RNA in - RNA out Puzzle Analysis

  • 4
  • Article
  • Updated 3 years ago
This is a new thread for analyzing RNA in - RNA out ouzzles, starting with the puzzles in R98 and R99

We are very excited about the results, which have been posted in Eterna and also summarized in a PDF (https://drive.google.com/file/d/0B_N0...)

and Google Docs
(Eterna">https://docs.google.com/spreadsheets/... R98 RNA in – RNA out).
(Eterna">https://docs.google.com/spreadsheets/... R99 RNA in – RNA out (Round 2).

Looking forward to hear your thoughts!
Photo of johana

johana, Researcher

  • 96 Posts
  • 45 Reply Likes

Posted 4 years ago

  • 4
Photo of Eli Fisker

Eli Fisker

  • 2236 Posts
  • 495 Reply Likes
Mir in, reporter in - Short reporter - 98

State1: 0 oligo binding

State 2: 2 oligo binding

Easy, oligo complement turns oligo complement off


This lab gets far more winners than its round twin.


The structures of both jandersonlee’s and Omei’s designs are in the majority of the winning designs.


ININshortjpg


Omei’s designs have better KD numbers. His designs keeps a static stem in between the oligo complements. Jandersonlee’s type have the oligo complements pair up directly.


It looks like oligo placement is that TB input wants to be early than the static stem and the reporter later. But both design types are possible, but it seems that the static stem is beneficial in this lab.


There are several minority types.



miRNA-in, reporter-out - Short reporter - 98


State 1: Reporter binding

State 2: TB B binding

Oligo exclusion type design, harder


The oligos are working in an exclusion manner. One oligo partly sharing lane with the other, so they kick each other out mutually.

This lab has far fewer winners than the miRNA-in, reporter-in lab that also have short reporter. I wonder if exclusion type labs are somehow harder.


The majority of the winning designs follow Omei’s design patterns.


INOUTshortjpg


There are only 2 of the minority type among the winners, but I’m dragging them in - not just because they are mine - but also since the pattern for its second state resemble the minority type pattern in the sibling round 2 lab, mirRNA- in, reporter-out of the long reporter type.


miRNA in, reporter in - Long reporter - 99

State1: 0 oligo binding

State 2: 2 oligo binding

Oligo complement binds with oligo complement, harder


Generally there are far more varied possible structures than usual for microRNA input labs. I’m guessing that while the reporter which is 20 nt long and of similar length to the MS2 (19 nt), is far weaker and thus probably less demanding in relation to how it is placed as long as it can hook up and get turned off. MS2 is harder to dance with, but then on the other hand far seem more consistent about what it wants.

So I can’t say much about preference of oligo binding sites on the static stem, as they shift. However one thing is generally in common among the winning designs.


1) A static stem between the stems which I find interesting since I have speculated about such earlier in my microRNA analysis.


2) If there is no static stem, then having the 2 oligo complements fairly close works as well. So far it seems like the preference is a relative short stretch of single bases in between.


In both cases the oligo complements tend to turnoff each other in state 1. So basically it seems that if there is a stem in between or just a single stretch of bases - it works. :) 


Broad general types that seems to work:

ININlongjpg



MiRNA in, reporter out - Long reporter - 99


State 1: Reporter binding

State 2: TB B binding

Oligo exclusion type, easy


Again there are many structural minority variations with legal ways of solving. Again it is possible to place both the reporter and the TB quite differently.  


But Omeis designs are clear in the majority. Nicely followed by a dozen Eized mods.

INOUTlongjpg


Minority type of Eized mod


Partial Overlap - sharing lanes, score

http://www.eternagame.org/game/browse/6171039/?filter1_arg2=6225504&filter1=Id&filter1_arg1=6225504


Yay, the exclusion strategy works! This is the first time I recall have seen partial lane working in lab - for real. :) I first saw it in eternaBot’s solve of one of the XOR puzzles.


I was starting to get my doubts, because I saw how bending between oligo and oligo complement were generally counter productive in the logic gates labs. And this was only slightly bending that seemed to cause trouble.

