schrodinger.application.matsci.msml.mlutils module¶

Utilities for managing ML prediction models (model data definitions and paths)

class schrodinger.application.matsci.msml.mlutils.ModelData(group, name, flag, prop, directory, skip_standardization, input_type, additional_model, model_file)¶

Bases: tuple

additional_model¶: Alias for field number 7

directory¶: Alias for field number 4

flag¶: Alias for field number 2

group¶: Alias for field number 0

input_type¶: Alias for field number 6

model_file¶: Alias for field number 8

name¶: Alias for field number 1

prop¶: Alias for field number 3

skip_standardization¶: Alias for field number 5

class schrodinger.application.matsci.msml.mlutils.SolventComposition(solvent_1, solvent_2, solvent_1_fraction)¶

Bases: tuple

solvent_1¶: Alias for field number 0

solvent_1_fraction¶: Alias for field number 2

solvent_2¶: Alias for field number 1

schrodinger.application.matsci.msml.mlutils.get_model_dir(subdir_name, info_dir=False)¶

Get the directory path for a model

Parameters:

subdir_name (str) – The subdir name for the model
info_dir (bool) – Whether the directory is for information about model

Return type:

str

Returns:

The full path to the model directory, default or custom

schrodinger.application.matsci.msml.mlutils.get_oled_models()¶

Return list of OLED ModelData objects

Return type:: list
Returns:: List of OLED model data

schrodinger.application.matsci.msml.mlutils.get_models_by_group(group)¶

Return list of ModelData objects for a given group

Parameters:: group (str) – Model group (SINGLE_MOL, POLYMER, OLED, etc.)
Return type:: list
Returns:: List of ModelData objects matching the group

schrodinger.application.matsci.msml.mlutils.get_model_by_flag(flag)¶

Return ModelData for a given flag

Parameters:: flag (str) – Model flag (e.g., ‘-density_predict’)
Return type:: ModelData or None
Returns:: ModelData object if found, None otherwise

schrodinger.application.matsci.msml.mlutils.read_model_info(model_data, model_dir=None)¶

Read the info.json metadata for one ML model.

Parameters:

model_data (ModelData) – ML model data definition
model_dir (str) – Optional explicit model/info directory containing info.json

Return type:

dict

Returns:

Parsed info.json contents

schrodinger.application.matsci.msml.mlutils.get_model_chemical_space(model_data, model_dir=None)¶

Return the supported element list for one ML model.

Parameters:

model_data (ModelData) – ML model data definition
model_dir (str) – Optional explicit model/info directory containing info.json

Return type:

list

Returns:

Ordered list of supported element symbols

schrodinger.application.matsci.msml.mlutils.fetch_model_from_URL(model_dir, url)¶

Fetch the model from the given URL

Parameters:

model_dir (str) – The directory to save the model to
url (str) – The URL to download the model from

Return type:

(bool, str)

Returns:

True, empty string if the model was downloaded successfully, False and error message if not

schrodinger.application.matsci.msml.mlutils.download_model(model_data)¶

Download the model if it is not already present

Parameters:: model_data (ModelData) – ML model data to download
Return type:: bool, str
Returns:: True, model directory if the model is present or downloaded successfully, False and error message if there was an error downloading the model

schrodinger.application.matsci.msml.mlutils.read_solvent_file(solvent_file='opto_training_solvents.csv')¶

Read in the solvent file and store it as dictionary

Parameters:: solvent_file (str) – The name of the solvent file to read
Return type:: dict, dict
Returns:: Dictionary of solvent names to SMILES strings, and SMILES string to names

schrodinger.application.matsci.msml.mlutils.get_smiles(struct, mark_head_tail=True, logger=None)¶

Get SMILES pattern of the structures

Parameters:

struct (Structure) – The structure to get SMILES for
mark_head_tail (bool) – Whether to replace polymer head and tail atoms if found

Return type:

str or None

Returns:

SMILES pattern, or None if there was an issue

schrodinger.application.matsci.msml.mlutils.get_smasher_data(flag='these')¶

Load the in slow-to-import ligand_ml information if ligand_ml is available

Parameters:: flag (str) – The model flag for the warning message

schrodinger.application.matsci.msml.mlutils.get_numpy_dataset_class(flag='these')¶

Load the Ligand ML NumpyDataset class required for crystal prediction.

