Ejemplo n.º 1
0
def make_map():

    # representation types
    morg2 = RepresentationType(name='morg2',
        representation_func=skchemize(morg, radius=2, nBits=2048),
        metadata="""Hashed Circular fingerprint generated by the Morgan algorithm, """
            """implemented in <a href="http://www.rdkit.org">RDKit</a>. <br/>"""
            """Parameters used: Radius = 2, Bit length = 2048""")

    targets = RepresentationType(name='targets',
        representation_func=PIDGIN(),
        metadata="""Bayes affinity fingerprint for 1080 human targets, produced """
        """using the <a href="https://github.com/lhm30/PIDGIN">PIDGIN (Prediction of targets IncluDinG INactives)</a>"""
        """Target Prediction algorithm, implemented in <a href="https://github.com/richlewis42/scikit-chem">scikit-chem</a>.""")

    random = RepresentationType(name='random',
        representation_func=lambda m: pd.Series(np.random.random(100)),
        metadata="""Uniformly distributed random feature vector of length 100"""
        """implemented using <a href="http://www.numpy.org">numpy</a> <a href="http://docs.scipy.org/doc/numpy/reference/generated/numpy.random.random.html#numpy.random.random">random</a> module""")


    representation_types = [ morg2, targets, random ]

    # reduction types
    pca = ReductionMethod(name='PCA',
        model=PCA(n_components=2),
        metadata="""<a href="http://en.wikipedia.org/wiki/Principal_component_analysis">Principal component analysis</a> implemented in <a href="http://scikit-learn.org/stable/" target="_blank">scikit-learn</a>\n"""
        """<br/>Default parameters used.""")


    mds = ReductionMethod(name='MDS',
        model=MDS(),
        metadata=
            """<a href="http://en.wikipedia.org/wiki/Multidimensional_scaling" target="_blank">Multidimensional Scaling</a> implemented in <a href="http://scikit-learn.org/stable/" target="_blank">scikit-learn</a>"""
        """<br/>Default parameters used.""")


    tsne = ReductionMethod(name='t-SNE',
        model=TSNE(perplexity=1),
        metadata=
            """<a href="http://lvdmaaten.github.io/tsne/">Student's t-distributed stochastic neighbour embedding</a>, """
            """implemented according to <a href="http://lvdmaaten.github.io/publications/papers/JMLR_2008.pdf">van der Maartin et al. 2008</a>\n"""
            """<br/>Parameters used: Perplexity = 1, theta=0""")

    reduction_types = [ pca, mds, tsne ]

    # activity types
    pIC20 = ActivityType(name='pIC20', metadata=
        """negative based-10 logarithm of the <a href="http://en.wikipedia.org/wiki/IC50">IC20</a>, the concentation of"""
        """compound required for 20% inhibition of growth of Lymphoma cells""")

    IC20 = ActivityType(name='IC20', metadata=
        """<a href="http://en.wikipedia.org/wiki/IC50">IC20</a>, the concentation of"""
        """compound required for 20% inhibition of growth of Lymphoma cells""")

    activity_types = [ pIC20, IC20 ]

    # synergy types
    excessOverBliss = SynergyType(name='ExcessOverBliss', metadata=
        """Difference in observed vs expected activity of the component compounds,"""
        """each at the IC20 concentration (when known) assuming the <a href="http://doi.wiley.com/10.1111/j.1744-7348.1939.tb06990.x">Bliss Independence model</a>""")

    synergy_types = [ excessOverBliss ]

    # data
    compound_df = skc.read_smiles(os.path.join(DIRNAME, 'compounds.smiles'), name_column=1, title_line=True)
    compound_df['pIC20'] = -np.log10(compound_df['IC20'])

    combination_df = pd.read_csv(os.path.join(DIRNAME, 'combinations.csv'))
    combination_df.set_index('id', inplace=True)

    synergy_map = SynergyMap(compound_df=compound_df,
        combination_df=combination_df,
        representation_types=representation_types,
        reduction_types=reduction_types,
        activity_types=activity_types,
        synergy_types=synergy_types,
        metadata='DREAM Drug Combination Challenge Data')

    return synergy_map
Ejemplo n.º 2
0
def make_map():

