RNA sequence design for a given target secondary structure. Sequences are designed by optimising a design score in a continuous sequence space, allowing a mixture of bases at each position during optimisation.

Marco C. Matthies, Stefan Bienert, Andrew E. Torda. Dynamics in Sequence Space for RNA Secondary Structure Design. J. Chem. Theory Comput., 2012, 8, 10, 3663--3670. https://doi.org/10.1021/ct300267j

The dynamics in sequence space optimisation is performed with the program opt-md. The opt-sd and opt-sd-gsl programs perform sequence design by gradient descent alone and are not described in the paper.

# show command-line arguments
./opt-md -h

# design a sequence for the target structure ((((...))))
./opt-md '((((...))))'

# you can often improve the quality of designed sequences by playing with
# the --kneg and --khet parameters, which change the weights of the negative
# design and sequence heterogeneity terms, as well as other parameters

The code can be compiled by running make in this directory. You need the GSL (GNU Scientific Library) header files and library installed (libgsl-dev package in Ubuntu).

Additionally, there is a little utility program called rna-ensemble-distance that uses the ViennaRNA library to compute the target probability and ensemble defect of a designed sequence with respect to a given target structure. You can find it in the rna-ensemble-distance-with-ViennaRNA/ directory. The README file in the rna-ensemble-distance-with-ViennaRNA/ directory contains instructions on how to build this utility.

The beginnings of a python interface can be found in the directory python/. Install it with

pip install python/

# alternatively, if you want to run the python unit tests, do
pip install python/[test]
# run the python unit tests
pytest

Example usage:

import dssopt
dssopt.opt_md('(((...)))')

The parameters for the RNA nearest neighbour energy model were converted from ViennaRNA, many thanks to them. The parameters were converted from the ViennaRNA-1.8.5 param/vienna17.par parameter file to C code with scripts/vienna_parse_param.py. The generated C code containing the parameters can be found in nn_param.c.

All code was written by Marco Matthies, Centre for Bioinformatics, Hamburg and is licensed under the GNU General Public License, version 3 or later.