Python library and utilities for constructing 3D models of RNA molecules, developed as part of my master's thesis with Professor Jérôme Waldispühl.

Provides the rna_builder, rna_insert_loops and bp2_bridge command line utilities, and the rna_bits Python library for programmatic access.

Note that BayesPairing2 is a separate utility developed by Roman Sarrazin-Gendron for predicting loop motif insertion sites.

Link to the original rna_bits repo on Jérôme's gitlab: https://jwgitlab.cs.mcgill.ca/pheneg1/rna_bits

Python >= 3.8

Requires numpy, biopython, mmtf-python, which will be installed automatically.

Running certain data generation pipelines requires MC-Annotate, which needs to be downloaded separately. For convenience, I have included the output of MC-Annotate for those pipelines. (TODO: MC-Annotate installation pointer.)

As of now, I have only tested on Linux.

Make sure you're inside the directory and run:

pip3 install -e .

It will automatically install the python requirements, and make the python library and command line utilities available.

Assembles a 3D RNA structure from a ".rass" file and .pdb files of fragments.

cd examples/rna_builder
rna_builder 1MMS_native_motifs.rass

Given an RNA secondary structure and sequence, will annotate loops, that can then be assembled using rna_builder. Outputs a .rass file.

Before using this, the loop database needs to be generated by running generate_loops.py (Or by running the numbered scripts in src/rna_bits/data/loops/) (Note: the rna_bits library must first be installed). This might take multiple hours.

python3 generate_loops.py

Then, we can insert loops and assemble structures:

cd examples/rna_builder
rna_insert_loops 1MMS_no_motifs.rass
rna_builder il_out/*.rass

Turns an output from BayesPairing2 (https://jwgitlab.cs.mcgill.ca/sarrazin/rnabayespairing2) into an input that can be passed to rna_builder.

First, install BayesPairing2 (follow instructions in above link.)

Next, run BayesPairing2 using the BayesPairing command (if you installed BayesPairing2 in a new conda environment, make sure that conda environment is active). Example command:

mkdir bp2b_example  # create an empty directory
cd bp2b_example
BayesPairing -seq "UUUUUUAAGGAAGAUCUGGCCUUCCCACAAGGGAAGGCCAAAGAAUUUCCUU" -samplesize 1000 -d RELIABLE

Next, use bp2_bridge to generate the rass files.

bp2_bridge has 2 required arguments: the output.json from BayesPairing2, and the models json file that BayesPairing2 used. (The --chefs_choice flag is optional, to restrict to using only the "chef's choice" motifs as output by BayesPairing2.)

bp2_bridge --bp2_result output.json --database ~/where_you_installed_bp2/rnabayespairing2/bayespairing/models/RELIABLE.json --chefs_choice

This might take some time because it will download PDB files to generate the 3d motifs.

Finally, assemble the 3d structures:

rna_builder bp2b_out/*.rass