    # representation types
    morg2 = RepresentationType(
        name='morg2',
        representation_func=skchemize(morg, radius=2, nBits=2048),
        metadata=
        """Hashed Circular fingerprint generated by the Morgan algorithm, """
        """implemented in <a href="http://www.rdkit.org">RDKit</a>. <br/>"""
        """Parameters used: Radius = 2, Bit length = 2048""")

    targets = RepresentationType(
        name='targets',
        representation_func=PIDGIN(),
        metadata=
        """Bayes affinity fingerprint for 1080 human targets, produced """
        """using the <a href="https://github.com/lhm30/PIDGIN">PIDGIN (Prediction of targets IncluDinG INactives)</a>"""
        """Target Prediction algorithm, implemented in <a href="https://github.com/richlewis42/scikit-chem">scikit-chem</a>."""
    )

    random = RepresentationType(
        name='random',
        representation_func=lambda m: pd.Series(np.random.random(100)),
        metadata="""Uniformly distributed random feature vector of length 100"""
        """implemented using <a href="http://www.numpy.org">numpy</a> <a href="http://docs.scipy.org/doc/numpy/reference/generated/numpy.random.random.html#numpy.random.random">random</a> module"""
    )

    representation_types = [morg2, targets, random]

    # reduction types
    pca = ReductionMethod(
        name='PCA',
        model=PCA(n_components=2),
        metadata=
        """<a href="http://en.wikipedia.org/wiki/Principal_component_analysis">Principal component analysis</a> implemented in <a href="http://scikit-learn.org/stable/" target="_blank">scikit-learn</a>\n"""
        """<br/>Default parameters used.""")

    mds = ReductionMethod(
        name='MDS',
        model=MDS(),
        metadata=
        """<a href="http://en.wikipedia.org/wiki/Multidimensional_scaling" target="_blank">Multidimensional Scaling</a> implemented in <a href="http://scikit-learn.org/stable/" target="_blank">scikit-learn</a>"""
        """<br/>Default parameters used.""")

    tsne = ReductionMethod(
        name='t-SNE',
        model=TSNE(perplexity=1),
        metadata=
        """<a href="http://lvdmaaten.github.io/tsne/">Student's t-distributed stochastic neighbour embedding</a>, """
        """implemented according to <a href="http://lvdmaaten.github.io/publications/papers/JMLR_2008.pdf">van der Maartin et al. 2008</a>\n"""
        """<br/>Parameters used: Perplexity = 1, theta=0""")

    reduction_types = [pca, mds, tsne]

    # activity types
    pIC20 = ActivityType(
        name='pIC20',
        metadata=
        """negative based-10 logarithm of the <a href="http://en.wikipedia.org/wiki/IC50">IC20</a>, the concentation of"""
        """compound required for 20% inhibition of growth of Lymphoma cells""")

    IC20 = ActivityType(
        name='IC20',
        metadata=
        """<a href="http://en.wikipedia.org/wiki/IC50">IC20</a>, the concentation of"""
        """compound required for 20% inhibition of growth of Lymphoma cells""")

    activity_types = [pIC20, IC20]

    # synergy types
    excessOverBliss = SynergyType(
        name='ExcessOverBliss',
        metadata=
        """Difference in observed vs expected activity of the component compounds,"""
        """each at the IC20 concentration (when known) assuming the <a href="http://doi.wiley.com/10.1111/j.1744-7348.1939.tb06990.x">Bliss Independence model</a>"""
    )

    synergy_types = [excessOverBliss]

    # data
    compound_df = skc.read_smiles(os.path.join(DIRNAME, 'compounds.smiles'),
                                  name_column=1,
                                  title_line=True)
    compound_df['pIC20'] = -np.log10(compound_df['IC20'])

    combination_df = pd.read_csv(os.path.join(DIRNAME, 'combinations.csv'))
    combination_df.set_index('id', inplace=True)

    synergy_map = SynergyMap(compound_df=compound_df,
                             combination_df=combination_df,
                             representation_types=representation_types,
                             reduction_types=reduction_types,
                             activity_types=activity_types,
                             synergy_types=synergy_types,
                             metadata='DREAM Drug Combination Challenge Data')

    return synergy_map
Ejemplo n.º 3
0
    def to_dict(self):
        """produce a dict representation of the type

