Detailed RNAFold analysis of all synthesized designs so far

  • 3
  • Article
  • Updated 9 years ago
I'm excited about the RNAFold thing. It seems to me like a reasonable thing to do would be to let RNAFold do some analysis of all the designs we have synthesized so far, then compare RNAFold's predictions with the "synthesis score" or maybe even the SHAPE data that we got from actually synthesizing the things.

However, it's a bit too much of a project for one person to undertake. I propose that a few of us get together (in this thread) to decide what data exactly we should get from RNAFold, then split up the actual work and post it together when it's done. If we find 5 people to do 1/5th of the work, it shouldn't be much of a task. So:

1) Want to help?
2) What exactly should we get? Obviously the numbers like the percentage and the ensemble diversity and such are easy, but is there any good way to get a spreadsheet output of what RNAFold thinks are the biggest problem areas of the RNA (maybe the percentages for each bond, or the entropy at each base)? That way we could compare it to the SHAPE data. People who are good at RNAFold should propose a process to follow and show us how to do it, then we can split up the work among people who answer question #1.

:)
Photo of Chris Cunningham [ccccc]

Chris Cunningham [ccccc]

  • 97 Posts
  • 13 Reply Likes
  • hopeful

Posted 9 years ago

  • 3
Photo of Ding

Ding

  • 94 Posts
  • 20 Reply Likes
I'd definitely be up for going through a bunch of old entries to contribute to a spreadsheet of all synthesized designs' stats.

I really like the idea of trying to get something that could be compared to the SHAPE data base-by-base but I think that I would need some lessons in using the standalone RNAfold rather than the server.
Photo of Berex NZ

Berex NZ

  • 116 Posts
  • 20 Reply Likes
This might help you, ccccc.
Picture is a summary, if you want the full data sheet, please follow the link underneath it. Should give you guys a headstart :)



http://www.editgrid.com/user/berex_nz...

If you are using my spreadsheet, conditional formatting on the shape data above 5 = yellow, if under 5 = blue and you should end up to something very similar to the yellow and blues in the lab.

Haven't been able to get entropy by base yet though. Good luck! :)
Photo of Chris Cunningham [ccccc]

Chris Cunningham [ccccc]

  • 97 Posts
  • 13 Reply Likes
Yes !!
Photo of Chris Cunningham [ccccc]

Chris Cunningham [ccccc]

  • 97 Posts
  • 13 Reply Likes
I ran the obvious correlation coefficients and scatter plots.... depending on how you look at it this is either really good or really bad news, because none of these are a quick-and-dirty way to predict Synthesis Score.



Edit: ummm, some of my correlation coefficients look backwards, specifically the ones relating to free energy. Maybe the signs are wrong somewhere. Regardless, that should only cause sign errors... hm.