[Market strategy] Penalize GU heavy regions


  • For  Guanine and Uracil (K) stretches 6+ base of length, begin penalize and penalize exponentially.
  • If there are more such stretches, penalize harshly.
  • Only 1⁄2 penalty if these stretches are in static area and with a matching partner strand. They are still bad, but not doing as much harm, as if in the switching area.

Strategy background

I have been taking a look at what characterizes designs that tends to score low.

One thing stands out in particular. Too much G and U in line. Above 5 G’s and U’s in line, and it starts to go counterproductive. G and U has the letter K as a collective naming. The longer and the more K regions, the worse. They are worse in combo than if just U’s, although I think too much U’s also creates trouble.

U and G have an Oxygen sticking out at the end where they normally pair up. Which according to my favorite periodic table - (see under coolest hack), makes them electronegative beasts. :slight_smile:

I suspect them to act like splitters when in too big amounts.

Long stretches of them are endemic in anything scoring low and anything not winning. I have been looking at a few old switch labs where we do have feedback on the status of the individual nucleotide via SHAPE data. And if there are long stretches of U and G, they trend towards being unbound in both states. They simply tends to be nonreactive. Despite I can also find them in past winners, but even there they don’t seem to happy to bind. Example below. 

Target state

How it is estimated to have folded

These K stretches seems to be less of a problem in the static labs, especially if in longer stems. But there they usually also always have a matching C and A stretch to bind with. Which isn’t the case in RNA switches. There is a rather small region in easier RNA labs that needs to get moving, and long stretches of G and U, seems to prevent from getting the job done.

Although the K stretches seems to be of a little more help in the hard labs (Exclusion 5 and 6), that needs to get moving and thus not bound too tight in any sections. (They do tend to employ both more GU base pairs and also GC pairs than more easy going switch labs.)

1 Like