This project contains the script to run the experiments and reproduce the figures in the manuscript Old dog, new tricks: Exact seeding strategy improves RNA design performances

Dependencies

The project is heavily based on ViennaRNA (RNAlib) and LinearBPDesign. The latter one is installed as a git submodule. Run git submodule update --init --recursive to initiate.

Other required dependencies for figures are numpy, pandas, seaborn, matplotlib, varnaapi, logomaker ...

Script

The folder script contains different scripts to run experiments described in the manuscript

• rundesign.py: design given target structure using RNAinverse with different seeds. For example, the command below returns 10 solutions of structure ..(((((..((.((((.......)))).))..)))))... with Biseparable seed with modulo up to 3

   python script/rundesign.py -n 10 --seed linearbp -m 3 "..(((((..((.((((.......)))).))..)))))..."

• immediate_solutions.py: sample seeds and check whether they are T-design (w/o RNAinverse). The command below sample each 20 Boltzmann sampled seeds for 1st to 10th structures in xxx.txt (one line for each structure in dot-bracket notation)

   python script/immediate_solutions.py xxx.txt -n 20 -s 1 -e 10 --seed bpenergy

• gen_ss.py: generate uniform or MFE structures. The command below generates uniformly 1,000 structures of size 100 nts with helix length of 3+

   python script/gen_ss.py 100 -n 1000 --helix_length 3

• parseDesignResult.py: evaluate (ensemble defect, diversity ...) resulting design produced by rundesign.py and stored as pandas DataFrame in pickle. The script assumes the result of each target indiced i is stored in puzzle_i.csv. The command below parses all puzzle_*.csv results under xxx and stores the DataFrame in yyy.pkl.

   python script/parseDesignResult.py xxx yyy.pkl

Notebooks

The folder notebooks contains jupyter notebooks as indicated individually by the file name to reproduce figures in the manuscript