Parameters:: flag (str) – Model flag to include in the missing-product error.
Return type:: type(NumpyDataset)
Returns:: NumpyDataset class

schrodinger.application.matsci.msml.mlutils.struct_to_pmg_crystal(struct)¶

Convert a periodic Schrodinger structure to a ligand_ml PMGStructure.

Parameters:

struct (Structure) – Periodic structure to convert

Raises:

TypeError – If the conversion fails
ValueError – If the structure is invalid

Return type:

ligand_ml.data.datasets.PMGStructure

Returns:

Converted crystal structure

schrodinger.application.matsci.msml.mlutils.get_molecular_weight(struct, is_polymer, range_max)¶

Get the molecular weight of a structure, removing dummy atoms if necessary

Parameters:

struct (Structure) – The structure to get molecular weight for
is_polymer (bool) – Whether the structure is a polymer
range_max (float) – The maximum molecular weight allowed

Return type:

float

Returns:

The molecular weight of the structure

exception schrodinger.application.matsci.msml.mlutils.MLPredictionError¶: Bases: RuntimeError

schrodinger.application.matsci.msml.mlutils.validate_structure_input_type(structs, input_type)¶

Validate that structures are compatible with a model input type.

Parameters:

structs (list(Structure)) – Structures to validate
input_type (str) – One of MODEL_INPUT_TYPES

Raises:

ValueError – If input_type is unknown
MLPredictionError – If any structure is incompatible

class schrodinger.application.matsci.msml.mlutils.MLPropPredictor(options, structs=None, input_file=None, force_cpu=False, warn_func=None, logger=None)¶

