skchem.data.converters package¶

Submodules¶

skchem.data.converters.base module¶

# skchem.data.converters.base

Defines the base converter class.

class skchem.data.converters.base.Converter(directory, output_directory, output_filename='default.h5')[source]¶

Bases: object

Create a fuel dataset from molecules and targets.

classmethod convert(**kwargs)[source]¶

create_file(path)[source]¶

classmethod fill_subparser(subparser)[source]¶

run(ms, y, output_path, splits=None, features=None, pytables_kws={'complib': 'bzip2', 'complevel': 9})[source]¶

Args:

ms (pd.Series):

The molecules of the dataset.

ys (pd.Series or pd.DataFrame):

The target labels of the dataset.

output_path (str):

The path to which the dataset should be saved.

features (list[Feature]):

The features to calculate. Defaults are used if None.

splits (iterable<(name, split)>):

An iterable of name, split tuples. Splits are provided as boolean arrays of the whole data.

save_features(ms)[source]¶

Save all features for the dataset.

save_frame(data, name, prefix='targets')[source]¶

Save the a frame to the data file.

save_molecules(mols)[source]¶

Save the molecules to the data file.

save_splits()[source]¶

Save the splits to the data file.

save_targets(y)[source]¶

source_names¶

split_names¶

class skchem.data.converters.base.Feature(fper, key, axis_names)¶

Bases: tuple

axis_names¶

Alias for field number 2

fper¶

Alias for field number 0

key¶

Alias for field number 1

class skchem.data.converters.base.Split(mask, name, converter)[source]¶

Bases: object

contiguous¶

indices¶

ref¶

save()[source]¶

to_dict()[source]¶

skchem.data.converters.base.contiguous_order(to_order, splits)[source]¶

Determine a contiguous order from non-overlapping splits, and put data in that order.

Parameters:	to_order (iterable<pd.Series, pd.DataFrame, pd.Panel>) – The pandas objects to put in contiguous order. splits (iterable<pd.Series>) – The non-overlapping splits, as boolean masks.
Returns:	The data in contiguous order.
Return type:	iterable<pd.Series, pd.DataFrame, pd.Panel>

skchem.data.converters.base.default_features()[source]¶

skchem.data.converters.base.default_pipeline()[source]¶

Return a default pipeline to be used for general datasets.

skchem.data.converters.bradley_open_mp module¶

class skchem.data.converters.bradley_open_mp.BradleyOpenMPConverter(directory, output_directory, output_filename='bradley_open_mp.h5')[source]¶

Bases: skchem.data.converters.base.Converter

static filter_bad(data)[source]¶

static fix_mp(data)[source]¶

static parse_data(path)[source]¶

skchem.data.converters.bursi_ames module¶

class skchem.data.converters.bursi_ames.BursiAmesConverter(directory, output_directory, output_filename='bursi_ames.h5')[source]¶

Bases: skchem.data.converters.base.Converter

skchem.data.converters.diversity_set module¶

# skchem.data.coverters.example

Formatter for the example dataset.

class skchem.data.converters.diversity_set.DiversityConverter(directory, output_directory, output_filename='diversity.h5')[source]¶

Bases: skchem.data.converters.base.Converter

Example Converter, using the NCI DTP Diversity Set III.

parse_file(path)[source]¶

synthetic_targets(index)[source]¶

skchem.data.converters.muller_ames module¶

class skchem.data.converters.muller_ames.MullerAmesConverter(directory, output_directory, output_filename='muller_ames.h5')[source]¶

Bases: skchem.data.converters.base.Converter

create_split_dict(splits, name)[source]¶

drop_indices(splits, indices)[source]¶

parse_splits(f_path)[source]¶

patch_data(data, patches)[source]¶

Patch smiles in a DataFrame with rewritten ones that specify diazo groups in rdkit friendly way.

skchem.data.converters.nmrshiftdb2 module¶

class skchem.data.converters.nmrshiftdb2.NMRShiftDB2Converter(directory, output_directory, output_filename='nmrshiftdb2.h5')[source]¶

Bases: skchem.data.converters.base.Converter

static combine_duplicates(data)[source]¶

Collect duplicate spectra into one dictionary. All shifts are collected into lists.

static extract_duplicates(data, kind='13c')[source]¶

Get all 13c duplicates.

static get_spectra(data)[source]¶

Retrieves spectra from raw data.

static log_dists(data)[source]¶

log_duplicates(data)[source]¶

static parse_data(filepath)[source]¶

Reads the raw datafile.

static process_spectra(data)[source]¶

Turn the string representations found in sdf file into a dictionary.

