User:ElNando888/Static Cloud Labs/CL-VII.Semicircle

 

 

Candidates

For my first candidate, I did my usual routine, searched FRABASE, found a hit, and ran Vinnie with constraints. It's only now that I realize that I neglected to check the PDB entry more carefully.

 

 

It would seem that Magnesium ions are required for this fold to happen. Or maybe, they are just helpful, and not absolutely necessary. We will see.

----

The second candidate, "Theory", is an experiment related to some insane ponderings of mine... Does anyone even have a clue what I'm getting at here?

 

 

Links

• Lab: http://eterna.cmu.edu/web/lab/3102650/
• Designs:
  1. http://eterna.cmu.edu/game/browse/3102650/?filter1_arg2=3190959&filter1=Id&filter1_arg1=3190959
  2. http://eterna.cmu.edu/game/browse/3102650/?filter1_arg2=3210385&filter1=Id&filter1_arg1=3210385

 

Results

The "Theory" experiment failed, so it seems my hypothesis is clearly falsified. Just for the sake of confirming this result, I'm submitting another sequence with swapped GC as closing pairs in the next round.

 

The other design, well, it is hard to say what happened, because the results in this lab are globally skewed by the SHAPE threshold.

The most visibly affected area is the barcode. I suspect that the processing of the raw SHAPE data includes more or less automatic processes, one of them being the determination of the threshold, which is simply deciding where is the bar below which a base is declared protected and above which it is declared reactive. In this lab, the long stacks obviously created a context where a majority of bases have an extremely low reactivity. The "poor" barcode in contrast can only have a low reactivity... 

Do stars stop shining when the sun rises? Of course not, but you can't see them by daylight. In this lab, the barcodes did fold correctly, but we can't see it, because the "shine" of the rest of the data blinds us.

Edit: This said, there seems to be a pattern here: of the 4-6 designs that did get a rather clean protection signal in the barcode, 4 show clear signs of relative weaknesses in the main design: Zanna used many GU pairs in two of her candidates, and Brourd and I attempted noncanonical combos, a AA/AA and a UC/UC. So it would seem that, for a decent barcode to appear in this lab, something had to be wrong elsewhere in the design. Isn't it strange that the weaker some areas of the candidate design, the better the signal in the barcode?

Thinking a little more about it, I wonder if this effect wouldn't have actual chemical origins (as opposed to a numerical artifact from the data processing). I mean, whether the chemical probe 1M7 attaches to O2' on mobile nucleobases must be mainly a function of two important factors: 1. how mobile the base is, and 2. how (locally) concentrated the chemical probe is. With a "normal" RNA sequence, a portion of the 1M7 present locally is "absorbed" by the reactive bases, and over time, the local concentration of 1M7 decreases because of that, and because it self-hydrolyzes in water. Now, with RNAs like the ones we made in this lab, the level of protection of the vast majority of the sequence is huge, the concentration locally keeps elevated all around the RNA molecule, and maybe, it is only a matter of time before the 1M7 probes find this brief instant of relaxation in the comparatively weaker barcode hairpin, resulting in a much more reactive signal there than should normally be.

Edit: After some more pondering, I do not believe anymore in a numerical or chemical artifact. I'm currently exploring another hypothesis, and I hope that the Don Quixote lab I recently proposed will shed some light.