Bases: object

Class to handle ML property prediction

SMILES_JOINER = '.[1S].'¶

SMILES_HEADER = 'SMILES'¶

FEATURE_0_HEADER = 'additional_feature_0'¶

MODEL_FILE = 'pkgd_model_qzip.qzip'¶

DATA_INFO_FILE = 'data_info.json'¶

PANEL_INFO_FILE = 'info.json'¶

LOGGER = None¶

FLAG_TO_DATA = {'-absorption_fwhm_predict': ModelData(group='optoelectronics', name='Absorption bandwidth', flag='-absorption_fwhm_predict', prop='r_matsci_Absorption_Bandwidth_(cm-1)', directory='opto_absorption_FWHM', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-absorption_lmax_predict': ModelData(group='optoelectronics', name='Absorption peak position', flag='-absorption_lmax_predict', prop='r_matsci_Absorption_Lmax_(nm)', directory='opto_absorption_lambda_max', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-aqua_solubility_predict': ModelData(group='single_molecule', name='Aqueous solubility', flag='-aqua_solubility_predict', prop='r_matsci_Aqueous_Solubility_(log_mol/L)', directory='aqueous_solubility', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-band_gap_predict': ModelData(group='crystal', name='Band gap', flag='-band_gap_predict', prop='r_matsci_Band_Gap_(eV)', directory='band_gap', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-bulk_modulus_predict': ModelData(group='crystal', name='Bulk modulus', flag='-bulk_modulus_predict', prop='r_matsci_Bulk_Modulus_(GPa)', directory='bulk_modulus', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-de_s1_t1_gap_predict': ModelData(group='oled', name='Singlet-triplet energy gap', flag='-de_s1_t1_gap_predict', prop='r_matsci_Singlet_Triplet_Energy_Gap_(eV)', directory='oled_singlet_triplet_gap', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-density_predict': ModelData(group='single_molecule', name='Density', flag='-density_predict', prop='r_matsci_Density_(g/cm3)_at_%.2f_K', directory='organic_density', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-df_predict': ModelData(group='polymer', name='Dielectric loss', flag='-df_predict', prop='r_matsci_Polymer_Df_at_%.2f_Hz', directory='polymer_df', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-dk_predict': ModelData(group='polymer', name='Dielectric constant', flag='-dk_predict', prop='r_matsci_Polymer_Dk_at_%.2f_Hz', directory='polymer_dk', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-electron_reorg_predict': ModelData(group='oled', name='Electron reorganization energy', flag='-electron_reorg_predict', prop='r_matsci_Electron_Reorganization_(eV)', directory='oled_electron_reorg', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-emission_fwhm_predict': ModelData(group='optoelectronics', name='Emission bandwidth', flag='-emission_fwhm_predict', prop='r_matsci_Emission_Bandwidth_(cm-1)', directory='opto_emission_FWHM', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-emission_lifetime_predict': ModelData(group='optoelectronics', name='Emission lifetime', flag='-emission_lifetime_predict', prop='r_matsci_Emission_Lifetime_(log_ns)', directory='opto_log_lifetime', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-emission_lmax_predict': ModelData(group='optoelectronics', name='Emission peak position', flag='-emission_lmax_predict', prop='r_matsci_Emission_Emax_(nm)', directory='opto_emission_lambda_max', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-energy_above_hull_predict': ModelData(group='crystal', name='Energy above hull', flag='-energy_above_hull_predict', prop='r_matsci_Energy_Above_Hull_(eV/atom)', directory='energy_above_hull', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-extinction_predict': ModelData(group='optoelectronics', name='Extinction coefficient', flag='-extinction_predict', prop='r_matsci_Extinction_Coefficient_(log_mol-1_dm3_cm-1)', directory='opto_log_e', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-formation_energy_predict': ModelData(group='crystal', name='Formation energy', flag='-formation_energy_predict', prop='r_matsci_Formation_Energy_(eV/atom)', directory='formation_energy', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-hole_reorg_predict': ModelData(group='oled', name='Hole reorganization energy', flag='-hole_reorg_predict', prop='r_matsci_Hole_Reorganization_(eV)', directory='oled_hole_reorg', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-melting_point_predict': ModelData(group='single_molecule', name='Melting point', flag='-melting_point_predict', prop='r_matsci_Melting_Point_(K)', directory='melting_temp', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-non_aqua_solubility_predict': ModelData(group='formulation', name='Non-aqueous solubility', flag='-non_aqua_solubility_predict', prop='r_matsci_NonAq_Solubility_(log_mol_frac)_at_%s_K', directory='nonaqueous_solubility', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-oxidation_potential_predict': ModelData(group='single_molecule', name='Oxidation potential', flag='-oxidation_potential_predict', prop='r_matsci_Oxidation_Potential_(V)', directory='oxidation_potential', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-plqy_predict': ModelData(group='optoelectronics', name='Photoluminescence quantum yield', flag='-plqy_predict', prop='r_matsci_Photoluminescence_Quantum_Yield', directory='opto_quantum_yield', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-reduction_potential_predict': ModelData(group='single_molecule', name='Reduction potential', flag='-reduction_potential_predict', prop='r_matsci_Reduction_Potential_(V)', directory='reduction_potential', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-scaled_homo_predict': ModelData(group='oled', name='Scaled HOMO', flag='-scaled_homo_predict', prop='r_matsci_scaled_HOMO_Energy_(eV)', directory='oled_scaled_homo', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-scaled_lumo_predict': ModelData(group='oled', name='Scaled LUMO', flag='-scaled_lumo_predict', prop='r_matsci_scaled_LUMO_Energy_(eV)', directory='oled_scaled_lumo', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-semi_organic_bp_predict': ModelData(group='single_molecule', name='Boiling point', flag='-semi_organic_bp_predict', prop='r_matsci_Evaporation_temperature_(K)_at_%.2f_Torr', directory='semi_organic_bp', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-semi_organic_vp_predict': ModelData(group='single_molecule', name='Vapor pressure', flag='-semi_organic_vp_predict', prop='r_matsci_Vapor_pressure_(Torr)_at_%s_K', directory='semi_organic_vp', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-semi_organometal_bp_predict': ModelData(group='single_molecule', name='Boiling point', flag='-semi_organometal_bp_predict', prop='r_matsci_Evaporation_temperature_(K)_at_%.2f_Torr', directory='semi_organometallic_bp', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-shear_modulus_predict': ModelData(group='crystal', name='Shear modulus', flag='-shear_modulus_predict', prop='r_matsci_Shear_Modulus_(GPa)', directory='shear_modulus', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-tg_predict': ModelData(group='polymer', name='Polymer glass transition temperature', flag='-tg_predict', prop='r_matsci_Polymer_Tg_(K)', directory='polymer_tg', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-triplet_energy_predict': ModelData(group='oled', name='Triplet Energy (E(S0T1))', flag='-triplet_energy_predict', prop='r_matsci_Triplet_Energy_(eV)', directory='oled_triplet_energy', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-triplet_reorg_predict': ModelData(group='oled', name='Triplet reorganization energy', flag='-triplet_reorg_predict', prop='r_matsci_Triplet_Reorganization_Energy_(eV)', directory='oled_triplet_reorg', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-viscosity_predict': ModelData(group='single_molecule', name='Viscosity', flag='-viscosity_predict', prop='r_matsci_Viscosity_(cP)_at_%.2f_K', directory='organic_viscosity', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip')}¶

