DNA Stability

[Resha Caner]

Depends on the randomness you are looking for.

OK, let me explain further, then. It has been shown that DNA sequences exhibit chaotic patterns. Simply stated, that means if you give me a DNA sequence and ask me to look forward and predict the next letter in the sequence, I can't (though a lot of people try). However, if I look backward over the sequence you've given me, I can identify patterns in the sequence.

This has been used to develop trees. It provides a way of distinguishing one sequence from another. So far, every DNA sequence that's been tried exhibits a chaotic pattern.

However, if you were to just draw letters from a hat to produce a random sequence, it won't exhibit a chaotic pattern. So, I wonder if junk DNA is random in this sense, or if it would produce a chaotic pattern. If so, would the pattern be the same as the coding region? Again, I would expect different patterns to have different structural characteristics.

It's an easy thing to check. Even I can write some code to do it ... at least I can do some simplistic comparisons. I get lost in some of the more advanced algorithms. My problem is if I pull something down from GenBank I'm never sure exactly what I've got. Further, I suspect someone has already looked at something like this, i.e. there's probably already a paper on it - but I haven't found one.

I would also like to bring up that position on a chromosome also impacts the likelihood of mutations; the farther from the center, the more vulnerable the sequences tend to be.

Would that distinguish junk & coding DNA or does it happen to both?

 

[Papias]

Caner wrote:

However, if you were to just draw letters from a hat to produce a random sequence, it won't exhibit a chaotic pattern. So, I wonder if junk DNA is random in this sense, or if it would produce a chaotic pattern. If so, would the pattern be the same as the coding region? Again, I would expect different patterns to have different structural characteristics.

Much of the non-coding DNA is debris from earlier DNA, accumulated over millions of years. For instance, pseudogenes make up around 20% of the genome. Pseudogenes are copies of working genes which, being copies made through the duplication mutation, are not needed, and hence often don't work due to some additional mutations (like changing a letter). Since nearly all of their base letters are still the same, they'd pretty much still show whatever patterns they showed before.

Similarly, around 10% of the genome is made up of the carcasses of virus DNA, which, for the same reason, would likely show whatever pattern the original virus DNA showed. etc.

So it seems that a lot of the junk DNA would show a lot of the same patterns, though likely degraded over time.

Just a thought.

Papias

 

[PsychoSarah]

Would that distinguish junk & coding DNA or does it happen to both?

Happens to both; junk is everywhere, you won't find a single human chromosome that isn't mostly junk, with just a few genes sprinkled throughout.

 

[Resha Caner]

Much of the non-coding DNA is debris from earlier DNA, accumulated over millions of years. For instance, pseudogenes make up around 20% of the genome. Pseudogenes are copies of working genes which, being copies made through the duplication mutation, are not needed, and hence often don't work due to some additional mutations (like changing a letter). Since nearly all of their base letters are still the same, they'd pretty much still show whatever patterns they showed before.

Similarly, around 10% of the genome is made up of the carcasses of virus DNA, which, for the same reason, would likely show whatever pattern the original virus DNA showed. etc.

So it seems that a lot of the junk DNA would show a lot of the same patterns, though likely degraded over time.

I'll have to think about this. You're right this could mean a pattern still exists, but I'll have to think about the extent. Further, I'd have to think about whether the structure would still be good if it's a jumble of discarded sequences. When I go to a junkyard, I can tell the junk was once cars, but that doesn't mean they're structurally sound cars.

Further, I would think evolution would have to posit a "building period" where the original sequnces were largely random, and were then downselected to those that work. If so, wouldn't some remnant of that randomness still be hanging around ... or, maybe it wouldn't be if random DNA isn't stable.

 

[PsychoSarah]

I'll have to think about this. You're right this could mean a pattern still exists, but I'll have to think about the extent. Further, I'd have to think about whether the structure would still be good if it's a jumble of discarded sequences. When I go to a junkyard, I can tell the junk was once cars, but that doesn't mean they're structurally sound cars.

Further, I would think evolution would have to posit a "building period" where the original sequnces were largely random, and were then downselected to those that work. If so, wouldn't some remnant of that randomness still be hanging around ... or, maybe it wouldn't be if random DNA isn't stable.

DNA is semi-stable, it has to be fragile enough that it can be split apart to be used and duplicated, hence many hydrogen bonds, which are relatively weak. It is held together by proteins that attach to the molecule and wind it up, and once these proteins are deactivated or removed, DNA is prone to falling apart.

