Unofficial CPython binding to LinearPartition

Use pip to install the module.

pip install linearpartition-unofficial

You may build from the source code for unsupported Python versions or platforms.

git clone --recursive https://github.com/ChangLabSNU/python-linearpartition
cd python-linearpartition
pip install .

The module currently only has one function called partition(seq). The seq parameter should be an RNA sequence in uppercase letters, and any T should be converted to U before passing it to the function.

>>> import linearpartition as lp
>>> seq = 'UGUCGGGGUUGGCUGUCUGACA'
>>> pred = lp.partition(seq)
>>> pred['free_energy']
-7.216465644007023
>>> pred['structure']
'(((((((........)))))))'
>>> import pandas as pd
>>> pd.DataFrame(pred['bpp']).sort_values('prob', ascending=False).head()
    i   j      prob
19  3  18  0.999201
18  2  19  0.998801
17  1  20  0.997717
21  5  16  0.996692
22  4  17  0.996508

The linearpartition.partition function is a Python C extension function that calls LinearPartition to perform a linear partitioning operation and get the base pairing probability matrix.

linearpartition.partition(seq, mode='eterna', beamsize=100, dangles=2)

• seq (required): A string containing the RNA sequence to be analyzed. The sequence must be in uppercase and only contain A, C, G, and U. This parameter is required.
• mode (optional): The name of free energy parameters to use. Use 'vienna' for Vienna RNA parameters, or 'eterna' for EternaFold parameters.
• beamsize (optional): An integer representing the beam size for the operation. Larger value requires more computational time and memory. The default value is 100.
• dangles (optional): An integer representing the number of dangles for the partitioning operation. The default value is 2.

This function returns a dictionary containing the MEA structure, base-pairing probability matrix and free energy of the ensemble structure in kcal/mol from the result of the partitioning operation.

Hyeshik Chang <[email protected]>

This Python binding is licensed under the MIT-style license. However, the compiled binary includes code from the LinearPartition package, which is licensed for non-commercial use.