Hey guys,
I’m currently working on a bit of a presentation, and I’d like to hear your strategies for solving riboswitches for lab puzzles, in as brief a description as possible. Primarily, I’m looking less for specifics about sequence, and more about the actual generation of structures that fulfill the requirements. If you have questions or need clarification, please feel free to post!
I don’t use a lot of strategy. I just keep trying, and flipping different pairs until something works. Not very scientific, I know, but it gets the job done. Once stable, I use the much the same method for raising or lowering energy levels in one puzzle vs. the other.
Thought of something else. Sometimes I paste a second ms2 around until it snaps together with the first one. Then I weaken the bond that holds them together until it lets go in one.
I don’t know what the difference is between specifics about sequence and the actual generation of structures, but I can tell you this much:
My general strategy is to focus on fulfilling the stability and switching requirements. For this I only use GC and AU basepairs because I don’t understand the purpose of GU pairs in labs. I use GC pairs at stack ends for no good reason (the bad reason being because I feel like an AU pair is more likely to split at a stack end than a GC pair is, based on puzzle experience in vienna energy model), and when stacks end in a multiloop or open loop, I orient them the way Eli toad us to because he’ll always be the mayor to me. In short stacks, I use about a 2:1 GC to AU ratio because I know short stacks generally require strong bonds but I don’t like stacks that are exclusively made of GC basepairs. In long stacks it is closer to 1:1 becase I think the demand for bond strength of basepairs is not as high, but past experience from results that I’ve filed away mentally but can’t immediately verify seem to support this. I try to asymmetrically stagger basepair orientations and quad clumps to reduce repeated patterns in stacks in an effort to reduce mispairing between stacks. Finally, I throw in a few G-A mismatches in some internal loops with the attitude that doing so might encourage the formation of those loops to some degree. I don’t pay a whole lot of attention to the models’ prediction of stability, but I do take note if I keep seeing the same misfolds over and over and may consider changing things to remedy that if I think doing so won’t affect the other things I’ve already just described too negatively.
In order to promote switching, I’ll check my total energy values in the in-game models, and try to get the difference between these values to be about half of whatever the energy value of the switching mechanism is.
Consecutive base requirements I don’t worry much about, usually it’s just a matter of mutating a few default adenine bases to guanine or uracil here and there.
All other requirements I can think of off the top of my head I don’t understand very well at all, so I tend to ignore them.
Hope that helps
For this round we already know a lot about what might work for the FMN sub-labs, so I submitted some designs that are mods of past winners. In same state sub-labs I am trying for some designs that display symmetry between the MS2 and the aptamer. In the exclusions I am going for some designs with the MS2 near the aptamer on the optimal side, and even as part of the aptamer. I am also experimenting with alternative MS2’s in most of the sublabs to see if I can improve fold change. I also like to try and have at least one static stem. As for generating the structures, that usually starts with identifying shared or complementary sequences, not necessarily exact, and working out to the ends from there. Sometimes I will keep working to sew up loose nts, as long as it will still switch. Most of what I’m doing this time comes as a result of reading Eli’s CRISPR tutorial puzzle descriptions.
I position the MS2 hairpin where I want it, then decide how I want to peel it apart to solve the puzzle.If I peel from one of the ends I get different configurations than if I peel from around the center of the MS2 hairpin. I then reduce the energy difference between the two by flipping or rearranging nts to solve the puzzle. Peeling from the center appears to give more stable or symmetrical looking solutions where the nts don’t need to move a large distance in order to switch.
So, the modern labs require you to fill in the structure. Do you just fill in nucleotides until something sticks, or do you try for a predetermined structure?