skchem.filters package¶

Submodules¶

skchem.filters.base module¶

# skchem.filters

Chemical filters are defined.

class skchem.filters.base.BaseFilter(agg='any', **kwargs)[source]¶

Bases: skchem.base.BaseTransformer

The base Filter class.

agg¶

callable – The aggregate function to use. String aliases for ‘any’, ‘not any’, ‘all’, ‘not all’ are available.

axes_names()[source]¶

columns¶

pd.Index – The column index to use.

filter(mols, y=None, neg=False)[source]¶

transform(mols, agg=True, **kwargs)[source]¶

class skchem.filters.base.Filter(func=None, agg='any', n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.BaseFilter, skchem.base.Transformer

Filter base class.

Examples

>>> import skchem

Initialize the filter with a function: >>> is_named = skchem.filters.Filter(lambda m: m.name is not None)

Filter results can be found with transform: >>> ethane = skchem.Mol.from_smiles(‘CC’, name=’ethane’) >>> is_named.transform(ethane) True

>>> anonymous = skchem.Mol.from_smiles('c1ccccc1')
>>> is_named.transform(anonymous)
False

Can take a series or dataframe: >>> mols = pd.Series({‘anonymous’: anonymous, ‘ethane’: ethane}) >>> is_named.transform(mols) anonymous False ethane True Name: Filter, dtype: bool

Using filter will drop out molecules that fail the test: >>> is_named.filter(mols) ethane <Mol: CC> dtype: object

Only failed are retained with the neg keyword argument: >>> is_named.filter(mols, neg=True) anonymous <Mol: c1ccccc1> dtype: object

class skchem.filters.base.TransformFilter(agg='any', **kwargs)[source]¶

Bases: skchem.filters.base.BaseFilter

Transform Filter object.

Implements transform_filter, which allows a transform, then a filter step returning the transformed values that are not False, None or np.nan.

transform_filter(mols, y=None, neg=False)[source]¶

skchem.filters.base.identity(x)[source]¶

The identity

skchem.filters.base.not_all(x)[source]¶

Not all x

skchem.filters.base.not_any(x)[source]¶

Not any x

skchem.filters.simple module¶

# skchem.filters.simple

Simple filters for compounds.

class skchem.filters.simple.AtomNumberFilter(above=3, below=60, include_hydrogens=False, n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter whether the number of atoms in a Mol falls in a defined interval.

above <= n_atoms < below

Examples

>>> import skchem

>>> data = [
...         skchem.Mol.from_smiles('CC', name='ethane'),
...         skchem.Mol.from_smiles('CCCC', name='butane'),
...         skchem.Mol.from_smiles('NC(C)C(=O)O', name='alanine'),
...         skchem.Mol.from_smiles('C12C=CC(C=C2)C=C1', name='barrelene')
... ]

>>> af = skchem.filters.AtomNumberFilter(above=3, below=7)

>>> af.transform(data)
ethane       False
butane        True
alanine       True
barrelene    False
Name: num_atoms_in_range, dtype: bool

>>> af.filter(data)
butane            <Mol: CCCC>
alanine    <Mol: CC(N)C(=O)O>
Name: structure, dtype: object

>>> af = skchem.filters.AtomNumberFilter(above=5, below=15, include_hydrogens=True)

>>> af.transform(data)
ethane        True
butane        True
alanine       True
barrelene    False
Name: num_atoms_in_range, dtype: bool

columns¶

class skchem.filters.simple.ElementFilter(elements=None, as_bits=False, agg='any', n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter by elements.

Examples

Basic usage on molecules:

>>> import skchem
>>> hal_f = skchem.filters.ElementFilter(['F', 'Cl', 'Br', 'I'])

Molecules with one of the atoms transform to True.

>>> m1 = skchem.Mol.from_smiles('ClC(Cl)Cl', name='chloroform')
>>> hal_f.transform(m1)
True

Molecules with none of the atoms transform to False.