__init__(options, structs=None, input_file=None, force_cpu=False, warn_func=None, logger=None)¶

Create a ML prediction class instance

type options: argparse.Namespace :param options: The command line options

Parameters:

structs (list of schrodinger.structure.Structure) – list of structures for prediction
input_file (str) – name of the input file
force_cpu (bool) – If true, force the model to run on CPU else LigandML will figure out the best device to run the model on. The default is False.
warn_func (callable) – The function to call to log/show warnings

Raises:

MLPredictionError – if the input structure is coarse-grained.

log(msg, **kwargs)¶

Add a message to the log file

Parameters:: msg (str) – The message to log

Additional keyword arguments are passed to the textlogger.log_msg function

run()¶

Run the workflow

Return type:: pandas.DataFrame
Returns:: Dataframe containing structures and predicted properties

getComputedValues()¶

Get computed values from data frame.

Return type:: list[float or None]
Returns:: Computed values for all the structures. None if a value wasn’t computed

getSolventComposition()¶

Normalize the solvent option to one solvent composition.

The legacy scalar solvent option is accepted as a pure-solvent composition. A missing option represents the pure thin-film condition.

Return type:: SolventComposition
Returns:: Solvent composition for the prediction
Raises:: MLPredictionError – If the solvent composition has an invalid type

classmethod getProperty(options)¶

Get the name of the computed property for the model.

Return type:: str
Returns:: The name of the computed property
Raises:: MLPredictionError – If no property is requested

validateAtomPresent()¶

Validate the input structure to check if structure is differing from trained chemical space. If not, remove the structure from prediction.

Raises:: MLPredictionError – If the atom in the input structure is not part of trained dataset.

featurizeStruct()¶: Prepare structures for the representation required by the model input type.

featurizeMolStructs()¶

Convert non-periodic structures into RDKit molecules and SMILES strings.

Structures that cannot be converted are skipped, and the successfully converted structures are used to initialize the prediction output.

Raises:: MLPredictionError – If no structures convert successfully

featurizeCrystalStructs()¶

Convert periodic structures into a Ligand ML crystal dataset.

