This week the developers added a new category of lab puzzles under the name RNA in – RNA out. The descriptions for the two new puzzles are currently miRNA-in, reporter-out and miRNA-in, reporter-in. I didn’t have any trouble understanding the intention of the miRNA-in, reporter-out name, but when it came to miRNA-in, reporter-in, I had to (and still do) go through some mental gymnastics to relate that name to the puzzle.
From my perspective, the essence of the difference from the previous logic gate puzzles is that the output of the puzzle is now an unbound miRNA molecule (the reporter) instead of a bound hairpin structure. That makes me think that it might be useful to unify the way we talk about (and hence think about) the two types of puzzles by continuing to identify both types as logic gates, with just the form of the output being different.
Viewed in that light, with the obvious encoding of molecular presence/absence to True/False, the miRNA-in, reporter-out puzzle is a simple 1 input, 1 output identity gate and the miRNA-in, reporter-in puzzle is a 1 input, 1 output inverter.
I agree that these new targets can be classified as logic gates, just like all the other currently active ones, actually. From the experimenter point of view though, the instrument is still the same, and will still be limited to one measurement: the amount of light emitted by the clusters. If the presence of light is 1, and its absence 0, then the miRNA-in reporter-in is the YES/TRUE/identity operator, and the miRNA-in reporter-out is the NO/FALSE/inverter operator.
I suppose it depends on how we frame the problem. Here’s the way I’m thinking of it.
For the miRNA-in, reporter-out puzzle:
Input: TB RNA (True if and only if the TB RNA is added)
Output: Reporter RNA (True if and only if the reporter RNA increases after the gate responds to the input)
Initial gate state is the switching RNA bound with the reporter RNA
Framed this way, if the input is True, the output should be True. Hence it is the identity function.
For the _miRNA-in, reporter-in _puzzle, the initial gate state is the unbound switch RNA. If TB RNA is added (representing True), the reporter RNA decreases (representing False). Hence it represents an inverter.
Do you have a different way of framing the mapping from experiment to interpretation as a logic gate?
The reporter RNA is the one that will have fluorescence attached to it, and it’s the light emitted from the reporter RNA that is detected.
miRNA-in, reporter-out =
* in the absence of TB (0), reporter is bound, there is light detected in the cluster (1)
* in the presence of TB (1), reporter can no longer bind, there’s no light in the cluster (0)
To me, this is an inverter, not an identity. Again, I’m talking about the experimental context and instrument. Johan’s instrument only sees light, and I would think that it is more natural to treat its presence as 1/TRUE and its absence as 0/FALSE, rather than the other way around.
Ah! Now I understand where our mental models differ. I had been assuming that the reporter RNA, when not bound to the switch RNA, would bind to a florescent protein, creating the light. But it sounds like it is the opposite; the reporter RNA is covalently bound to the florescent protein (not shown in the puzzle) and when the the reporter RNA also binds to the switch RNA, we get light.
Is that the way the chemistry actually works? Or some other variation that has the same effect? If so, then yes, I agree with your identification of which is the inverter.
Well, Johan could give us the last word on this, and if it turns out that I’m wrong about the fluorescence being covalently attached to the reporter RNA, then you will be the correct one here, of course
This said, this is a detail compared to what you were stating about logic in general, which I completely agree with. In fact, I very recently submitted to Rhiju that these puzzles strongly remind me of a nowadays nearly forgotten branch of mathematics and computer science, that used to be “en vogue” a couple decades ago: “fuzzy logic”.
These days, most computer scientists are taught that ANNs advantageously replace this type of computations, and it’s only in Asia, mostly in Japan, that you find related applications, mostly in control systems used for instance in appliances (the best rice cooking machines are “fuzzy logic” ones)
Maybe I’m deviating a bit from your original intent in this thread, but I wonder if some form of know-how about “fuzzy logic”, or simply some theorems in it, could find an application in what we’re attempting to do…
This does strike a chord with me. Fuzzy logic (as I am sure you know) is based on the mathematical concept of fuzzy sets, and fuzzy sets generalize conventional sets by assigning each member of the set a number from 0 to 1 that represents that element’s degree of membership. So for example, a specific dead, but still standing, tree might be assigned to the set of all trees, but with a membership value of only 0.4 instead of 1.0.
I bring this up because I am working toward a model for RNA switching which is based on identifying discreet base subsequences that form “attraction pairs”. For example, any base sequence forms a perfect attraction pair with its reverse complement sequence, but imperfect matches form attraction pairs also. So treating the “set of all attraction patterns in a RNA” as a fuzzy set, and using fuzzy logic to find the best matching of attraction patterns for a given puzzle, could well get better results than defining a sharp Yes/No criteria for what constitutes the classical set of attraction patterns.
Nando, another thought about whether the In/Out or the In/In puzzles should be seen as the inverter.
Thinking of the riboswitches switches as boolean logic gates is a convenience for us mostly because we have a lot of experience applying that theory to computing devices, and the main power of that theory is in building complex computational structures out of a handful of basic gates.
With this in mind, it seems helpful to categorize our RNA molecules as being either switches (the riboswitches we’re designing) or “wires”, such as the TB and reporter RNAs. These wires could represent input, output or intermediate values. Each wire would be represented by a unique RNA that is distinct enough that cross-talk between wires would be negligible. And by convention, we would probably choose to have the presence/absence (at least higher/lower concentrations) of each of these “wire” RNAs represent True/False.
So what happens if we start connecting In/Out gates in series? With the proposed convention, the value of the output of the second and third gates are the same as the first, so the In/Out behaves as though it were a pass-through gate. On the other hand, if we connect In/In gates in series, the output values of odd numbered gates have one value, while the those of the even numbered gate have the other value. So the In/In gate behaves as if it were an inverter.
I think the reporter RNA, and its association with the light/no light condition, is a special case. It combines an RNA with a protein (in some way that you and I aren’t sure of at the moment) to produce light, but we could equally well assign True to No Light as to Light. It’s just another convention.
My conclusion is that we’re better off in the long run (assuming we are looking forward to building more complex logic out of these elementary gates) to think of the In/In gate as the inverter, regardless of whether the light goes on or off when the reporter RNA is bound in the single-gate case.
