mordred Documentation

The mordred package provides the primary interfaces for molecular descriptor calculation:

• mordred.Calculator: The main engine used to compute descriptors for molecules.
• mordred.Descriptor: Represents an individual molecular descriptor.
• mordred.Result: Represents the outcome of a descriptor calculation for a specific molecule.

The Calculator class is the primary interface for computing molecular descriptors.

• Initialization: You pass a collection of descriptors (usually the descriptors object) to the constructor.
• 3D Descriptors: By default, Calculator may include 3D descriptors. To restrict to 2D descriptors, use ignore_3D=True. To explicitly use 3D descriptors, ensure your input molecules have 3D coordinates (from SDF or MOL files) and use the -3 flag in CLI or appropriate settings in the library.
• Output Formats: When used as a library, the calculator can return results as a list/sequence or directly as a pandas.DataFrame using the .pandas(mols) method.

To install via pip, you must first install the rdkit python package. You can then install mordred either with extra requirements (recommended for full functionality) or as a standard package.

# Install with extra requirements
$ pip install 'mordred[full]'

# Or standard install
$ pip install mordred

The recommended way to install mordred is using conda. Ensure you have miniconda or anaconda installed first, then run the following command to install mordred from the mordred-descriptor channel.

$ conda install -c rdkit -c mordred-descriptor mordred

You can test if the installation was successful by running the built-in tests using the following command.

$ python -m mordred.tests

To use mordred in your code, import Calculator and descriptors. You typically initialize a Calculator with a set of descriptors and then pass RDKit molecule objects to it.

from rdkit import Chem
from mordred import Calculator, descriptors

# Create a calculator (ignore_3D=True for 2D descriptors)
calc = Calculator(descriptors, ignore_3D=True)

# Calculate descriptors for a single molecule
mol = Chem.MolFromSmiles('c1ccccc1')
results = calc(mol)
print(results[:3])  # Access first 3 descriptors

# Calculate descriptors for multiple molecules
mols = [Chem.MolFromSmiles(smi) for smi in ['c1ccccc1Cl', 'c1ccccc1O', 'c1ccccc1N']]
results_list = calc.pandas(mols)
print(results_list['SLogP'])

Use the following functions to discover and collect descriptors available within specific Python modules:

• get_descriptors_in_module(module): Retrieves descriptors contained within a given module.
• get_descriptors_from_module(module): Retrieves descriptors from a specified module.

The following flags are available when running python -m mordred:

You can use mordred directly from the command line to calculate descriptors for molecules provided in files (SMILES, SDF, or MOL).

# Calculate all descriptors for a SMILES file
$ python -m mordred example.smi

# Save results to a CSV file
$ python -m mordred example.smi -o example.csv

# Stream read (low memory usage, no molecule count info)
$ python -m mordred example.smi -s -o example.csv

# Calculate specific descriptors (e.g., ABCIndex)
$ python -m mordred example.smi -d ABCIndex

# Calculate multiple descriptors (e.g., ABCIndex and AcidBase)
$ python -m mordred example.smi -d ABCIndex -d AcidBase

# Process multiple input files
$ python -m mordred example.smi example2.smi -d ABCIndex