Accuracy of the lab results in loops

I’ve noticed in my design (…) that my loops were evaluated as partly base-paired.

I’ve used UUCG and GAAA loops that are known to be stable, so it is not unexpected that they might be not as accessible as less stable loops. But this result is probably an artifact of the method used to determine secondary structure, the loop may be stabilized, but it still is a loop.

So to get the maximum score I would have to deliberately destabilize the loops to work around the quirks of that assay.

Hi Fabian. I can’t comment on the accuracy of the results as I’m not familiar with the method, however:

  1. for the one left arm that was mostly unformed, you accidentally reversed the closing CG basepair of your UUCG tetraloop. See dimension 9’s new tetraloop reference here, which explicitly denotes the identities of the closing basepair.…

There are experimental references in that thread which also show the drastic effect of reversing the closing CG pair, so this is probably a realistic feature of eteRNA.

  1. The UUCG tetraloop is hypothesized to be ultra-stable because of a non-canonical UG wobble interaction (this is an unusual non-watson crick base pair, i.e. something not covered by the ETERNA energy function). The G ends up being flipped upside down and backwards and is very conformationally restricted, so I’m not surprised that the SHAPE analysis tagged the G as being inaccessible.

The image below is from a recent structural analysis of UUCG tetraloops doi: 10.1261/rna.866408

He’s talking about the other loops, where the G’s show up *blue* despite the shape being (probably) perfect.

I am also curious whether a G in a tetraloop turning blue represents a score penalty, and on a somewhat related note, whether it actually represents a flaw in the shape of the sequenced RNA…

@ccccc; Yup, me too, at least in part. The upper loop in his design is a UUCG with the G marked blue, I’m hypothesizing here that many loop stabilizing G’s will show up this way because they are in fact not accessible as they form a real interaction, but it’s not of the watson-crick type so it’s not shown in the 2d plot. So the reason we are adding them in the first place is consistent with the reason they show up this way experimentally.

The original SHAPE paper also indicates one can expect this behavior, this is from doi: 10.1038/nchembio010

Note the low accessibility scores at the same place in every loop:

Thinking about it some more, I agree that the score should probably be manually adjusted in such cases, because we WANT people to make good loops, right?

I noticed in the latest round of lab results that we no longer seem to be penalized for these loop-stabilizing Gs showing up as bonded in the results (in the scores; they still show up as blue on the graphic of course).

Thanks, EteRNA team!

Is this still the case that loop-stabilizing Gs are showing up blue when they should not in the shape graphic?

There is an older thread on why loops sometimes appear to be misformed when it is just partial bonding that adds to the loop’s stability. It seems most pronounces in quad loops. I’m not sure that the SHAPE results can distinguish this case.

Perhaps Adrien or Jee or Rhijuu would know if they compensate for this, but I haven’t seen any evidence to support that assumption.