https://github.com/rdkit/automatedseriesclassification

https://github.com/rdkit/automatedseriesclassification

Last synced: about 2 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/rdkit/automatedseriesclassification
• Owner: rdkit
• License: mit
• Created: 2020-09-30T11:03:24.000Z (about 4 years ago)
• Default Branch: master
• Last Pushed: 2021-06-23T12:33:52.000Z (over 3 years ago)
• Last Synced: 2024-09-26T02:01:39.969Z (3 months ago)
• Language: Jupyter Notebook
• Size: 729 KB
• Stars: 22
• Watchers: 3
• Forks: 7
• Open Issues: 2
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# AutomatedSeriesClassification

This is code for automated chemical series classification

## Original article

Automated Identification of Chemical Series: Classifying like a Medicinal Chemist

https://pubs.acs.org/doi/abs/10.1021/acs.jcim.0c00204

## Installation

First, you should have RDKit installed. Then, the code can be downloaded and installed with:

```bash

git clone https://github.com/rdkit/AutomatedSeriesClassification

cd AutomatedSeriesClassification

pip install -e .

```

The ``-e`` flag means it gets installed in editable mode.

## Example usage

### Data Preparation

1. The following script will download chembl27.sdf.gz and make substructurefingerprint library.

   If you want to use an alternate version of chembl, specify the `--chebml-version` flag. You

   can run `python -m automated_series_classification.dataprep --help` in your shell to see all options.

```

$ python -m automated_series_classification.dataprep  # it'll take ~30 or more minutes on my PC

```

2. Then launch jupyter notebook, the notebook use same dataset as original articles. But you'll get different results compared to the article. This is because I used more newer version of ChEMBL for this code. If you would like to use same dataset to original article it is easy, just changing download link of chembl

## Acknoledgements

- Greg Landrum

## etc

Any comments, requests and suggestions will be greatly appreciated.

## License

[MIT](https://choosealicense.com/licenses/mit/)