 

[Papias]

Caner wrote:

I'll have to think about this.

Enjoy!

Further, I would think evolution would have to posit a "building period" where the original sequnces were largely random, and were then downselected to those that work. If so, wouldn't some remnant of that randomness still be hanging around ..

I think that that original sequences - the first self-replicating molecules - were very short, maybe only a few dozen bases or less. From there, functional sections would have sometimes been duplicated, making longer whole genomes, and duplication would hence build up the genome. After all duplication can double whole chromosomes.

So their may not have been any "original randomness" even as soon as a few million years from the start of life ~4,000 million years ago, and hence non left today other than randomness reintroduced from recent mutations. Just my guess though.

Papias

 

[PsychoSarah]

Caner wrote:



Enjoy!



I think that that original sequences - the first self-replicating molecules - were very short, maybe only a few dozen bases or less. From there, functional sections would have sometimes been duplicated, making longer whole genomes, and duplication would hence build up the genome. After all duplication can double whole chromosomes.

So their may not have been any "original randomness" even as soon as a few million years from the start of life ~4,000 million years ago, and hence non left today other than randomness reintroduced from recent mutations. Just my guess though.

Papias

Don't forget viruses and how they like to integrate themselves into DNA and even cause mutations. Hence why some viruses are associated with certain cancers.

 

[sfs]

Given what I do as an engineer, I'm curious about some of the structural and dynamic aspects of DNA. I've found papers that have studied certain structural aspects (beads on a string) and certain vibratory aspects. To me that implies some variation in strength along the DNA chain that might determine a higher likelihood of defects in certain regions.

You're probably better off thinking of DNA as what it is: a molecule. A really, really long, highly dynamic molecule. The DNA in your 23 chromosomes is about 2 meters long. It's bundled together with other molecules (proteins) and wrapped around still others and forms a million or more larger loops. The whole thing is being jostled by physical forces, pulled apart and rearranged by proteins and is constantly being broken and repaired.

"Your genome" is actually an idealization: you've got lots of slightly different genomes, scattered throughout your body, the result of countless somatic mutations in your cells. Different tissues accumulate different kinds of characteristic mutations, and yes, there are definitely regions of the genome that are more prone to different kinds. Repeated sequence may cause repair or recombination machinery to delete the intervening sequence, or invert it or double it, for example. Frequently transcribed regions are frequently pulled apart and exposed to mutagenic forces, so they tend to have more mutations. If the chromosomes in one of your cells become too badly damaged, the cell will commit suicide -- unless that machinery itself has been damaged by mutation, in which case the cell may start replicating uncontrollably and kill you.

There is a way of processing DNA to look for patterns called the "chaos game". I haven't quite wrapped my head around all the implications of the patterns it creates, but coding DNA produces definite patterns and random DNA doesn't. Again, to me that would imply one has a different structural makeup than the other.

Are you talking about patterns in linear genome space? If so, then junk DNA (which is a large proportion of noncoding DNA) forms clearer patterns than coding DNA. Junk DNA includes simple repeats and larger units that may be repeated hundreds or hundreds of thousands of times. This is in a way the opposite of random; that's why you frequently mask off "low complexity" regions of the genome from analysis, since they're not interesting. (Nor are they easy to analyze.) Mutations, especially substitution mutation, will gradually turn the simple patterns into random mush, however, so lots of junk DNA does also look random.

 

[Resha Caner]

I think that that original sequences - the first self-replicating molecules - were very short, maybe only a few dozen bases or less. From there, functional sections would have sometimes been duplicated, making longer whole genomes, and duplication would hence build up the genome. After all duplication can double whole chromosomes.

So their may not have been any "original randomness" even as soon as a few million years from the start of life ~4,000 million years ago, and hence non left today other than randomness reintroduced from recent mutations. Just my guess though.

This is why I struggle sometimes to separate biogenesis from evolution. The two seem to overlap in some spaces.

It is an interesting line of research - determining the shortest sequence that will replicate and sustain itself. However, as far as I know the shortest known sequences are still pretty long.

 

[PsychoSarah]

This is why I struggle sometimes to separate biogenesis from evolution. The two seem to overlap in some spaces.

It is an interesting line of research - determining the shortest sequence that will replicate and sustain itself. However, as far as I know the shortest known sequences are still pretty long.