>>> m2 = skchem.Mol.from_smiles('CC', name='ethane')
>>> hal_f.transform(m2)
False

Can see the atom breakdown by passing agg == False: >>> hal_f.transform(m1, agg=False) has_element F 0 Cl 3 Br 0 I 0 Name: ElementFilter, dtype: int64

Can transform series.

>>> ms = [m1, m2]
>>> hal_f.transform(ms)
chloroform     True
ethane        False
dtype: bool

>>> hal_f.transform(ms, agg=False)
has_element  F  Cl  Br  I
chloroform   0   3   0  0
ethane       0   0   0  0

Can also filter series:

>>> hal_f.filter(ms)
chloroform    <Mol: ClC(Cl)Cl>
Name: structure, dtype: object

>>> hal_f.filter(ms, neg=True)
ethane    <Mol: CC>
Name: structure, dtype: object

columns¶

elements¶

class skchem.filters.simple.MassFilter(above=3, below=900, n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter whether the molecular weight of a molecule is outside a range.

above <= mass < below

Examples

>>> import skchem

>>> data = [
...         skchem.Mol.from_smiles('CC', name='ethane'),
...         skchem.Mol.from_smiles('CCCC', name='butane'),
...         skchem.Mol.from_smiles('NC(C)C(=O)O', name='alanine'),
...         skchem.Mol.from_smiles('C12C=CC(C=C2)C=C1', name='barrelene')
... ]

>>> mf = skchem.filters.MassFilter(above=31, below=100)

>>> mf.transform(data)
ethane       False
butane        True
alanine       True
barrelene    False
Name: mass_in_range, dtype: bool

>>> mf.filter(data)
butane            <Mol: CCCC>
alanine    <Mol: CC(N)C(=O)O>
Name: structure, dtype: object

columns¶

class skchem.filters.simple.OrganicFilter(n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.simple.ElementFilter

Whether a molecule is organic.

For the purpose of this function, an organic molecule is defined as having atoms with elements only in the set H, B, C, N, O, F, P, S, Cl, Br, I.

Examples

Basic usage as a function on molecules: >>> import skchem >>> of = skchem.filters.OrganicFilter() >>> benzene = skchem.Mol.from_smiles(‘c1ccccc1’, name=’benzene’)

>>> of.transform(benzene)
True

>>> ferrocene = skchem.Mol.from_smiles('[cH-]1cccc1.[cH-]1cccc1.[Fe+2]',
...                                    name='ferrocene')
>>> of.transform(ferrocene)
False

More useful on collections:

>>> sa = skchem.Mol.from_smiles('CC(=O)[O-].[Na+]', name='sodium acetate')
>>> norbornane = skchem.Mol.from_smiles('C12CCC(C2)CC1', name='norbornane')

>>> data = [benzene, ferrocene, norbornane, sa]
>>> of.transform(data)
benzene            True
ferrocene         False
norbornane         True
sodium acetate    False
dtype: bool

>>> of.filter(data)
benzene          <Mol: c1ccccc1>
norbornane    <Mol: C1CC2CCC1C2>
Name: structure, dtype: object

>>> of.filter(data, neg=True)
ferrocene         <Mol: [Fe+2].c1cc[cH-]c1.c1cc[cH-]c1>
sodium acetate                  <Mol: CC(=O)[O-].[Na+]>
Name: structure, dtype: object

skchem.filters.simple.mass(mol, above=10, below=900)[source]¶

Whether a the molecular weight of a molecule is lower than a threshold.

above <= mass < below

Parameters:	mol – (skchem.Mol): The molecule to be tested. above (float) – The lower threshold on the mass. Defaults to None. below (float) – The higher threshold on the mass. Defaults to None.
Returns:	Whether the mass of the molecule is lower than the threshold.
Return type:	bool

Examples

Basic usage as a function on molecules:

>>> import skchem
>>> m = skchem.Mol.from_smiles('c1ccccc1') # benzene has M_r = 78.
>>> skchem.filters.mass(m, above=70)
True
>>> skchem.filters.mass(m, above=80)
False
>>> skchem.filters.mass(m, below=80)
True
>>> skchem.filters.mass(m, below=70)
False
>>> skchem.filters.mass(m, above=70, below=80)
True

skchem.filters.simple.n_atoms(mol, above=2, below=75, include_hydrogens=False)[source]¶

Whether the number of atoms in a molecule falls in a defined interval.

above <= n_atoms < below

Parameters:	mol – (skchem.Mol): The molecule to be tested. above (int) – The lower threshold number of atoms (exclusive). below (int) – The higher threshold number of atoms (inclusive). include_hydrogens (bool) – Whether to consider hydrogens in the atom count.
Returns:	Whether the molecule has more atoms than the threshold.
Return type:	bool

Examples

Basic usage as a function on molecules:

>>> import skchem
>>> m = skchem.Mol.from_smiles('c1ccccc1') # benzene has 6 atoms.

Lower threshold:

>>> skchem.filters.n_atoms(m, above=3)
True
>>> skchem.filters.n_atoms(m, above=8)
False

Higher threshold:

>>> skchem.filters.n_atoms(m, below=8)
True
>>> skchem.filters.n_atoms(m, below=3)
False

Bounds work like Python slices - inclusive lower, exclusive upper:

>>> skchem.filters.n_atoms(m, above=6)
True
>>> skchem.filters.n_atoms(m, below=6)
False

Both can be used at once:

>>> skchem.filters.n_atoms(m, above=3, below=8)
True

Can include hydrogens:

>>> skchem.filters.n_atoms(m, above=3, below=8, include_hydrogens=True)
False
>>> skchem.filters.n_atoms(m, above=9, below=14, include_hydrogens=True)
True

skchem.filters.smarts module¶

# skchem.filters.smarts

Module defines SMARTS filters.

class skchem.filters.smarts.PAINSFilter(n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.smarts.SMARTSFilter

Whether a molecule passes the Pan Assay INterference (PAINS) filters.

These are supplied with RDKit, and were originally proposed by Baell et al.

_pains¶

pd.Series – a series of smarts template molecules.

References

[The original paper](http://dx.doi.org/10.1021/jm901137j)

Examples

Basic usage as a function on molecules:

>>> import skchem
>>> benzene = skchem.Mol.from_smiles('c1ccccc1', name='benzene')
>>> pf = skchem.filters.PAINSFilter()
>>> pf.transform(benzene)
True
>>> catechol = skchem.Mol.from_smiles('Oc1c(O)cccc1', name='catechol')
>>> pf.transform(catechol)
False

>>> res = pf.transform(catechol, agg=False)
>>> res[res]
names
catechol_A(92)    True
Name: PAINSFilter, dtype: bool

More useful in combination with pandas DataFrames:

>>> data = [benzene, catechol]
>>> pf.transform(data)
benzene      True
catechol    False
dtype: bool

>>> pf.filter(data)
benzene    <Mol: c1ccccc1>
Name: structure, dtype: object

class skchem.filters.smarts.SMARTSFilter(smarts, agg='any', merge_hs=True, n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter a molecule based on smarts.

Examples

>>> import skchem

>>> data = [
...         skchem.Mol.from_smiles('CC', name='ethane'),
...         skchem.Mol.from_smiles('c1ccccc1', name='benzene'),
...         skchem.Mol.from_smiles('c1ccccc1-c2c(C=O)ccnc2', name='bg')
... ]

>>> f = skchem.filters.SMARTSFilter({'benzene': 'c1ccccc1',
...                                  'pyridine': 'c1ccccn1',
...                                  'acetyl': 'C=O'})
>>> f.transform(data, agg=False)
        acetyl benzene pyridine
ethane   False   False    False
benzene  False    True    False
bg       True    True     True

