I would file this under the “Own Goal” category. Because, once one digs into the mechanisms by which small RNAs work, one realizes that this system is actually a low-information (by ID parlance) way to dramatically alter gene expression networks, and thus of growth and development. The implications vis-a-vis evolution are obvious, and not at all kind to the standard ID party line.
Low information is exactly right @Art. Can you explain this further for people to understand?
I’m going to take a stab at what I think Art is saying and he call tell me if I am mistaken. The RNA sequences are themselves rather short. As such their information carrying capacity is less than that of longer sequences.
But how much information carrying capacity do you need in order to perform a simple on/off function? So whether it is low-information or high-information doesn’t seem all that relevant to me. We would need to look at the system as a whole, and not just one part of it.
Exactly 1 bit of information and thus there is exactly 1 bit of entropy in such a system.
I have done some work with micro-RNA (miRNA), but information theory is something I am very ignorant of. Looking at how miRNA works, is it possible to measure Shannon information in any meaningful way? The reason I ask is there is a sender-receiver relationship found in miRNA function.
In the majority of cases, miRNA down regulates translation by binding to the 3’ untranslated region (UTR) of the mRNA through complementary bases. This means there is a direct relationship between the base sequence of both the miRNA and the 3’ UTR of the mRNA. Mutations in either the miRNA or 3’ UTR region of the mRNA can affect which miRNA binds to which mRNA. Can this be modeled as a sender/receiver relationship?
I can try. Basically, microRNAs bind to other target RNAs by canonical base-pairing, and in so doing bring other functionality (mRNA degradation, translational regulation, to name two) to the target. This happens because microRNAs are guides that are associated with protein complexes that mediate these effects. Depending on the organism and target, productive base-pairing may involve between 8 and 12 bases, situated on the 5’-half of the miRNA (that will be between 20 and 24 nucleotides in length). The relatively modest sizes of the target recognition sequences means that the dynamics of regulatory networks can be easily altered during evolution by the appearance, or disappearance, of microRNA targets in mRNAs. For the audience here, this also means that this sort of remodeling of regulatory networks requires almost no information (again, in the colloquial ID sense of the word - I am not going to pretend that I can convey any sort of proper information theory sense to this).
I would say first that the mathematics of “Shannon information” works regardless of whether there is any model of a communications system. IOW we can dispense with sender-receiver as any sort of limit on the applicability. That just happens to be the historical context in which it arose. I am speaking loosely here. When we start talking about coding and channel capacity and noise then perhaps it becomes more relevant.
There is another field of information theory that deals specifically with sequences, algorithmic information theory. So it’s important to keep in mind which one is being used.
Well, now I am confused. So the “low-information” comment had nothing to do with the length of the sequences involved?
Could you expand on your comment that “this sort of remodeling of regulatory networks requires almost no information (again, in the colloquial ID sense of the word,” because you appear to be conflating two different senses of the word “information.” More specifically, what is “the colloquial ID sense of the word” and how is it unrelated to the length of the sequences involved?
No there isn’t another field of information theory. There is just one field of information theory and it is called information theory. Here is a link to the IEEE Information Theory Society. You can look but you can’t join as you don’t meet the membership requirements.
Just a small point of clarification - when I refer to the “colloquial ID sense” of information, I refer to the inevitable conclusion of any discussion of information by ID proponents, namely that information is something that cannot be created by evolutionary mechanisms. I think that, while certainly not a technical term, this is a pretty fair representation of what information really is in ID circles. I would ask that we not stray here into debates or discussions about the many, many flavors of information and theory that might be (yet again, for the umpteenth time) explored. These are not pertinent to my point.
I don’t understand then what the length of the RNA has to do with it. Could you elucidate?
Maximum, there are 2 bits of information per base. Of course, the real value is much lower. This means, nonetheless, that the max amount of information is bounded by length of the of the RNA.
But not the maximum amount of ID colloquial information (IDCI), which, as defined by Art, has nothing whatsoever to do with the length of the RNA, given that it [IDCI] is defined not by information carrying capacity in the Shannon sense, but rather by what evolutionary mechanisms cannot do.
I’m just left wondering what Art’s original post would look like if he took out the information sciencey sounding bits that are, as he says, not pertinent to his point. (Such as the length of the RNA.)
@mung this is a classic example of goal post shifting and begging the question. I really hope you are joking. If not, please stop speaking authoritatively about this highly important and technical topic.
If you think I have taken a misstep here please be specific and I think we can work it out, not only to my own benefit but also to the benefit of others who may be watching. Vague generalities are not helpful.
Let’s not overlook the fact that I give a clear description of what I thought Art was trying to convey in bringing up the fact that these are short-length RNAs and thus in his words “low-information (by ID parlance).” For whatever reason Art didn’t say whether I got it right or I got it wrong. Patrick seemed to be saying that I got it right but you seem to be saying that I got it wrong.
Later Art goes on to explain what he meant and it has nothing at all to do with what he said previously.
Here’s Art’s concept of the ID concept of information:
Perhaps that belongs in the thread on the ID definition of information.
But back to your original point. I do not speak as an authority on Information Theory. But I would propose that if I say something that is incorrect the best thing to do is point out that it is incorrect and why it is incorrect.
I am more than willing to work with you.
not really. The correct term would be that a short sequence has low entropy.
I got to tell you that I was almost bamboozled by Eric and you. I have an real rigorous Bell Labs Shannon taught background and degrees in real Information Theory. When Eric arrived here with his freshly minted PhD in Information Theory from Baylor Dept of EE and CE, I thought I was Issac Newton in a General Relativity class. I knew my fundamentals were rock solid but 40 years old. Surely Eric would be operating at a level that benefited from 40 years of progress in Information Theory. But Eric (and you) made absolutely no sense to me. I thought that the field advanced so much that I couldn’t follow it anymore.
But then Dr. Swamidass did his tutorial on definitions using Venn diagrams. This was basic to me. This was primary Information Theory. of course I could understand it, it was like riding a bike. So I wasn’t a fossil in Information Theory, I was an emeritus Information Theory expert (PhD, IEEE Fellow).
When Eric (and you) couldn’t grasp information was entropy and didn’t know Shannon’s 1948 paper line by line, I knew instantly what had happened.
So now, I am proclaiming that I am the Information Theory expert at PS who has the genealogy to Claude Shannon. I am ready to take on Eric at all website that he is posting at. He will be seeing me there soon.
And what I wrote is that “their information carrying capacity is less than that of longer sequences.” Was I wrong?
If I said that a short sequence has low entropy and a long sequence has high entropy would anyone been the wiser? Would people be telling me I was wrong to say that as well?
No, but it trivially obvious to someone who knows Information Theory that a shorter sequence has less information carrying capacity than that of longer sequences.
A large pale holds more water than a smaller pale. It is trivially obvious.