PSearch - 3D ligand-based pharmacophore modeling
PSearch is a Python application to automatically generate 3D pharmacophore models based on a supplied data set of compounds with measured activity values.
git clone https://github.com/meddwl/psearch git submodule init git submodule update
rdkit >= 2017.09
networkx >= 1.11
It is recommended to create an empty dir which would be your
$PROJECT_DIR and copy an input file to that location.
There are two steps of pharmacophore model generation.
- Data set preparation. It takes as input a comma-separated SMILES file containing
activity value. It splits the input on active and inactive subsets, generates stereoisomers and conformers, creates databases of active and inactive compounds with labeled pharmacophore features.
python3 prepare_datatset.py -i $PROJECT_DIR/input.smi -l 6 -u 8 -c 4
-i - path to the input file;
-u - treshold to define active compounds (compounds with
activity value >= threshold are considered active);
-l - treshold to define inactive compounds (compounds with
activity value <= threshold are considered inactive);
-c - number of CPUs to use.
There are other arguments available to tweak data set preparation. To get the full list of agruments run
python3 prepare_datatset.py -h
- Model building.
python3 psearch.py -p $PROJECT_DIR -t 0.4 -c 4
-p - path to the project dir;
-t - threshold for compound clustering to create training sets;
-c- number of CPUs to use
Alina Kutlushina, Pavel Polishchuk