[DIYbio] Re: Any use for RNA secondary structure prediction? (slash help getting started?)

There are actually very good reasons to use the partition function over the minimum free energy, at least for strands of length N > 100. If you look at the Boltzmann distribution P(s) = e^{-E(s)}/Z, the probability of being in any given state goes like 1.1*10^{-N/59} (see attached pMFE2.pdf). For many of these strands, there exist multiple stable macrostates, the most stable of which may not contain the MFE state because they get their entropic weight from multiplicity of states near them, rather than having the lowest energies (see attached ensemble-centroid.jpg). Predicting structure by sampling from the boltzmann distribution and clustering structures has been shown to be more accurate than just the MFE calculation. Check out Ding and Lawrence (2005) http://www.ncbi.nlm.nih.gov/pubmed/16043502. There are also many other reasons not to trust MFE computations, mostly because the energy model of RNA has tons of error and uncertainty which the MFE is *really* sensitive to, while partition function probabilities are a little more robust. I wrote about this extensively in my thesis. You are correct though, it is a difficult computation because there are many states, specifically O(1.8^n), but there exist dynamic programming algorithms that compute it in O(N^3). 

I was thinking more of basic DIY projects that I could do and maybe extend with RNA structure prediction, or other molecular dynamics tools. I recently read about BioBricks, thought I might be able to help design some standardized parts, but I would have a hard time extending something I haven't touched yet. I want to get my hands dirty, and through that figure out something I could use my computational knowledge for. I am just completely in the dark with how to get started experimentally. 

Does this book look good: http://www.amazon.com/Synthetic-Biology-Manual-Josefine-Liljeruhm/dp/9814579548/ref=sr_1_4?ie=UTF8&qid=1439325077&sr=8-4&keywords=biobuilder  ?

Mike


On Tuesday, August 11, 2015 at 12:20:49 PM UTC-4, Josiah Zayner wrote:
Interesting thesis topic. I still can't wrap my head around the use of a partition function vs just "pure" free energy calculations. Especially because of non-local base pairing leading to an ridonculously incalculable number of possible states but I guess the goal is to try and predict more canonical secondary structures? And probably I just haven't spent enough time reading on the topic. Previous to this though I never knew that mfold(unafold) used a partition function calculation. 

The question you ask is difficult to answer, why is RNA secondary structure prediction important? RNA secondary structure prediction is important because it provides a tool for Scientists to use, to better understand how biology works. Science can't have exact answers in some(most?) cases. Think about using SHAPE for RNA secondary structure determination it only gives someone info on what bases are "flexible" or "unpaired"(the reason these are in quotes is because those definitions can vary). The only ways to really determine a "macrostate" conformation in RNA secondary structure is through things like NMR or X-ray crystallography and even those are just a local energy minimum based on context. So programs like unafold are used widely by RNA biochemists and biologist to understand how and why RNA works the way it does. The more we understand how things work the better we can predict why things are doing what they are doing and the better we can do cool things with Science. Helping with unafold doesn't necessarily contribute anything directly to our knowledge of the universe but it helps others contribute.

As for what you should do, the field of Molecular Dynamics simulations lends itself to someone with knowledge of physics and computer science. Molecular Dynamics around protein function and dynamics is a field that is being extensively and successfully explored. If you run a windows box I would suggest looking up tutorials on NAMD and proteins, if you use linux I would suggest looking up tutorials on GROMACS and proteins. Simulations can also be run with the 3D structures of RNA but the energy functions aren't quite optimized and so I would generally stick to proteins.

If you have any questions feel free to send them my way.


Josiah Zayner




On Monday, August 10, 2015 at 2:37:16 PM UTC-7, Michael Flynn wrote:
Hi all,

I recently wrote my undergraduate thesis on RNA secondary structure prediction algorithms. In the process I've become intimately familiar with the algorithms and code bases behind Unafold and RNAstructure, which I have been told comprise 90% of the market share of RNA structure prediction software. My thesis is located here, if you are interested.  

However, I was a physics and computer science student, and I have minimal (no) training in biology. This meant that most of the time while I was writing my thesis I was in great want of background knowledge and motivation. I was often asked why my research was important, but I could never give a strong answer. When I asked my advisor for motivations behind our work, he did not answer to my satisfaction. So I thought, what better way to learn than DIYbio?

So are there ways I could help this community? What would be the best path for me to get started, in a way that my background would be the most relevant? 

Mike

--
-- You received this message because you are subscribed to the Google Groups DIYbio group. To post to this group, send email to diybio@googlegroups.com. To unsubscribe from this group, send email to diybio+unsubscribe@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/diybio?hl=en
Learn more at www.diybio.org
---
You received this message because you are subscribed to the Google Groups "DIYbio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to diybio+unsubscribe@googlegroups.com.
To post to this group, send email to diybio@googlegroups.com.
Visit this group at http://groups.google.com/group/diybio.
To view this discussion on the web visit https://groups.google.com/d/msgid/diybio/9a43c14d-0600-4d0f-a80a-4754ac2f5463%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

  • Digg
  • Del.icio.us
  • StumbleUpon
  • Reddit
  • RSS

0 comments:

Post a Comment