Raises:: MLPredictionError – If no structures convert successfully

convertToMol(struct)¶

Convert the structure to a rdkit mol that includes the solvent and necessary properties

Parameters:: struct (Structure) – The structure to convert
Return type:: dict
Returns:: Row data for the structure, including the RDKit molecule under the RDKIT_MOL key, or an empty dictionary if conversion failed.

getFormulationMol(smiles, struct, solvent_composition)¶

Create a formulation molecule for a formulation-type prediction.

Parameters:

smiles (str) – smiles of the input structure
struct (Structure or None) – Structure corresponding to smiles. Its title is updated for non-solubility formulations.
solvent_composition (SolventComposition) – Solvent composition for this molecule

Return type:

Chem.rdchem.Mol or None

Returns:

Formulation molecule, or None if the SMILES are invalid.

getFormulationComponents(smiles, struct, solvent_composition)¶

Return formulation components, composition, and solid indicator.

Parameters:

smiles (str) – SMILES of the input structure.
struct (Structure or None) – Input structure whose title is updated for the formulation.
solvent_composition (SolventComposition) – Solvent composition for the formulation.

Return type:

tuple(list(str), list(float), str)

Returns:

Component SMILES, component fractions, and solid indicator.

Raises:

MLPredictionError – If two solvents are supplied for a model that does not support binary solvents.

getRDKitMol(smiles)¶

Get RDKIT mol object from SMILES string

Parameters:: smiles (str) – The SMILES string
Return type:: Chem.rdchem.Mol or None
Returns:: The RDKIT mol object or None if smiles to mol object conversion failed

getAdditionFeaturesVectors(feature_flag)¶

Get the additional features for the model

Parameters:: feature_flag (str) – The flag for the model
Return type:: float
Returns:: The additional features for the model

getModel()¶

Get the ligand_ml model object

Raises:: MLPredictionError – If the user did not select a model to run

getModelFromData(model_data, Smasher)¶

Load a model package from its metadata.

Parameters:

model_data (ModelData) – Model metadata
Smasher (type) – Ligand ML Smasher class.

Return type:

Smasher

Returns:

Loaded model with metadata attributes

getFullModelPath(dirname, model_filename=None)¶

Get the full path to the model file, using the custom version if found

Parameters:

dirname (str) – The path to built-in model
model_filename (str) – Model package filename. If omitted, use the standard model package filename.

Return type:

tuple(str, str)

Returns:

The actual path to the model directory and model package.

Raises:

MLPredictionError – If the model file is not found

getInfoFromModelDir(model_directory)¶

Get molecular weight range and chemical space from a model directory.

Parameters:: model_directory (str) – The path to the model directory
Return type:: tuple(list(float), list(str))
Returns:: Molecular-weight range and chemical space

getAdditionalFeaturesFromModelDir(model_directory)¶

Get the additional features from a model directory.

Parameters:: model_directory (str) – The path to the model directory
Return type:: dict or None
Returns:: Additional model features, if defined

predictMols(mols, model)¶

Run a batched prediction on RDKit molecules.

Parameters:

mols (list(Chem.rdchem.Mol)) – Molecules for prediction.
model (Smasher) – Loaded model to evaluate.

Return type:

list(tuple(float, float))

Returns:

Prediction and uncertainty pairs.

Raises:

MLPredictionError – If model prediction fails or returns an unexpected number of results.

predictCrystal(model)¶

Run a batched prediction on the crystal dataset.

Parameters:: model (Smasher) – Loaded model to evaluate.
Return type:: list(tuple(float, float))
Returns:: Prediction and uncertainty pairs.
Raises:: MLPredictionError – If model prediction fails or returns an unexpected number of results.

postProcessResults(results, model)¶

Apply model-specific post-processing to prediction results.

Parameters:

results (list(tuple(float, float))) – Prediction and uncertainty pairs.
model (Smasher) – Model used for the predictions.

Return type:

list(tuple(float, float))

Returns:

Post-processed prediction and uncertainty pairs.

predict()¶: Run the predictions

cleanDataFrame()¶

Remove internal prediction data and rows without numerical results.