static squash_duplicates(data)[source]¶

Take the mean of all the duplicates. This is where we could do a bit more checking.

static to_frame(data)[source]¶

Convert a series of dictionaries to a dataframe.

skchem.data.converters.physprop module¶

class skchem.data.converters.physprop.PhysPropConverter(directory, output_directory, output_filename='physprop.h5')[source]¶

Bases: skchem.data.converters.base.Converter

drop_inconsistencies(data)[source]¶

extract(directory)[source]¶

static fix_temp(s, mean_range=5)[source]¶

process_bp(data)[source]¶

process_logP(data)[source]¶

process_logS(data)[source]¶

process_mp(data)[source]¶

process_sdf(path)[source]¶

process_targets(data)[source]¶

process_txt(path)[source]¶

skchem.data.converters.tox21 module¶

## skchem.data.transformers.tox21

Module defining transformation techniques for tox21.

class skchem.data.converters.tox21.Tox21Converter(directory, output_directory, output_filename='tox21.h5')[source]¶

Bases: skchem.data.converters.base.Converter

Class to build tox21 dataset.

extract(directory)[source]¶

static fix_assay_name(s)[source]¶

static fix_id(s)[source]¶

static patch_test(test)[source]¶

read_test(test, test_data)[source]¶

read_train(train)[source]¶

read_valid(valid)[source]¶

Module contents¶

class skchem.data.converters.DiversityConverter(directory, output_directory, output_filename='diversity.h5')[source]¶

Bases: skchem.data.converters.base.Converter

Example Converter, using the NCI DTP Diversity Set III.

parse_file(path)[source]¶

synthetic_targets(index)[source]¶

class skchem.data.converters.BursiAmesConverter(directory, output_directory, output_filename='bursi_ames.h5')[source]¶

Bases: skchem.data.converters.base.Converter

class skchem.data.converters.MullerAmesConverter(directory, output_directory, output_filename='muller_ames.h5')[source]¶

Bases: skchem.data.converters.base.Converter

create_split_dict(splits, name)[source]¶

drop_indices(splits, indices)[source]¶

parse_splits(f_path)[source]¶

patch_data(data, patches)[source]¶

Patch smiles in a DataFrame with rewritten ones that specify diazo groups in rdkit friendly way.

class skchem.data.converters.PhysPropConverter(directory, output_directory, output_filename='physprop.h5')[source]¶

Bases: skchem.data.converters.base.Converter

drop_inconsistencies(data)[source]¶

extract(directory)[source]¶

static fix_temp(s, mean_range=5)[source]¶

process_bp(data)[source]¶

process_logP(data)[source]¶

process_logS(data)[source]¶

process_mp(data)[source]¶

process_sdf(path)[source]¶

process_targets(data)[source]¶

process_txt(path)[source]¶

class skchem.data.converters.BradleyOpenMPConverter(directory, output_directory, output_filename='bradley_open_mp.h5')[source]¶

Bases: skchem.data.converters.base.Converter

static filter_bad(data)[source]¶

static fix_mp(data)[source]¶

static parse_data(path)[source]¶

class skchem.data.converters.NMRShiftDB2Converter(directory, output_directory, output_filename='nmrshiftdb2.h5')[source]¶

Bases: skchem.data.converters.base.Converter

static combine_duplicates(data)[source]¶

Collect duplicate spectra into one dictionary. All shifts are collected into lists.

static extract_duplicates(data, kind='13c')[source]¶

Get all 13c duplicates.

static get_spectra(data)[source]¶

Retrieves spectra from raw data.

static log_dists(data)[source]¶

log_duplicates(data)[source]¶

static parse_data(filepath)[source]¶

Reads the raw datafile.

static process_spectra(data)[source]¶

Turn the string representations found in sdf file into a dictionary.

static squash_duplicates(data)[source]¶

Take the mean of all the duplicates. This is where we could do a bit more checking.

static to_frame(data)[source]¶

Convert a series of dictionaries to a dataframe.

class skchem.data.converters.Tox21Converter(directory, output_directory, output_filename='tox21.h5')[source]¶

Bases: skchem.data.converters.base.Converter

Class to build tox21 dataset.

extract(directory)[source]¶

static fix_assay_name(s)[source]¶

static fix_id(s)[source]¶

static patch_test(test)[source]¶

read_test(test, test_data)[source]¶

read_train(train)[source]¶

read_valid(valid)[source]¶

class skchem.data.converters.ChEMBLConverter(directory, output_directory, output_filename='chembl.h5')[source]¶

Bases: skchem.data.converters.base.Converter

Converter for the ChEMBL dataset.

parse_infile(filename)[source]¶