Long perhaps in that the number seems long, but in molecular terms, not very large. DNA as a whole is a big molecule and has adapted as a big molecule: why assume modern DNA can sustain itself in as small of pieces as it could in the past?

 

[Resha Caner]

Are you talking about patterns in linear genome space? If so, then junk DNA (which is a large proportion of noncoding DNA) forms clearer patterns than coding DNA. Junk DNA includes simple repeats and larger units that may be repeated hundreds or hundreds of thousands of times. This is in a way the opposite of random; that's why you frequently mask off "low complexity" regions of the genome from analysis, since they're not interesting. (Nor are they easy to analyze.) Mutations, especially substitution mutation, will gradually turn the simple patterns into random mush, however, so lots of junk DNA does also look random.

Each code in the sequence is plotted as a point on a plane. There are rules determining how the code is plotted based on what codes preceeded it. If the pattern becomes very repetitive it will reduce to a "limit cycle" in very short order. However, most sequences produce a cloudy pattern.

For example, the sequence AAAAAAAAAA converges to a single point in the plane. The sequence ACACACACAC converges to 2 points in the plane. A random sequence just looks like someone spray-painted the plane. Most sequences form patterns of clouds somewhere between distinct points and complete randomness.

The seminal paper on this is: Chaos Game Representation of Gene Structure by Jeffrey, but there are multiple ideas about how to approach it.

So ... do you have a coding sequence and a junk sequence from the same genome I could use see what happens?

 

[Resha Caner]

Long perhaps in that the number seems long, but in molecular terms, not very large. DNA as a whole is a big molecule and has adapted as a big molecule: why assume modern DNA can sustain itself in as small of pieces as it could in the past?

I'm not. Papias was suggesting that's how it started. Maybe, but if it was too short to replicate and sustain, then that's not how it started. It's a question of what critical mass was necessary to kick the whole thing off.

For my question, the longer the better.

 

[PsychoSarah]

I'm not. Papias was suggesting that's how it started. Maybe, but if it was too short to replicate and sustain, then that's not how it started. It's a question of what critical mass was necessary to kick the whole thing off.

For my question, the longer the better.

Incorrect, a longer molecule is less stable than a shorter one (in the case of DNA) without the associated stabilizing proteins. Additionally, why would it have to originally sustain itself, when any environment that would allow its formation would also likely be an environment in which it was stable?

 

[Resha Caner]

Incorrect, a longer molecule is less stable than a shorter one (in the case of DNA) without the associated stabilizing proteins. Additionally, why would it have to originally sustain itself, when any environment that would allow its formation would also likely be an environment in which it was stable?

I think you've missed my point, as I'm not trying to say what you're infering.

Let's start with this. Suppose I have all the cellular machinery in place to copy DNA, and the sequence it copies into the embryo for the new organism is GCAT. Will that new organism be viable? Will it live? Grow? Continue to reproduce?

 

[PsychoSarah]

I think you've missed my point, as I'm not trying to say what you're infering.

Let's start with this. Suppose I have all the cellular machinery in place to copy DNA, and the sequence it copies into the embryo for the new organism is GCAT. Will that new organism be viable? Will it live? Grow? Continue to reproduce?

Just GCAT and no other sequences? That won't ever be an embryo. I can also tell you honestly that you couldn't stop more than that sequence from being copied no matter how hard you tried.

But, there are plenty of replicating entities that are exceedingly simple, they just aren't considered alive, for various reasons, but not in ways that couldn't make such things precursors to life. For example, living things have to process their own energy and metabolize. Viruses don't do that, but they reproduce and evolve. Same applies to things like prions, which are literally just proteins that happen to incite other proteins to take their shape (don't ask me how, I don't know, and I don't think anyone has really figured that one out). Prions are so tough that anything that would destroy them would pretty much destroy anything organic.

 

[Resha Caner]

Just GCAT and no other sequences? That won't ever be an embryo.

Exactly. So what about 6 codes? 100 codes? 1000 codes? What is the shortest DNA (or RNA) sequence you know of that has been observed to replicate and produce an organism?

But, there are plenty of replicating entities that are exceedingly simple, they just aren't considered alive, for various reasons, but not in ways that couldn't make such things precursors to life.