Raises:: MLPredictionError – if the dataframe is empty after removal

cleanModelFiles()¶: Clean temporary files created by the loaded models.

checkMolecularWeights()¶: Check the molecular weight of the structures and add a warning property if the molecular weight is outside the range

classmethod getDefaultOptions(model_data, option_overrides=None)¶

Build default options dict for MLPropPredictor.

Parameters:

model_data (ModelData) – ML model data
option_overrides – Option values that override the model defaults

Return type:

dict

Returns:

Options dictionary for MLPropPredictor to run prediction

class schrodinger.application.matsci.msml.mlutils.SolubilityMLPropPredictor(options, structs=None, input_file=None, force_cpu=False, warn_func=None, logger=None)¶

Bases: MLPropPredictor

ML property predictor with solubility-specific processing.

FLAG_TO_DATA = {'-absorption_fwhm_predict': ModelData(group='optoelectronics', name='Absorption bandwidth', flag='-absorption_fwhm_predict', prop='r_matsci_Absorption_Bandwidth_(cm-1)', directory='opto_absorption_FWHM', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-absorption_lmax_predict': ModelData(group='optoelectronics', name='Absorption peak position', flag='-absorption_lmax_predict', prop='r_matsci_Absorption_Lmax_(nm)', directory='opto_absorption_lambda_max', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-aqua_solubility_predict': ModelData(group='single_molecule', name='Aqueous solubility', flag='-aqua_solubility_predict', prop='r_matsci_Aqueous_Solubility_(log_mol/L)', directory='aqueous_solubility', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-band_gap_predict': ModelData(group='crystal', name='Band gap', flag='-band_gap_predict', prop='r_matsci_Band_Gap_(eV)', directory='band_gap', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-bulk_modulus_predict': ModelData(group='crystal', name='Bulk modulus', flag='-bulk_modulus_predict', prop='r_matsci_Bulk_Modulus_(GPa)', directory='bulk_modulus', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-de_s1_t1_gap_predict': ModelData(group='oled', name='Singlet-triplet energy gap', flag='-de_s1_t1_gap_predict', prop='r_matsci_Singlet_Triplet_Energy_Gap_(eV)', directory='oled_singlet_triplet_gap', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-density_predict': ModelData(group='single_molecule', name='Density', flag='-density_predict', prop='r_matsci_Density_(g/cm3)_at_%.2f_K', directory='organic_density', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-df_predict': ModelData(group='polymer', name='Dielectric loss', flag='-df_predict', prop='r_matsci_Polymer_Df_at_%.2f_Hz', directory='polymer_df', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-dk_predict': ModelData(group='polymer', name='Dielectric constant', flag='-dk_predict', prop='r_matsci_Polymer_Dk_at_%.2f_Hz', directory='polymer_dk', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-electron_reorg_predict': ModelData(group='oled', name='Electron reorganization energy', flag='-electron_reorg_predict', prop='r_matsci_Electron_Reorganization_(eV)', directory='oled_electron_reorg', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-emission_fwhm_predict': ModelData(group='optoelectronics', name='Emission bandwidth', flag='-emission_fwhm_predict', prop='r_matsci_Emission_Bandwidth_(cm-1)', directory='opto_emission_FWHM', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-emission_lifetime_predict': ModelData(group='optoelectronics', name='Emission lifetime', flag='-emission_lifetime_predict', prop='r_matsci_Emission_Lifetime_(log_ns)', directory='opto_log_lifetime', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-emission_lmax_predict': ModelData(group='optoelectronics', name='Emission peak position', flag='-emission_lmax_predict', prop='r_matsci_Emission_Emax_(nm)', directory='opto_emission_lambda_max', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-energy_above_hull_predict': ModelData(group='crystal', name='Energy above hull', flag='-energy_above_hull_predict', prop='r_matsci_Energy_Above_Hull_(eV/atom)', directory='energy_above_hull', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-extinction_predict': ModelData(group='optoelectronics', name='Extinction coefficient', flag='-extinction_predict', prop='r_matsci_Extinction_Coefficient_(log_mol-1_dm3_cm-1)', directory='opto_log_e', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-formation_energy_predict': ModelData(group='crystal', name='Formation energy', flag='-formation_energy_predict', prop='r_matsci_Formation_Energy_(eV/atom)', directory='formation_energy', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-hole_reorg_predict': ModelData(group='oled', name='Hole