UNAFold and mfold

3 replies [Last post]
tj2015
Offline
Joined: 09/24/2012

Dear Developer,

I recently tried to use UNAFold to predict the secondary structure for several RNA sequences, but I found it gave different result for the same RNA sequence from mfold (online version). Could you tell me how to setup the parameter for UNAFold to get the similar result as mfold? I used default parameters both for UNAFold and the online mfold.

I also downloaded mfold version 3.5, but I always met this problem when I run it on linux (with default parameters):
mfold version 3.5
REUSE= NO
temp2.pnt created.
Sequence length is 401
Folding at 37 degrees using version 3.0 dat files.
Save file created using nafold.
Save file is empty. No foldings.
Job Aborted

Is there a method to produce a picture showing the bases for the secondary structure by UNAFold?

Thank you for your help.

zukerm
Offline
Joined: 11/12/2010
UNAFold versus mfold

UNAFold and mfold differ slightly because the free energy parameters had to be modified for UNAFold to avoid overcounting when computing partition functions. You should not expect identical results. The energy dot plots should be almost identical. You can perform a visual inspection. If you are only looking at a single folding such as the minimum free energy structure, then you are not using the software correctly. You need to look at multiple foldings as well as the energy dot plot to get an idea of which parts of the predictions are more reliable than others.

To address your concerns, I am making available a modified set of version 3.0 RNA free energy parameters. They are named "unafold_ala_mfold", which means that they have been designed so that UNAFold will predict the same structures as mfold as much as possible. There are still some small algorithm differences. You may download these parameters at the following URL:
unafold_ala_mfold.tar.gz
The command: tar -zxvf unafold_ala_mfold.tar.gz will create a directory containing the modified free energy files. Be careful not to mix them with the standard free energy files with the same names (suffix is .DAT). If you want to use them, set the environment variable UNAFOLDDAT to point to the directory where they are stored. For example: export UNAFOLDDAT=/usr/local/share/unafold_ala_mfold will cause the UNAFold software to look in the named directory first for energy files. If you use these files, please report back on this forum about your experience.

The error you report when running mfold suggests that the software has not been properly installed. Did you first use "make" and then "make install"? What are the contents of the log file? Did you finally get mfold running locally or are you relying solely on computations from the web application?

UNAFold has no display software to draw secondary structures. You should download the mfold_util package from our web site (version 4.6). The program you want is sir_graph. If you have mfold installed, then you already have the non-interactive version of sir_graph (possibly named sir_graph_ng) installed. You may also upload a ".ct" or ".b" file to the Structure Display and Free Energy Determination application. This will only draw one structure at a time. You almost certainly want sir_graph running locally.

tj2015
Offline
Joined: 09/24/2012
More information. Actually

More information.
Actually only one RNA sequence I tested by UNAfold and mfold show some difference at a perticular place that I am interested in. For all other sequences I tested, they produce almost the same result.

zukerm
Offline
Joined: 11/12/2010
That is good news. Please

That is good news. Please read the reply to your previous post