 'bits', 'algorithms' , 'endcoding', 'decoding', 'storage' etc. In my opinion, I believe it's a mistake to draw such artificial distinctions. I would prefer a more holistic approach, where information is defined as what makes something as a whole different or unique from other things. Can information perhaps be defined and quantified as degree of uniqueness?So, let me understand what you're saying here Paul ... you acknowledge uniqueness, within a backdrop of Chemical Law, (presumably on the basis of evidence for a mechanism, (ie: genetic reproduction). You call for a concept in information or complexity theory, to explain this 'uniqueness of information'.

So does this mean you accept that uniqueness of information can be generalised, even perhaps, in the absence of a defining mechanism?

Also, are the terms: 'hardware', 'software', 'information flow', 'bits', 'algorithms' , 'endcoding', 'decoding', 'storage' etc really just analogies and 'artificial' distinctions? I wonder ..?.. Turing's models all seemed to have had, at their cores, a life/biology perspective. I often wonder just how dissimilar the architecture of a computer really is, compared with life/biology. It may well be a more accurate analogy than we think, for this reason.

Selfsim
2012-Dec-14, 06:44 AM
I don't think they're saying that life is a purely digital system. On the contrary, they say that life processes information in both digital and analogue ways (section 2, first paragraph):

"In informational language, genetics and metabolism may be unified under a common conceptual framework by regarding metabolism as a form of analogue information processing... to be contrasted with the digital information of genetics."

In section 2.2 they say that "modern life" is a "hybrid" of digital and analogue...Its always kind of amused me that the technology used to manipulate information in the digital domain, has analogue circuits at it core. For example, the basics of digital logic technologies, are things like AND, OR, NAND, XOR, etc 'gates', which in turn, have transistors, resistors and capacitors cleverly connected in a circuit, deliberately designed to emulate binary digital states, determined by a rigid set of mathematical rules relating cause and effect (a truth table). I wouldn't call a transistor a 'digital' device, but when incorporated into a circuit which forces it into perform the function of a switching element, it suddenly appears as 'digital'.

The distinctions of 'analogue' and 'digital', I think, are very much dependent on the scale at which one is thinking and this case, I think they're attempting to use it as a metaphor for the model they're campaigning for.

Paul Wally
2012-Dec-14, 11:01 AM
I don't think they're saying that life is a purely digital system. On the contrary, they say that life processes information in both digital and analogue ways (section 2, first paragraph):

"In informational language, genetics and metabolism may be unified under a common conceptual framework by regarding metabolism as a form of analogue information processing... to be contrasted with the digital information of genetics."

In section 2.2 they say that "modern life" is a "hybrid" of digital and analogue...

There it is again: "Life processes information". That is what I find problematic. You see, if we say life processes information then we commit to the computer paradigm. I have no problem with accepting it as a way of understanding how life works or to use it as a modeling tool. The problem is when we say life really processes information. Immediately, if we do that, we introduce a dichotomy which seems to be entirely of our own making. Once we say life processes information, then there are two 'substances' that we must now look out for now - information and that which processes it, software vs hardware, genetics vs metabolism, chicken vs egg. I think these are pseudo-problems arising from applying the computer paradigm to life.

Paul Wally
2012-Dec-14, 10:31 PM
So, let me understand what you're saying here Paul ... you acknowledge uniqueness, within a backdrop of Chemical Law, (presumably on the basis of evidence for a mechanism, (ie: genetic reproduction). You call for a concept in information or complexity theory, to explain this 'uniqueness of information'.

I'd say, I'm looking for a formulation of the concept of information in terms of uniqueness. I know the idea is not clear, but maybe there is some information measure relating information content to uniqueness. The reason I'm thinking about this is because if something or some property lacks uniqueness, i.e. if it is pervasive then there is no need to have a gene for it. So non-unique properties don't need to be memorised but unique properties do need to be memorised if they are to persist. I almost want to say uniqueness is (new) information.

Selfsim
2012-Dec-15, 12:19 AM
I'd say, I'm looking for a formulation of the concept of information in terms of uniqueness. I know the idea is not clear, but maybe there is some information measure relating information content to uniqueness. The reason I'm thinking about this is because if something or some property lacks uniqueness, i.e. if it is pervasive then there is no need to have a gene for it. So non-unique properties don't need to be memorised but unique properties do need to be memorised if they are to persist. I almost want to say uniqueness is (new) information.Hmm .. interesting …

I'm reminded of the Dawkins response to the Creationist 'no new information' challenge. It might be worthwhile having a quick re-read of it, (I will .. when I get a chance), as I think it goes into some detail about the 'ins' and 'outs' of this. Here is the link (http://www.skeptics.com.au/publications/articles/the-information-challenge/), if you're interested.

