Biodeg

Predict if a molecule is generally degraded when in the wild.

Install

From source

conda create -n biodeg python=3.11 && \
    conda activate biodeg && \
    pip install git+https://github.com/research11111/biodeg.git#egg=biodeg

From github

Download a release from releases, then

conda create -n biodeg python=3.11 && \
    conda activate biodeg && \
    pip install biodeg-*.whl

Run

Make some real time prediction with the default installed prediction model

biodeg predict -s 'O=C1CCCCCO1' -o '/dev/stdout'

SMILES,BioDegradability
O=C1CCCCCO1,1

Train a new prediction model with 10 epoch on a small dataset

biodeg train -e 10 -i data/data/All-Public_dataset_Mordred_tail_10.csv -o /tmp/e.pt

Test accuracy of this new model

biodeg test -m /tmp/e.pt -i data/data/All-Public_dataset_Mordred_tail_10.csv

Developp

Install

conda create -n biodeg python=3.11 && \
    conda activate biodeg && \
    pip install poetry && \
    poetry install --no-root && \
    pip install -e .

Build

poetry build

Test

poetry run pytest

Credits

All the initial work credits go to Myeonghun Lee and Kyoungmin Min (https://pubs.acs.org/doi/10.1021/acsomega.1c06274).
biodeg is just a refactoring of this work with some small improvments