However in the reporter labs, I see extreme bending also in winners, sometimes even bending in both states. And split oligo binding site, just like in the microRNA turnoff labs.

However it seems as if bending can be very beneficial, especially when an oligo needs to be kicked out by another. I think we can use and abuse this fact, in the future when needed. :)


General thoughts

  • Asymmetric internal loops seems to be used for destabilizing and helping switches - this is something I have been seeing in some microRNA labs too.

  • Bending both between oligo, but also elsewhere are used particularly in state 1 for kick off of oligo :)

  • Dangles still present but mostly aimed at only one oligo if there are two inputs.

  • There are far more structure variation among the winners, for these labs, compared to what is usual for MS2/FMN labs. I’m guessing it is because the reporters are not MS2. :) Both reporters are both far weaker, leaving more options for alternative pairings. Whereas MS2 is far more specific in what it wants, but on the other hand also leaves fewer options open.


In both reporter type labs, the round 98 has better KD. I’m guessing it has probably to do with the length of the reporter, since the miRNA-in, reporter-out labs have a similar difference between round 98 and 99. So it is easier getting a long and short oligo switch than two longer oligos. It is easier making a few bases switch compared to many.
(Edited)
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 980 Posts
  • 308 Reply Likes
@johana: For these experiments, was it the input or the reporter RNA that was experimentally varied?  And what was the concentration of whichever one was held static? 
Photo of johana

johana, Researcher

  • 96 Posts
  • 45 Reply Likes
The reporter was varied from 0.025 nM to 3000 nM and the binding curves represent that. The two states for the switch were determined by the input which was either absent or present at 100 nM.
Photo of Eli Fisker

Eli Fisker

  • 2236 Posts
  • 495 Reply Likes
Things to watch out for

I have been saying that in general bends between oligo and oligo complement seems to be counterproductive. That is unless it is a lab where one needs to kick out one oligo for another, as in the two microRNA-in, reporter-out labs.


However I noted something interesting. I have been wondering about the one of my designs in the OR lab, that did get fairly well away with a bend between the oligo and oligo complement.


Score 85%

http://www.eternagame.org/game/browse/6096395/?filter1_arg1=6134890&filter1_arg2=6134890&fil...

Now I notice a similar pattern with the designs in the kickout labs. (microRNA in/Reporter out)


There seems to be something about a specific spot the bend pops up. Right around the strong G magnet segment. Which kind of makes sense. It is likely the strongest binding spot, so it might have a harder time to let go. I see this pattern turn up elsewhere. It’s endemic in the microRNA in/Reporter out labs, which are exclusion labs where one oligo needs to kick the other out.


Winner by Jieux, score 100%

http://www.eternagame.org/game/browse/6171039/?filter1_arg2=6235914&filter1=Id&filter1_arg1=6235914


And as I started to suspect and wanted to know, a G magnet dangling ends works just as well as pyrimidine end dangles.


Instead of a bulge bend it can also be a mismatch in or next to the magnet segment.


Winner by Omei, score 100%

http://www.eternagame.org/game/browse/6171039/?filter1_arg2=6235869&filter1=Id&filter1_arg1=6235869


Not all designs have magnet segments in their oligo sequences - it depends on the sequence inputs. However as can be seen, bending can also be done in both states and with internal loops between. Should be used with much care if not in an exclusion type lab - as too much bending seems to result in bad scores.


Mat mod, score 98%

http://www.eternagame.org/game/browse/6171039/?filter1_arg2=6225504&filter1=Id&filter1_arg1=6225504


Here I think the reason why the real big loop gap in state 2 is tolerated is because this sequence has far more strong bases than the sequence in state 1. So the gap is giving a huge disadvantage to the stronger state 2 sequence to leveling the ground between the inputs.

It even happens in the earliest of the miRNA-in, reporter-out labs.