>>> f.transform(data)
ethane     False
benzene     True
bg         True
dtype: bool

>>> f.filter(data)
benzene                <Mol: c1ccccc1>
bg        <Mol: O=Cc1ccncc1-c1ccccc1>
Name: structure, dtype: object

>>> f.agg = all
>>> f.filter(data)
bg    <Mol: O=Cc1ccncc1-c1ccccc1>
Name: structure, dtype: object

columns¶

skchem.filters.stereo module¶

# skchem.filters.stereo

Stereo filters for scikit-chem.

class skchem.filters.stereo.ChiralFilter(check_meso=True, n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter chiral compounds.

Examples

>>> import skchem
>>> cf = skchem.filters.ChiralFilter()
>>> ms = [
...    skchem.Mol.from_smiles('F[C@@H](F)[C@H](F)F', name='achiral'),
...    skchem.Mol.from_smiles('F[C@@H](Br)[C@H](Br)F', name='chiral'),
...    skchem.Mol.from_smiles('F[C@H](Br)[C@H](Br)F', name='meso'),
...    skchem.Mol.from_smiles('FC(Br)C(Br)F', name='racemic')
... ]
>>> cf.transform(ms)
achiral    False
chiral      True
meso       False
racemic    False
Name: is_chiral, dtype: bool

columns¶

static is_meso(mol)[source]¶

Determines whether the molecule is meso.

Meso compounds have chiral centres, but has a mirror plane allowing superposition.

Examples

>>> import skchem

>>> cf = skchem.filters.ChiralFilter()

>>> meso = skchem.Mol.from_smiles('F[C@H](Br)[C@H](Br)F')

>>> cf.is_meso(meso)
True

>>> non_meso = skchem.Mol.from_smiles('F[C@H](Br)[C@@H](Br)F')
>>> cf.is_meso(non_meso)
False

Module contents¶

# skchem.filters

Molecule filters for scikit-chem.

class skchem.filters.ChiralFilter(check_meso=True, n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter chiral compounds.

Examples

>>> import skchem
>>> cf = skchem.filters.ChiralFilter()
>>> ms = [
...    skchem.Mol.from_smiles('F[C@@H](F)[C@H](F)F', name='achiral'),
...    skchem.Mol.from_smiles('F[C@@H](Br)[C@H](Br)F', name='chiral'),
...    skchem.Mol.from_smiles('F[C@H](Br)[C@H](Br)F', name='meso'),
...    skchem.Mol.from_smiles('FC(Br)C(Br)F', name='racemic')
... ]
>>> cf.transform(ms)
achiral    False
chiral      True
meso       False
racemic    False
Name: is_chiral, dtype: bool

columns¶

static is_meso(mol)[source]¶

Determines whether the molecule is meso.

Meso compounds have chiral centres, but has a mirror plane allowing superposition.

Examples

>>> import skchem

>>> cf = skchem.filters.ChiralFilter()

>>> meso = skchem.Mol.from_smiles('F[C@H](Br)[C@H](Br)F')

>>> cf.is_meso(meso)
True

>>> non_meso = skchem.Mol.from_smiles('F[C@H](Br)[C@@H](Br)F')
>>> cf.is_meso(non_meso)
False

class skchem.filters.SMARTSFilter(smarts, agg='any', merge_hs=True, n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter a molecule based on smarts.