But what must be included in the system that allows these things to replicate? If you subtract something from the system, and the entity henceforth fails to replicate, then it's not a viable system. So what is the minimum system that will replicate?

For example, living things have to process their own energy and metabolize. Viruses don't do that, but they reproduce and evolve.

Only because they can cannibalize another system. So, if you want to define a "viral system" per the question I'm asking, it would need to include the host.

 

[Loudmouth]

Exactly. So what about 6 codes? 100 codes? 1000 codes? What is the shortest DNA (or RNA) sequence you know of that has been observed to replicate and produce an organism?

"Seven families of RNA ligases, previously isolated from random RNA sequences, fall into three classes on the basis of secondary structure and regiospecificity of ligation. Two of the three classes of ribozymes have been engineered to act as true enzymes, catalyzing the multiple-turnover transformation of substrates into products. The most complex of these ribozymes has a minimal catalytic domain of 93 nucleotides."
Structurally complex and highly active RNA ligases derived from random RNA sequences. - PubMed - NCBI

Somwhere around 100 nucleotides might be enough for a simple RNA ligase.

So what is the minimum system that will replicate?

No one knows, which makes the probability calculations trumpeted by ID/Creationists a bit hollow.

 

[Loudmouth]

This is why I struggle sometimes to separate biogenesis from evolution. The two seem to overlap in some spaces.

For 99.9999999% of biology, they don't. We don't need to know anything about abiogenesis in order to work out phylogenies for mammals or vertebrates. Even for work on reconstructing LUCA (last universal common ancestor), it really doesn't enter into the equation much.

 

[Loudmouth]

OK, let me explain further, then. It has been shown that DNA sequences exhibit chaotic patterns. Simply stated, that means if you give me a DNA sequence and ask me to look forward and predict the next letter in the sequence, I can't (though a lot of people try).

With the tools I have available, I could make very strong predictions. Without looking up the actual genome being sequenced, I could BLAST the genomes of closely related species and give you very good predicitons of what the next base would be.

However, if I look backward over the sequence you've given me, I can identify patterns in the sequence.

We can find patterns in all sorts of stochastic and spontaneous processes.

This has been used to develop trees. It provides a way of distinguishing one sequence from another. So far, every DNA sequence that's been tried exhibits a chaotic pattern.

However, if you were to just draw letters from a hat to produce a random sequence, it won't exhibit a chaotic pattern. So, I wonder if junk DNA is random in this sense, or if it would produce a chaotic pattern. If so, would the pattern be the same as the coding region? Again, I would expect different patterns to have different structural characteristics.

I am having a hard time understanding how this would apply to DNA. I guess I view it more from a mechanistic point of view. DNA is like anything else, it is a product of its environment and the applicable natural mechanisms. Its like asking if weather is chaotic or random. If it is random, does this mean that all places on the planet should get the same rainfall? How would you describe the chaotic and random nature of weather with respect to rainfall across the globe?

My problem is if I pull something down from GenBank I'm never sure exactly what I've got. Further, I suspect someone has already looked at something like this, i.e. there's probably already a paper on it - but I haven't found one.

What exactly are you looking for? What patterns are you looking for, and why would you expect to find them?

 

[Resha Caner]

We can find patterns in all sorts of stochastic and spontaneous processes.

Sure. Everybody and their mother thinks they have found the secret sauce, but rarely do any of those ideas yield anything. Still, recognizing and understanding patterns are a very important step in predicting behavior.

What exactly are you looking for? What patterns are you looking for, and why would you expect to find them?

I'm asking if the stability of DNA is related to the sequence such that we could put a measure on it and conclude sequence A is x% more stable than sequence B.

Second, if that were possible, I would be interested in comparing the stability of junk DNA to coding DNA.

With the tools I have available, I could make very strong predictions. Without looking up the actual genome being sequenced, I could BLAST the genomes of closely related species and give you very good predicitons of what the next base would be.

Of all the papers I've read about prediciting patterns in DNA, your claim seems the most confident. If I thought I could do it myself, I would. I'd like to better understand what BLAST does. My impression is that it uses some sort of fuzzy (statistical/empirical) basis similar to a neural net, but I'm not sure.

But since you seem to think it an easy thing, maybe you could try a few things for us here at CF:
1) What % confidence is there that BLAST will predict the next base correctly?
2) Does the prediction improve as you give it more related sequences?
3) Is the % confidence different for coding and junk DNA?