Winner by Omei, score 100%

http://www.eternagame.org/game/browse/6116601/?filter1_arg1=6142984&filter1_arg2=6142984&filter1=Id


There are no clear magnet segments here. But the bend turns up exactly at the shifting spot where the A and B segments share lane. So the bend is aiding the shift between states. Also notice that the bend turns up between the long sequence (B) and not the shorter sequence A.


The oligos are of uneven length. One much shorter than the other. A bend can be introduced in the longer of the oligos as a way to evening out the playground and get a better balance between the switching partners.


The shorter oligo need all the pairing up it can get, whereas the longer can live with a weakened grip because it is stronger and can get a longer stretch of pairing. I think it is also why the bigger gap between the oligo segment in state two is tolerated.

(Edited)
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 980 Posts
  • 308 Reply Likes
I added the switch graphs and made fusion tables for the Round 98 and Round 99 results.
Photo of Eli Fisker

Eli Fisker

  • 2236 Posts
  • 495 Reply Likes

Crossed GU in reporter labs


I noticed that crossed GU's and excess GU in general was highly present in a good deal of the logic gate labs. That made me look back to earlier two input labs to see if they followed a similar trend.

The tails in the first state needs to split. I suspect the crossed GU is helping. Also as this one don’t hold any internal loops as the cooperativity labs naturally do when their two MS2’s pair up with each other for turnoff.


miRNA-in, reporter-in, Round 2


96%, fc 572, fce 1.16

http://www.eternagame.org/game/browse/6171040/?filter1_arg2=6228422&filter1=Id&filter1_arg1=6228422


Actually the designs that score well in this round have a rather high amount of GU’s in their long static stem, compared to what is usual.


As can be seen it has a rather high GU content in general. 10 GU’s are normally not the usual amount for a lab.


Round 2, long reporter

http://www.eternagame.org/game/browse/6171041/



miRNA-in, reporter-in, round 1


Score 100%, fc 88, fce 1.09

http://www.eternagame.org/game/browse/6116602/?filter1_arg2=6153838&filter1=Id&filter1_arg1=...

Omei even have the crossed GU’s in the static stem area not involved in the switching. This one and its sibling scored real well.


100%, fc 354, fce 1.10

http://www.eternagame.org/game/browse/6116602/?filter1_arg2=6143292&filter1=Id&filter1_arg1=...


Round 1, short reporter

http://www.eternagame.org/game/puzzle/6116602/


The round 1 lab doesn’t call for nearly as many GU’s, in the switching area, if any. Likely because the reporter is short so there isn’t space or need for it.


Whereas the round 2 with the equally long reporter and input often actively uses them for splitting up the pairing between the reporter and input complement binding with each other. So there the GU’s goes both in static stem and the switching area.
Photo of Eli Fisker

Eli Fisker

  • 2236 Posts
  • 495 Reply Likes
While I was looking at Omei’s designs in the round 1 of the miRNA-in, reporter-in lab data, I took notice of a certain pattern in relation to a few similar designs, that seems to be related to the extreme high fold changes.

I have been wondering if switches would benefit from weakening of the end of other stems than necks. This holds relation to Salish’s end bit investigation.


Omei shifted the closing base pair for 3 designs going over GC, AU and GU. The weaker ones receive a far higher fold change than the stronger ones.


Score 100%, fold change (fc) 354, fold change error (fec) 1.10

http://www.eternagame.org/game/browse/6116602/?filter1_arg2=6143292&filter1=Id&filter1_arg1=6143292


Score 100%, fc 324, fce 1.10

http://www.eternagame.org/game/browse/6116602/?filter1_arg2=6143296&filter1=Id&filter1_arg1=6143296


Score 100%, fc 217, fce 1.13

http://www.eternagame.org/game/browse/6116602/?filter1_arg2=6143273&filter1=Id&filter1_arg1=...


The weaker closing base pair for the stem in state 1 may help allow the stem split a bit in state 2. The GU may be doing the splitting job a little too well, since AU result in a higher fold change. The design with a GC closing still do rather good.


I find all this worth noting and repeating.