The distinctions of phenotype and genotype, would also seem to cover the 'sameness' and 'variations' within a given species. (There's also a mechanism explanation in this as well).

Digital information transmission theory goes into some detail about extraction of a particular digital sequences from a noisy information channel. I'm pretty sure there are theoretical measures for quantifying unique sequences, as this leads onto error correction ...

Then I guess, there are all the usual Mathematics definitions of uniqueness (mainly from set theory) ...

Colin Robinson
2012-Dec-15, 08:14 AM
There it is again: "Life processes information". That is what I find problematic. You see, if we say life processes information then we commit to the computer paradigm. I have no problem with accepting it as a way of understanding how life works or to use it as a modeling tool. The problem is when we say life really processes information.

Whether "information" is a metaphor or a reality may depend on which definition of "information" you start from. I found the following on the WP page "Information"

"Systems theory at times seems to refer to information in this sense, assuming information does not necessarily involve any conscious mind, and patterns circulating (due to feedback) in the system can be called information...For example, Gregory Bateson defines 'information' as a 'difference that makes a difference'."


Immediately, if we do that, we introduce a dichotomy which seems to be entirely of our own making. Once we say life processes information, then there are two 'substances' that we must now look out for

It's not necessarily a distinction between two substances -- rather, distinction between a structure (e.g. a DNA or RNA molecule) and a function (e.g. storage and replication of inherited information). The distinction is important, because it enables us to consider questions like whether the same function was originally performed in a different (perhaps less centralized) way.


- information and that which processes it, software vs hardware, genetics vs metabolism, chicken vs egg. I think these are pseudo-problems arising from applying the computer paradigm to life.

The metabolism-first versus genes-first question is a pseudo-problem?

I can't agree. I think it is a historical question just as valid and important as e.g. whether the Sun and the planets formed at the same time? or whether human beings started to draw and paint before they started to write?

The difference is that the latter two questions have been pretty well resolved, whereas metabolism-first versus genes-first is still a matter of serious scientific debate...

Colin Robinson
2012-Dec-15, 10:05 AM
Its always kind of amused me that the technology used to manipulate information in the digital domain, has analogue circuits at it core. For example, the basics of digital logic technologies, are things like AND, OR, NAND, XOR, etc 'gates', which in turn, have transistors, resistors and capacitors cleverly connected in a circuit, deliberately designed to emulate binary digital states, determined by a rigid set of mathematical rules relating cause and effect (a truth table). I wouldn't call a transistor a 'digital' device, but when incorporated into a circuit which forces it into perform the function of a switching element, it suddenly appears as 'digital'.

I agree that the transistor is not inherently digital. The operational amplifier or op-amp, which is the basic unit of an analogue computer (roughly corresponding to the logic gate in a digital one) also has transistors in it.


The distinctions of 'analogue' and 'digital', I think, are very much dependent on the scale at which one is thinking

Yes, it is a matter of scale... it depends on the level of system one is thinking about. When you consider an electronic computer as a whole, the distinction between a digital computer and an analogue one is not at all problematic, it is quite easy to apply. Even when you go down to the level of logic gates or op amps, it is easy to say that the logical gates are digital devices, whereas the op amps are analogue. And yet both are made of transistors etc.


and this case, I think they're attempting to use it as a metaphor for the model they're campaigning for.

Metaphor or not, I think there is a real similarity between the way information is stored in a nucleic acid molecule (as a string of chemical units, each of which has just 4 possible states) and the way information is stored in a digital computer (as a string of electric units, each of which has just 2 possible states).

If I remember correctly, there is a theory (within the "metabolism-first" school) that the original function of RNA was not as a carrier of genetic information, but simply as a member of a team of catalysts — a co-enzyme. Which would perhaps bear comparison with non-digital functions of the transistor...

Paul Wally
2012-Dec-15, 02:17 PM
The metabolism-first versus genes-first question is a pseudo-problem?


Yes, metabolism is just chemistry. One could define an organism as a system of metabolism, so the firsts organism whatever it was, was a form of metabolism. A gene is just memory, so if metabolism replicates itself then that implies memory. Replication is memory, because what was replicated is a memory of what it is replicated from. So if metabolism replicates itself, then there is memory and therefore there is a gene. Hence, both metabolism and genes came first, they go together as one system of metabolism replicating itself. Therefore the problem of which came first is not a real problem, because it doesn't make sense to ask which came first.