reorganization energy', flag='-hole_reorg_predict', prop='r_matsci_Hole_Reorganization_(eV)', directory='oled_hole_reorg', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-melting_point_predict': ModelData(group='single_molecule', name='Melting point', flag='-melting_point_predict', prop='r_matsci_Melting_Point_(K)', directory='melting_temp', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-non_aqua_solubility_predict': ModelData(group='formulation', name='Non-aqueous solubility', flag='-non_aqua_solubility_predict', prop='r_matsci_NonAq_Solubility_(log_mol_frac)_at_%s_K', directory='nonaqueous_solubility', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-oxidation_potential_predict': ModelData(group='single_molecule', name='Oxidation potential', flag='-oxidation_potential_predict', prop='r_matsci_Oxidation_Potential_(V)', directory='oxidation_potential', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-plqy_predict': ModelData(group='optoelectronics', name='Photoluminescence quantum yield', flag='-plqy_predict', prop='r_matsci_Photoluminescence_Quantum_Yield', directory='opto_quantum_yield', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-reduction_potential_predict': ModelData(group='single_molecule', name='Reduction potential', flag='-reduction_potential_predict', prop='r_matsci_Reduction_Potential_(V)', directory='reduction_potential', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-scaled_homo_predict': ModelData(group='oled', name='Scaled HOMO', flag='-scaled_homo_predict', prop='r_matsci_scaled_HOMO_Energy_(eV)', directory='oled_scaled_homo', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-scaled_lumo_predict': ModelData(group='oled', name='Scaled LUMO', flag='-scaled_lumo_predict', prop='r_matsci_scaled_LUMO_Energy_(eV)', directory='oled_scaled_lumo', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-semi_organic_bp_predict': ModelData(group='single_molecule', name='Boiling point', flag='-semi_organic_bp_predict', prop='r_matsci_Evaporation_temperature_(K)_at_%.2f_Torr', directory='semi_organic_bp', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-semi_organic_vp_predict': ModelData(group='single_molecule', name='Vapor pressure', flag='-semi_organic_vp_predict', prop='r_matsci_Vapor_pressure_(Torr)_at_%s_K', directory='semi_organic_vp', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-semi_organometal_bp_predict': ModelData(group='single_molecule', name='Boiling point', flag='-semi_organometal_bp_predict', prop='r_matsci_Evaporation_temperature_(K)_at_%.2f_Torr', directory='semi_organometallic_bp', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-shear_modulus_predict': ModelData(group='crystal', name='Shear modulus', flag='-shear_modulus_predict', prop='r_matsci_Shear_Modulus_(GPa)', directory='shear_modulus', skip_standardization=False, input_type='crystal', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-solubility_predict': ModelData(group='formulation', name='Solubility prediction', flag='-solubility_predict', prop='r_matsci_Solubility_(log_g/L)_at_%s_K', directory='solubility', skip_standardization=False, input_type='molecule', additional_model=[ModelData(group='single_molecule', name='Solubility density', flag='-solubility_density_predict', prop='r_matsci_Solubility_Density_(g/cm3)_at_%.2f_K', directory='solubility', skip_standardization=False, input_type='molecule', additional_model=None, model_file='density_prod_package.tar.gz')], model_file='pkgd_model_qzip.qzip'), '-tg_predict': ModelData(group='polymer', name='Polymer glass transition temperature', flag='-tg_predict', prop='r_matsci_Polymer_Tg_(K)', directory='polymer_tg', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-triplet_energy_predict': ModelData(group='oled', name='Triplet Energy (E(S0T1))', flag='-triplet_energy_predict', prop='r_matsci_Triplet_Energy_(eV)', directory='oled_triplet_energy', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-triplet_reorg_predict': ModelData(group='oled', name='Triplet reorganization energy', flag='-triplet_reorg_predict', prop='r_matsci_Triplet_Reorganization_Energy_(eV)', directory='oled_triplet_reorg', skip_standardization=True, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip'), '-viscosity_predict': ModelData(group='single_molecule', name='Viscosity', flag='-viscosity_predict', prop='r_matsci_Viscosity_(cP)_at_%.2f_K', directory='organic_viscosity', skip_standardization=False, input_type='molecule', additional_model=None, model_file='pkgd_model_qzip.qzip')}¶