Examples

>>> import skchem

>>> data = [
...         skchem.Mol.from_smiles('CC', name='ethane'),
...         skchem.Mol.from_smiles('c1ccccc1', name='benzene'),
...         skchem.Mol.from_smiles('c1ccccc1-c2c(C=O)ccnc2', name='bg')
... ]

>>> f = skchem.filters.SMARTSFilter({'benzene': 'c1ccccc1',
...                                  'pyridine': 'c1ccccn1',
...                                  'acetyl': 'C=O'})
>>> f.transform(data, agg=False)
        acetyl benzene pyridine
ethane   False   False    False
benzene  False    True    False
bg       True    True     True

>>> f.transform(data)
ethane     False
benzene     True
bg         True
dtype: bool

>>> f.filter(data)
benzene                <Mol: c1ccccc1>
bg        <Mol: O=Cc1ccncc1-c1ccccc1>
Name: structure, dtype: object

>>> f.agg = all
>>> f.filter(data)
bg    <Mol: O=Cc1ccncc1-c1ccccc1>
Name: structure, dtype: object

columns¶

class skchem.filters.PAINSFilter(n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.smarts.SMARTSFilter

Whether a molecule passes the Pan Assay INterference (PAINS) filters.

These are supplied with RDKit, and were originally proposed by Baell et al.

_pains¶

pd.Series – a series of smarts template molecules.

References

[The original paper](http://dx.doi.org/10.1021/jm901137j)

Examples

Basic usage as a function on molecules:

>>> import skchem
>>> benzene = skchem.Mol.from_smiles('c1ccccc1', name='benzene')
>>> pf = skchem.filters.PAINSFilter()
>>> pf.transform(benzene)
True
>>> catechol = skchem.Mol.from_smiles('Oc1c(O)cccc1', name='catechol')
>>> pf.transform(catechol)
False

>>> res = pf.transform(catechol, agg=False)
>>> res[res]
names
catechol_A(92)    True
Name: PAINSFilter, dtype: bool

More useful in combination with pandas DataFrames:

>>> data = [benzene, catechol]
>>> pf.transform(data)
benzene      True
catechol    False
dtype: bool

>>> pf.filter(data)
benzene    <Mol: c1ccccc1>
Name: structure, dtype: object

class skchem.filters.ElementFilter(elements=None, as_bits=False, agg='any', n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter by elements.

Examples

Basic usage on molecules:

>>> import skchem
>>> hal_f = skchem.filters.ElementFilter(['F', 'Cl', 'Br', 'I'])

Molecules with one of the atoms transform to True.

>>> m1 = skchem.Mol.from_smiles('ClC(Cl)Cl', name='chloroform')
>>> hal_f.transform(m1)
True

Molecules with none of the atoms transform to False.

>>> m2 = skchem.Mol.from_smiles('CC', name='ethane')
>>> hal_f.transform(m2)
False

Can see the atom breakdown by passing agg == False: >>> hal_f.transform(m1, agg=False) has_element F 0 Cl 3 Br 0 I 0 Name: ElementFilter, dtype: int64

Can transform series.

>>> ms = [m1, m2]
>>> hal_f.transform(ms)
chloroform     True
ethane        False
dtype: bool

>>> hal_f.transform(ms, agg=False)
has_element  F  Cl  Br  I
chloroform   0   3   0  0
ethane       0   0   0  0

Can also filter series:

>>> hal_f.filter(ms)
chloroform    <Mol: ClC(Cl)Cl>
Name: structure, dtype: object

>>> hal_f.filter(ms, neg=True)
ethane    <Mol: CC>
Name: structure, dtype: object

columns¶

elements¶

class skchem.filters.OrganicFilter(n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.simple.ElementFilter

Whether a molecule is organic.

For the purpose of this function, an organic molecule is defined as having atoms with elements only in the set H, B, C, N, O, F, P, S, Cl, Br, I.

Examples

Basic usage as a function on molecules: >>> import skchem >>> of = skchem.filters.OrganicFilter() >>> benzene = skchem.Mol.from_smiles(‘c1ccccc1’, name=’benzene’)

>>> of.transform(benzene)
True

>>> ferrocene = skchem.Mol.from_smiles('[cH-]1cccc1.[cH-]1cccc1.[Fe+2]',
...                                    name='ferrocene')
>>> of.transform(ferrocene)
False

More useful on collections:

>>> sa = skchem.Mol.from_smiles('CC(=O)[O-].[Na+]', name='sodium acetate')
>>> norbornane = skchem.Mol.from_smiles('C12CCC(C2)CC1', name='norbornane')

