What is it?

RNAmutants is a software designed to explore the consequences of mutation(s) on the secondary structures of RNA sequences. Instead of running separately a classical secondary structure prediction software (mfold, RNAfold or RNAstructure), on all possible RNA sequences with k mutations, RNAmutants computes all these predictions in once.

Why would I need it?

Usage of RNAmutants includes (but is not restricted to):

RNAmutants allows us to study the resilience of an RNA molecule to pointwise mutations and to predict mutations that will tend to stabilize or modify the secondary structure (a.k.a. deleterious mutations). Analysis of the results can leads us to detect regions presumably under evolutionary pressure (see References).
Using sequence or folding constraints, RNAmutants can also be used as an efficient and realistic RNA design program (see Tutorial).

What is new?

RNAmutants use efficient dynamic programming algorithms allowing to compute in polynomial time and space the minimum free energy structures and the Boltzmann partition function for each k-mutants (sequence with k mutations).
Using classical structure prediction softwares such as mfold, RNAfold or RNAstructure on each sequences would require an exponential time and then cannot be applied on real size sequence with a large number number of mutations. For instance, we provide in (Waldispühl et al., 2008) an illustration of the full mutation landscape of sequences with 37 nucleotides (all sequences of length 37). Without RNAmutants such results would have been impossible to compute.

How to run it?

You can download a binary distribution of RNAmutants and run it on your own computer. Alternatively, a webserver is running in P. Clote's laboratory and is available at bioinformatics.bc.edu/clotelab/RNAmutants
A tutorial showing how to install and run RNAmutants is available on this website. We illustrate the versatility of the program by presenting some potential uses of the program.

How does it work?

RNAmutants computes the partition function of the grand canonical ensemble of all secondary structures that can be built over all mutants of the given sequence. Then, it rigorously samples from this ensemble, mutant sequences together with a secondary structures on this sequence.

The input RNA sequence is represented at the center while the k-neighbourhoods (Here k = 1, 2) are represented by concentric rings. Each individual RNA sequence is associated with a set of secondary structures that can be mapped onto it (the boxed structures). These comprise the set of structure that have to be enumerated to compute the Boltzmann partition function).
From this illustration, it is easy to see how RNAmutants generalize the previous structure prediction algorithms.