Reinvent Scoring with Enhanced Shape Similarity Components

This package is a fork of the original reinvent-scoring package with additional shape similarity components including RDKit-based and enhanced Roshambo implementations.

Added Features

RDKit Shape Components

• rdkit_shape_similarity: An open-source alternative to ROCS similarity using RDKit
• parallel_rdkit_shape_similarity: A parallel version for faster processing

Enhanced Roshambo Shape Components

• roshambo_shape_similarity: GPU-accelerated shape similarity using Roshambo with RDKit conformer generation
• parallel_roshambo_shape_similarity: Multi-GPU parallel version for high-throughput processing
• RoshamboConformerGenerator: Dedicated RDKit-based conformer generator for Roshambo
• Environment path support for conda environments
• Enhanced parameter management with dedicated enums
• Multiple reference file support for PROTAC design

Why This Fork?

The original package relies on OpenEye's ROCS for 3D shape similarity, which requires a commercial license. This fork adds both RDKit-based alternatives and enhanced Roshambo implementations that provide similar or superior functionality:

• RDKit components: Completely open-source alternative to ROCS
• Enhanced Roshambo: GPU-accelerated with RDKit conformer generation for optimal performance
• Environment flexibility: Support for conda environments and different installation setups
• PROTAC support: Multiple reference files for complex molecular design scenarios

Installation

pip install reinvent-scoring-rdkit

Usage

RDKit Shape Similarity

{
    "component_type": "rdkit_shape_similarity",
    "name": "RDKit Shape Similarity",
    "weight": 1.0,
    "specific_parameters": {
        "reference_file": "path/to/reference.sdf",
        "method": "usrcat",
        "shape_weight": 0.5,
        "color_weight": 0.5,
        "max_confs": 50
    }
}

Enhanced Roshambo Shape Similarity

{
    "component_type": "roshambo_shape_similarity",
    "name": "Roshambo Shape Similarity",
    "weight": 1.0,
    "specific_parameters": {
        "reference_file": "path/to/reference.sdf",
        "shape_weight": 0.6,
        "color_weight": 0.4,
        "n_confs": 20,
        "environment_path": "conda activate roshambo_env &&",
        "gpu_id": 0,
        "save_overlays": true,
        "overlays_dir": "shape_overlays"
    }
}

Parallel Roshambo for Multi-GPU

{
    "component_type": "parallel_roshambo_shape_similarity",
    "name": "Parallel Roshambo",
    "weight": 1.0,
    "specific_parameters": {
        "reference_file": "path/to/reference.sdf",
        "shape_weight": 0.6,
        "color_weight": 0.4,
        "n_confs": 20,
        "max_gpus": 2,
        "gpu_ids": [0, 1],
        "debug": true
    }
}

PROTAC Design with Multiple References

{
    "component_type": "roshambo_shape_similarity",
    "name": "PROTAC Shape Similarity",
    "weight": 1.0,
    "specific_parameters": {
        "reference_file": "protac_reference.sdf",
        "warhead1_reference": "warhead1.sdf",
        "warhead2_reference": "warhead2.sdf",
        "shape_weight": 0.6,
        "color_weight": 0.4,
        "n_confs": 30,
        "save_overlays": true
    }
}

Original Package

This package is based on the original reinvent-scoring package. All credit for the original code goes to the original authors.

Developing

Setup environment

You can use Conda to create an environment with all the necessary packages installed.

$ conda env create -f reinvent_scoring
[...]
$ conda activate reinvent_scoring

Run tests

The tests use the unittest package testing framework. Before you can run the tests make sure that you have created a config.jsonfile in the reinvent_scoring/configs directory. There is an example config in the same directory, which you can base your own config off of. Make sure that you set MAIN_TEST_PATH to a non-existent directory; it is where temporary files will be written during the tests; if it is set to an existing directory, that directory will be removed once the tests have finished.

Some tests require a proprietary OpenEye license; you have to set up a few things to make the tests read your license. The simple way is to just set the OE_LICENSE environment variable to the path of the file containing the license. If you just want to set the license in the reinvent_scoring Conda environment, it is a bit more complicated, but you only have to do it once.

(reinvent-scoring) $ cd $CONDA_PREFIX
$ mkdir -p etc/conda/activate.d
$ mkdir -p etc/conda/deactivate.d

Put the following in etc/conda/activate.d/env_vars.sh.

#!/bin/sh
export OE_LICENSE='</path/to/your/oe_license/file>'

And put the following in etc/conda/deactivate.d/env_vars.sh.

#!/bin/sh
unset OE_LICENSE

Once you have created the files, deactivate and re-activate the environment, and echo $OE_LICENSE should output the path to the license file.

Once you have created and configured your environment, you can run unittests by running

python main_test.py --unittests

If you have a valid Open eye license and other dependencie configured, like Icolos and AZDOCK - you can also run integration tests, by running command (remember to submit this configuration, since the default one is test):

python main_test.py --integration --base_config <path to your configuration>

Building

• Building: python setup.py sdist bdist_wheel
• Upload build to test: python -m twine upload --repository testpypi dist/*
• Upload build: python -m twine upload dist/*