Single to switch try#2 Lab followup.

  • 1
  • Article
  • Updated 6 years ago
Nando asked some questions on a lab that reminded me that a writeup was necessary.
Lab in question: Single to Switch try #2

The design in question is unique in that the molecule NTs are outside of the stacks
and the stacks are nice and long.

When I reviewed past switch labs several questions come up:

the obvious-what makes a good switch?

1-can you get better results with GUs in first state?
2-is one second state configuration better than another?
3-can a super clean dot plot give a higher switch score?
4-does a 3-D catcher mitt design produce better switches?

Number #1 and #2 discussion:
my "Single to Switch" design had the privilege of coming in flat out last place. A 77.
a failure, well......... not exactly.

The top 100%ers had no stack GUs, at most 1 GU.

Looking at the GU percentages:
The conclusion looks like: the more stack GUs, the lower the score.

So why the need for GU pairs at all.

It always looks like the GCs control the puzzle, but is it really the GUs
manipulating the pairings instead?

When writing my puzzle solver script I noticed that GUs aren't even needed to
solve most puzzles. So again, why are GUs even there?

Maybe like 3 siblings. The GC is the oldest and the bossiest.
The AU, the youngest, hangs around for the ride.
And the middle one, the GU,
always needs to manipulate the other two to get what it wants, not pushy enough to do the job itself..

Then is the role of the GU (specifically the GU/GC interaction) that helps the RNA
form into it's various functional roles?

The few GUs = lower # tricks an RNA can do. Um.

Back to the experiment.
The second part of the experiment is yet to be run.

Behind the "Single to Switch" lab I have around 6 switches modeled from this original
configuration in different GU percentages and different configurations.
All 6 switches would be submitted and hopefully run together so that the results can be quantified.

Or backwards

Number #3 discussion:

When reviewing past switch lab designs I noticed that top scoring designs had
pretty good Dot Plots but no "Perfect" Dot Plots.

So the question arises: can a perfect switch dot plot make for the top scoring design.

The criteria would be a dot plot with the first state at 100% level and the second state
with exactly 90% level or at 80% level. With zero mismatches, none, zippo, nada.

I've designed around 15 switches that meet these requirements using an online dot plotter
and would submit these for the next switch lab.

Nice Dot Plot

Number #4 discussion:

The 3-D catcher mitt design is still a work in progress but the picture below is
so cool I have to persue designs, then find a way to "track" the seqments into the
mitt. Is this a correlation between the "mitt" and lab score?

The bottom line: I have a slug of switches ready to be dumped upon the next
unsuspecting switch lab. HeHeHe.
Photo of JR


  • 241 Posts
  • 20 Reply Likes

Posted 6 years ago

  • 1
Photo of eternacac


  • 274 Posts
  • 19 Reply Likes
I would think that perfect dot plots could make the 100 score switch, we don't see them because our designs aren't that good as a rule.

You mention an online dot plotting tool that seems to show each switch conformation in the same plot, color coded. That would be a nice addition to our dot plot methods.