Melimda: Machine Learning Improved Docking (scoring) Algorithm

Melimda is a simple tool to re-score molecular docking results generated using Autodock Vina. Melimda is trained using experimentally validated data of molecular binding derived from the PDBBind database, and employs Morgan Fingerprints and Binding Site descriptors, along with the Vina raw score, to estimate a more accurate energy of interaction. It requires the .pdbqt files of the Vina output (--ligand) and receptor (--receptor).

Requirements

• python (3.8)
• pdbfixer
• Autodock Vina
• OpenBabel

Setup

$ pip install https+git@github.com:omixlab/melimda-cli.git

Running

$ melimda-predict \
    --ligand result.pdbqt  \
    --receptor receptor.pdbqt \
    --model model.pkl \
    --output result.txt > /dev/null

Cite Us

Goulart, L (2025). Melimda: Machine Learning Improved Docking (scoring) Algorithm. Available at: https://github.com/omixlab/melimda-cli.