TooMany
2012-Dec-15, 10:21 PM
There are "non-living" things called nanons that are self-propagating mineral proteins that have the appearance of ultra-small bacteria. They form spontaneously. They were originally claimed to be nano-bacteria but they have no DNA and apparently their structures are very simple. But they do propagate. There is something that makes this happen and accounts for the self-similarity when they propagate, but they contain no DNA or RNA. They are formed from calcium compounds and a protein called fetuin. They show up all kinds of places in animals. They are not life in accepted sense of the word.

Anyway, it just makes me wonder whether the origin of life started with something sophisticated (DNA or RNA), or whether some simpler information/reproduction processes came first, but DNA evolved later to revolutionize the information structure. RNA replicates itself with the help of enzymes. We just don't know how things got started.

TooMany
2012-Dec-15, 10:41 PM
An approach to understanding how life could start, might be to play with cellular automata or similar simulations in computers. Come up with some abstract atoms or molecules and assign them properties and rules of combination that allow them to form into bigger molecules. Not a full blow stimulation of organic chemistry, but something like that yet highly simplified.

Throw in some randomization and see what happens. Use a super-computer and try lots of variations. Select properties and rules that work best in producing something interesting. Use a "genetic algorithm" to make random variations in the rules/properties. Use other algorithms to watch for the development of interesting structures or self-replication. Maybe by randomly manipulating the properties and rules, more complex forms or even self-reproducing forms might arise.

Although this has little to do with the real world, the effort might provide some insight into how life can form. If it's extremely difficult to make anything interesting happen, well that's disappointing but tells us something about how special the rules have to be or perhaps about how much time it takes for the right random events to occur.

Colin Robinson
2012-Dec-15, 10:59 PM
Yes, metabolism is just chemistry. One could define an organism as a system of metabolism, so the firsts organism whatever it was, was a form of metabolism. A gene is just memory, so if metabolism replicates itself then that implies memory. Replication is memory, because what was replicated is a memory of what it is replicated from. So if metabolism replicates itself, then there is memory and therefore there is a gene. Hence, both metabolism and genes came first, they go together as one system of metabolism replicating itself. Therefore the problem of which came first is not a real problem, because it doesn't make sense to ask which came first.

That may be true if you use a broad enough definition of "gene" and "metabolism".

However, the issue of "genes first" versus "metabolism first" is a longstanding scientific debate.

Even if you think it is a pseudo-problem, the existence of this pseudo-problem can hardly be blamed on the paper we're discussing, and its use of terms "analogue" and "digital". Because the debate began long before this paper.

As I understand it, the debate has been basically about whether the history of life begins with the appearance of the first nucleic acids (the "genes first" view), or with catalytic systems did not involve nucleic acids (the "metabolism first" view).

Paul Wally
2012-Dec-16, 09:14 AM
An approach to understanding how life could start, might be to play with cellular automata or similar simulations in computers. Come up with some abstract atoms or molecules and assign them properties and rules of combination that allow them to form into bigger molecules. Not a full blow stimulation of organic chemistry, but something like that yet highly simplified.

Throw in some randomization and see what happens. Use a super-computer and try lots of variations. Select properties and rules that work best in producing something interesting. Use a "genetic algorithm" to make random variations in the rules/properties. Use other algorithms to watch for the development of interesting structures or self-replication. Maybe by randomly manipulating the properties and rules, more complex forms or even self-reproducing forms might arise.

Although this has little to do with the real world, the effort might provide some insight into how life can form. If it's extremely difficult to make anything interesting happen, well that's disappointing but tells us something about how special the rules have to be or perhaps about how much time it takes for the right random events to occur.

Yes, that it is the artificial life approach to abiogenesis. It's all about how self-replication emerges from a situation of non-self-replication; that transition process. Interestingly, I haven't yet seen even a very simple program doing that. Even just a simple program could give some conceptual understanding of the process.

I think it's easier to write a program that can simulate self-replication, but the problem is to let self-replication itself emerge without explicitly defining it and without "cheating" by somehow defining the rules in a clever way such that it's no real surprise when it happens.


That may be true if you use a broad enough definition of "gene" and "metabolism".

Yes, I'm using broad definitions. I do have a tendency to approach abiogenesis from the general to the particular; start with the simpler more general concepts and see where it leads. The other approach is to try and work backwards from our current particular state to it's past beginnings, i.e. to reverse engineer Earth-life. I'm not a big fan of the latter approach, but it does seem to be the most popular approach. I guess it's just different styles of thinking. Some like to immerse themselves in the details and particulars - entangle themselves in the complexity of a particular matter and thus lose sight of the broader fundamental principles.



As I understand it, the debate has been basically about whether the history of life begins with the appearance of the first nucleic acids (the "genes first" view), or with catalytic systems did not involve nucleic acids (the "metabolism first" view).

Ok, I see.