Suboptimal folding

The text output shows an energy sorted list (option -s) of all secondary structures within 1 kcal/mol of the MFE structure. Our sequence actually has a ground state structure (-5.40) and three structures within 1 kcal/mol range.

MFE folding alone gives no indication that there are actually a number of plausible structures. Remember that RNAsubopt cannot automatically plot structures, therefore you can use the tool RNAplot.

Note that the number of suboptimal structures grows exponentially with sequence length and therefore this approach is only tractable for sequences with less than 100 nt. To keep the number of suboptimal structures manageable the option -noLP can be used, forcing RNAsubopt to produce only structures without isolated base pairs. While RNAsubopt produces all structures within an energy range, mfold produces only a few, hopefully representative, structures. Try folding the sequence on the mfold server at
http://mfold.rna.albany.edu/?q=mfold.

Sometimes you want to get information about unusual properties of the Boltzmann ensemble (the sum of all RNA structures possible) for which no specialized program exists. For example you want to know all fractions of a bacterial mRNA in the Boltzmann ensemble where the Shine-Dalgarno (SD) sequence is unpaired. If the SD sequence is concealed by secondary structure the translation efficiency is reduced.

In such cases you can resort to drawing a representative sample of structures from the Boltzmann ensemble by using the option -p. Now you can simply count how many structures in the sample possess the feature you are looking for. This number divided by the size of your sample gives you the desired fraction.

The following example calculates the fraction of structures in the ensemble that have bases 6 to 8 unpaired.

Sven Findeiss 2013-11-22