RNAsketch Library for designing RNA molecules

Installation

Download the lastes RNAsketch package from https://github.com/ViennaRNA/RNAsketch/releases/latest, exctract and simply use setuptools to install the module and the few sample scripts:

python setup.py install

I highly recommend to install the module locally into a directory specified in your PYTHONPATH variable. This can be achieved either by using the --prefix=$HOME/local_path argument, or in case PYTHONUSERBASE is set, just use the --user flag.

Build the documentation

Documentation available online: http://viennarna.github.io/RNAsketch

Call these commands to build the html documentation:

cd doc
make html

Cite

Stefan Hammer, Birgit Tschiatschek, Christoph Flamm, Ivo L. Hofacker, and Sven Findeiß. “RNAblueprint: Flexible Multiple Target Nucleic Acid Sequence Design.” Bioinformatics, 2017. https://doi.org/10.1093/bioinformatics/btx263.

If you used the multi-dimensional boltzmann sampling scripts design-redprint-multistate.py or design-energyshift.py please cite:

Stefan Hammer, Yann Ponty, Wei Wang, and Sebastian Will. “Fixed-Parameter Tractable Sampling for RNA Design with Multiple Target Structures.” In RECOMB 2018. Paris, France, 2018. https://arxiv.org/abs/1804.00841.

Provided Example Scripts

Design a Multistate Design

This simple script generates a multistate design, which is a RNA molecule that is able to fold into all the given structural conformation. In case of a bistable molecule just call:

echo -e '(((((....)))))....\n....(((((....)))))' | design-multistate.py -i -m random -s 500

The program barriers can the be used to visualize the energy landscape to confirm the design goals:

  GUGACCGCGGUCACGUGG
1 (((((....)))))....  -7.00    0  10.00
2 ....(((((....)))))  -7.00    1   9.50
3 ..................   0.00    2   1.60
4 ....(((......)).).   0.80    1   2.00
5 ((....))..........   1.10    1   1.50
6 .......((......)).   1.50    1   1.10
7 ...((.........))..   2.40    1   0.40
8 .((....)).........   2.60    1   0.20

Barriers Tree showing the two desired states as deep minima (1, 2) and the open chain (3) as neighbouring minimum.

Design a sRNA mediated translational regulation

echo '...(((((((((((((((((&)))))))))))))))))((((....))));...(((((((((((((((((&)))))))))))))))))............
....................&.................((((....))));...xxxxxxxxxxxxxxxxx&xxxxxxxxxxxxxxxxx............
NNNNAAGGAGNNNNNNNAUG&NNNNNNNNNNNNNNNNNNNNNNNNNNNNN' | design-cofold.py -n 1 -s 1000

This small example will design a simple device consisting of a 5'UTR region which can be translationally controlled by a sRNA molecule. In this case the sRNA will shut down translation by directly binding the RBS (Ribosome Binding Site) and the AUG start codon.

RNAcofold -a -p -d2 calculates three dot-plots showing the base pair probabilities in the ensemble of states which confirms the design objective:

RNAcofold Dot-Plots, ViennaRNA v2.2.9, AAAUAAGGAGUAAAUGAAUG&CAUUCAUUUACUCCUUACCGCACUCGCGG Plots were assembled in a single picture for better comparison. Only base pair probabilities are shown in the plots.

Score: 0.89; complex concentration: 1.00; P(5UTR unpaired): 0.97; P(sRNA unpaired): 0.99; P(mRNA context): 0.18

Design a multistate Thermoswitch

echo -e "(((((((((((((....))))))))))))) 5.0\n(((((.....)))))(((((.....))))) 10.0\n(((((.....)))))............... 37.0" | design-thermoswitch.py -m random -s 1000

This results e.g in a sequence like GAUCUGUGUGGGGUCGAUUUUGUGUGGGUU which has the given MFE structures at the specified temperatures (lower plot). Folding it at all Temeratures from 10 to 100 degree Celsius shows, that the first structural change happens at ~7.0 degree Celsius and the second one at ~26 degrees. After _72 degrees, the sequence occurs only in the open chain conformation.

RNAheat further confirms that the designed sequence is indeed a three-stable thermoswitch:

RNAheat Plot, ViennaRNA v2.2.9, GAUCUGUGUGGGGUCGAUUUUGUGUGGGUU

Design a ligand triggered switch

echo -e "(((((...((((((((.....)))))...)))...)))))........................\n.........................(((((((((((......)))))))))))...........\nAAGUGAUACCAGCAUCGUCUUGAUGCCCUUGGCAGCACUUCANNNNNNNNNNNNNNNNNNNNNN" | design-ligandswitch.py -r 70:30 --ligand "GAUACCAG&CCCUUGGCAGC;(...((((&)...)))...);-9.22"

This designs a simple theophylline triggered switch. It adapts to a certain ratio of the aptamer structure (ligand competent state) and an alternative state as specified with the --ratio option. To model ligand binding we use the soft-constraints framework of the ViennaRNA package, similar to the --motif option of RNAfold. The specified objective function calculates the probabilities of the structural features with and without ligand. Thus, we optimize towards the given ratio without the ligand and maximize the ligand binding competent state in the presence of the ligand.

Design a multistate riboswitch with equal target energies

export REDPRINT=</path/to/redprint/>
echo -e "(((((((((((((....)))))))))))))\n(((((.....)))))(((((.....)))))\n(((((.....)))))..............." | design-redprint-multistate.py -i -n 10

Using this script you can design a multistate riboswitch similar to the design-multistate.py script with fundamental differences in the sampling technology. Here, we use RNARedPrint to gain Boltzmann-weighted sampling of sequences given the structural inputs. This script then implements a strategy to achieve multi-dimensional boltzmann sampling with the target energy being the mean achieveable energy of the target structures.

The resulting sequences exhibit equal energies for the given target structurs. This does not mean that the target structures are highly populated in the ensemble. To additionally achieve this, specify the -s 200 option to subsequently do a optimization towards the multi-target objective function.

Either put the complete RNARedPrint project folder next to the scripts, or set a REDPRINT environmental variable before executing this script!

Design a multistate riboswitch with specific target energies

export REDPRINT=</path/to/redprint/>
echo -e ".((((((......)))))).((((...((((((...((((...(((.......)))..........(((....))).))))..))))))...))))....\n.((((((......)))))).((((...((((((...((((((.(((.......((........))..)))....)).))))..))))))...))))....\n......((.((((.(((((((.((.((...((((..((.....(((.......))).....))..)))).)).)))))))))..)).)).))........" | design-energyshift.py -i -e 40,40,20

Design a multistate riboswitch with specific target energies and GC-content. This script uses RNARedPrint for Boltzmann-weighted sampling and implements a strategy to achieve multi-dimensional boltzmann sampling.

Either put the complete RNARedPrint project folder next to the scripts, or set a REDPRINT environmental variable before executing this script!

The resulting sequences are nicely distributed around the specified target energies:

Display the Dependency Graph

We generated two example scripts which can dump the Dependency Graph in the common GraphML format and, by using the igraph python library, render these files as images. Following example input is possible:

echo -e '(((((....)))))....\n....(((((....)))))' | design-generategraphml.py -i > dependency-graph.gml
design-printgraphml.py -g dependency-graph.gml -o dependency-graph.png

Or use the second script directly:

echo -e '(((((....)))))....\n....(((((....)))))\n(((((((....)))))))' | design-printgraphml.py -i

This results in a nice representation of the dependency graph:

Very simple dependency graph visualized using igraph.