predictFromJson(prediction_data)¶

Predict solubility from JSON-compatible data without Structures.

Each value contains the solute and solvent SMILES, the complete semicolon-separated composition in additional_feature_0, and the temperature in Kelvin. The caller is responsible for validating the data.

Parameters:: prediction_data (dict(str, dict)) – Maps prediction identifiers to SMILES, composition, and temperature values.
Return type:: list(tuple(float, float))
Returns:: Prediction and uncertainty pairs.
Raises:: MLPredictionError – If model loading or prediction fails.

getFormulationComponents(smiles, struct, solvent_composition)¶

Return solubility components and their composition.

Parameters:

smiles (str) – SMILES of the solute.
struct (Structure or None) – Input structure whose title is updated for the formulation.
solvent_composition (SolventComposition) – Solvent composition for the formulation.

Return type:

tuple(list(str), list(float), None)

Returns:

Component SMILES, component fractions, and solid indicator.

postProcessResults(results, model)¶

Apply the density correction to solubility results.

Parameters:

results (list(tuple(float, float))) – Prediction and uncertainty pairs.
model (Smasher) – Model used for the predictions.

Return type:

list(tuple(float, float))

Returns:

Predictions converted to log10(g/L) for the solubility model, or unchanged predictions for other models.

alignWithDensity(results)¶

Refine solubility predictions using solvent-density predictions.

Parameters:: results (list(tuple(float, float))) – Solubility predictions.
Return type:: list(tuple(float, float))
Returns:: Solubility predictions converted to log10(g/L).

predictSolventDensity(density_model)¶

Predict the pure-solvent densities in g/cm3.

Parameters:: density_model (Smasher) – Loaded solvent-density model.
Return type:: dict(tuple, tuple(float, float))
Returns:: Solvent and feature keys mapped to finite density prediction and uncertainty pairs.
Raises:: MLPredictionError – If the model returns an unexpected number of results.

getUniqueSolventMols(density_model)¶

Return density-model inputs keyed by solvent and feature values.

Parameters:: density_model (Smasher) – Loaded solvent-density model.
Return type:: dict(tuple, Chem.rdchem.Mol)
Returns:: Unique solvent and feature keys mapped to RDKit molecules.

static getDensityKey(density_model, solvent, row)¶

Return the density-result key for one solvent dataframe row.

Parameters:

density_model (Smasher) – Loaded solvent-density model.
solvent (str) – Solvent SMILES.
row (pandas.Series) – Formulation data containing density-model feature values.

Return type:

tuple(str, tuple)

Returns:

Solvent SMILES and ordered feature-name/value pairs.

postProcessWithDensity(results, density_model, density_results)¶

Convert mole-fraction solubility predictions to log10(g/L).

Parameters:

results (list(tuple(float, float))) – Solubility predictions.
density_model (Smasher) – Loaded density model.
density_results (dict) – Maps solvent and feature values to their pure-solvent density predictions.

Return type:

list(tuple(float, float))

Returns:

Converted solubility predictions.