>>> data = [benzene, ferrocene, norbornane, sa]
>>> of.transform(data)
benzene            True
ferrocene         False
norbornane         True
sodium acetate    False
dtype: bool

>>> of.filter(data)
benzene          <Mol: c1ccccc1>
norbornane    <Mol: C1CC2CCC1C2>
Name: structure, dtype: object

>>> of.filter(data, neg=True)
ferrocene         <Mol: [Fe+2].c1cc[cH-]c1.c1cc[cH-]c1>
sodium acetate                  <Mol: CC(=O)[O-].[Na+]>
Name: structure, dtype: object

class skchem.filters.AtomNumberFilter(above=3, below=60, include_hydrogens=False, n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter whether the number of atoms in a Mol falls in a defined interval.

above <= n_atoms < below

Examples

>>> import skchem

>>> data = [
...         skchem.Mol.from_smiles('CC', name='ethane'),
...         skchem.Mol.from_smiles('CCCC', name='butane'),
...         skchem.Mol.from_smiles('NC(C)C(=O)O', name='alanine'),
...         skchem.Mol.from_smiles('C12C=CC(C=C2)C=C1', name='barrelene')
... ]

>>> af = skchem.filters.AtomNumberFilter(above=3, below=7)

>>> af.transform(data)
ethane       False
butane        True
alanine       True
barrelene    False
Name: num_atoms_in_range, dtype: bool

>>> af.filter(data)
butane            <Mol: CCCC>
alanine    <Mol: CC(N)C(=O)O>
Name: structure, dtype: object

>>> af = skchem.filters.AtomNumberFilter(above=5, below=15, include_hydrogens=True)

>>> af.transform(data)
ethane        True
butane        True
alanine       True
barrelene    False
Name: num_atoms_in_range, dtype: bool

columns¶

class skchem.filters.MassFilter(above=3, below=900, n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.Filter

Filter whether the molecular weight of a molecule is outside a range.

above <= mass < below

Examples

>>> import skchem

>>> data = [
...         skchem.Mol.from_smiles('CC', name='ethane'),
...         skchem.Mol.from_smiles('CCCC', name='butane'),
...         skchem.Mol.from_smiles('NC(C)C(=O)O', name='alanine'),
...         skchem.Mol.from_smiles('C12C=CC(C=C2)C=C1', name='barrelene')
... ]

>>> mf = skchem.filters.MassFilter(above=31, below=100)

>>> mf.transform(data)
ethane       False
butane        True
alanine       True
barrelene    False
Name: mass_in_range, dtype: bool

>>> mf.filter(data)
butane            <Mol: CCCC>
alanine    <Mol: CC(N)C(=O)O>
Name: structure, dtype: object

columns¶

class skchem.filters.Filter(func=None, agg='any', n_jobs=1, verbose=True)[source]¶

Bases: skchem.filters.base.BaseFilter, skchem.base.Transformer

Filter base class.

Examples

>>> import skchem

Initialize the filter with a function: >>> is_named = skchem.filters.Filter(lambda m: m.name is not None)

Filter results can be found with transform: >>> ethane = skchem.Mol.from_smiles(‘CC’, name=’ethane’) >>> is_named.transform(ethane) True

>>> anonymous = skchem.Mol.from_smiles('c1ccccc1')
>>> is_named.transform(anonymous)
False

Can take a series or dataframe: >>> mols = pd.Series({‘anonymous’: anonymous, ‘ethane’: ethane}) >>> is_named.transform(mols) anonymous False ethane True Name: Filter, dtype: bool

Using filter will drop out molecules that fail the test: >>> is_named.filter(mols) ethane <Mol: CC> dtype: object

Only failed are retained with the neg keyword argument: >>> is_named.filter(mols, neg=True) anonymous <Mol: c1ccccc1> dtype: object