mehc-curation 1.0.4

A comprehensive toolkit for molecular data curation, validation, cleaning, and normalization

MEHC Curation

A comprehensive Python toolkit for molecular data curation, including validation, cleaning, normalization, and refinement pipelines.

Features

• Validation: Validate SMILES strings and remove unwanted molecular types (mixtures, inorganics, organometallics)
• Cleaning: Remove salts and neutralize charged molecules
• Normalization: Normalize tautomers and stereoisomers
• Refinement: Complete pipeline orchestrating all stages
• Parallel Processing: Efficient parallel processing using all available CPUs by default
• Comprehensive Reporting: Generate detailed reports for each processing stage

Installation

Prerequisites

Before installing mehc-curation, you need to install RDKit, which is best installed via conda:

conda install -c conda-forge rdkit

Install from PyPI

pip install mehc-curation

Install from source

git clone https://github.com/biochem-data-sci/mehc-curation.git
cd mehc-curation
pip install -e .

Quick Start

Python API

import pandas as pd
from mehc_curation.validation import ValidationStage
from mehc_curation.cleaning import CleaningStage
from mehc_curation.normalization import NormalizationStage
from mehc_curation.refinement import RefinementStage

# Load your SMILES data
df = pd.read_csv("your_data.csv")

# Validation
validator = ValidationStage(df)
validated_df = validator.complete_validation()

# Cleaning
cleaner = CleaningStage(validated_df)
cleaned_df = cleaner.complete_cleaning()

# Normalization
normalizer = NormalizationStage(cleaned_df)
normalized_df = normalizer.complete_normalization()

# Complete refinement pipeline
refiner = RefinementStage(df)
refined_df = refiner.complete_refinement(
    output_dir="./output",
    get_report=True
)

Command Line Interface

# Validation
python -m mehc_curation.validation -i input.csv -o output/ -c 5

# Cleaning
python -m mehc_curation.cleaning -i input.csv -o output/ -c 3

# Normalization
python -m mehc_curation.normalization -i input.csv -o output/ -c 3

# Complete refinement
python -m mehc_curation.refinement -i input.csv -o output/ --get_report

Modules

Validation Module

Validates SMILES strings and removes unwanted molecular types:

• validate_smi(): Validate SMILES strings
• rm_mixture(): Remove mixture compounds
• rm_inorganic(): Remove inorganic compounds
• rm_organometallic(): Remove organometallic compounds
• complete_validation(): Run all validation steps

Cleaning Module

Cleans SMILES strings:

• cl_salt(): Remove salts from SMILES
• neutralize(): Neutralize charged molecules
• complete_cleaning(): Run all cleaning steps

Normalization Module

Normalizes SMILES strings:

• detautomerize(): Normalize tautomers
• destereoisomerize(): Remove stereoisomers
• complete_normalization(): Run all normalization steps

Refinement Module

Complete refinement pipeline:

• complete_refinement(): Orchestrates validation, cleaning, and normalization stages

Configuration

CPU Usage

By default, the library uses all available CPUs (n_cpu=-1). You can specify the number of CPUs:

# Use all CPUs (default)
refiner.complete_refinement(n_cpu=-1)

# Use specific number of CPUs
refiner.complete_refinement(n_cpu=4)

# Use single CPU
refiner.complete_refinement(n_cpu=1)

Output Directories

• output_dir is optional for every stage. If you omit it, data stays in memory and any generated reports are written to the current working directory.
• When you do provide an output_dir, the folder will be created automatically if it does not exist, and both CSV outputs and reports are saved beneath it.

Duplicate Handling

• param_deduplicate now defaults to True for all validation, cleaning, and normalization entry points so that duplicate rows are removed automatically unless you opt out.

Requirements

• Python >= 3.7
• pandas >= 1.3.0
• parallel-pandas >= 0.2.8
• RDKit (install via conda: conda install -c conda-forge rdkit)

License

MIT License - see LICENSE file for details

Citation

If you use this library in your research, please cite:

@software{mehc_curation,
  title={MEHC-curation: An Automated Python Framework for High-Quality Molecular Dataset Preparation},
  author={Chinh Pham and Nhat-Anh Nguyen-Dang and Thanh-Hoang Nguyen-Vo and Binh P. Nguyen},
  month={dec},
  year={2025},
  version={1.0.4},
  url={https://github.com/biochem-data-sci/mehc-curation},
  license={MIT},
  doi={10.5281/zenodo.17567530}, 
  publisher={Zenodo}
}

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Support

For issues and questions, please open an issue on GitHub.