        Args:
            None

        Returns:
            dict: A dictionary of the synergy type's name and metadata.
        """
        return {"name": self.name, "metadata": self.metadata}


morg2 = RepresentationType(
    name='morg2',
    representation_func=skchemize(morg, radius=2, nBits=2048),
    metadata=
    """Hashed Circular fingerprint generated by the Morgan algorithm, """
    """implemented in <a href="http://www.rdkit.org">RDKit</a>. <br/>"""
    """Parameters used: Radius = 2, Bit length = 2048""")

targets = RepresentationType(
    name='targets',
    representation_func=PIDGIN(),
    metadata="""Bayes affinity fingerprint for 1080 human targets, produced """
    """using the <a href="https://github.com/lhm30/PIDGIN">PIDGIN (Prediction of targets IncluDinG INactives)</a>"""
    """Target Prediction algorithm, implemented in <a href="https://github.com/richlewis42/scikit-chem">scikit-chem</a>."""
)

random = RepresentationType(
    name='random',
Ejemplo n.º 4
0
def make_map():

    morg2 = RepresentationType(name='morg2',
        representation_func=skchemize(morg, radius=2, nBits=2048),
        metadata="""Hashed Circular fingerprint generated by the Morgan algorithm, """
            """implemented in <a href="http://www.rdkit.org">RDKit</a>. <br/>"""
            """Parameters used: Radius = 2, Bit length = 2048""")

    targets = RepresentationType(name='targets',
        representation_func=PIDGIN(),
        metadata="""Bayes affinity fingerprint for 1080 human targets, produced """
        """using the <a href="https://github.com/lhm30/PIDGIN">PIDGIN (Prediction of targets IncluDinG INactives)</a>"""
        """Target Prediction algorithm, implemented in <a href="https://github.com/richlewis42/scikit-chem">scikit-chem</a>.""")

    random = RepresentationType(name='random',
        representation_func=lambda m: pd.Series(np.random.random(100)),
        metadata="""Uniformly distributed random feature vector of length 100"""
        """implemented using <a href="http://www.numpy.org">numpy</a> <a href="http://docs.scipy.org/doc/numpy/reference/generated/numpy.random.random.html#numpy.random.random">random</a> module""")


    representation_types = [ morg2, targets, random ]

    # reduction types

    pca = ReductionMethod(name='PCA',
        model=PCA(n_components=2),
        metadata="""<a href="http://en.wikipedia.org/wiki/Principal_component_analysis">Principal component analysis</a> implemented in <a href="http://scikit-learn.org/stable/" target="_blank">scikit-learn</a>\n"""
        """<br/>Default parameters used.""")


    mds = ReductionMethod(name='MDS',
        model=MDS(),
        metadata=
            """<a href="http://en.wikipedia.org/wiki/Multidimensional_scaling" target="_blank">Multidimensional Scaling</a> implemented in <a href="http://scikit-learn.org/stable/" target="_blank">scikit-learn</a>"""
        """<br/>Default parameters used.""")


    tsne = ReductionMethod(name='t-SNE',
        model=TSNE(perplexity=10),
        metadata=
            """<a href="http://lvdmaaten.github.io/tsne/">Student's t-distributed stochastic neighbour embedding</a>, """
            """implemented according to <a href="http://lvdmaaten.github.io/publications/papers/JMLR_2008.pdf">van der Maartin et al. 2008</a>\n"""
            """<br/>Parameters used: Perplexity = 10, theta=0""")

    reduction_types = [ pca, mds, tsne ]


    # activity types

    pIC50 = ActivityType(name='pIC50', metadata=
        """<a href="http://en.wikipedia.org/wiki/IC50">IC50</a>, the concentation of"""
        """compound required for 50% inhibition of growth of Malarial cells""")

    activity_types = [ pIC50 ]
    # synergy types

    MedianExcess = SynergyType(name='MedianExcess', metadata="")
    NumExcess = SynergyType(name='NumExcess', metadata="")
    LS3x3 = SynergyType(name='LS3x3', metadata="")
    DBSumPos = SynergyType(name='DBSumPos', metadata="")
    DBSumNeg = SynergyType(name='DBSumNeg', metadata="")
    pGamma = SynergyType(name='pGamma', metadata="")
    ExcessHSA = SynergyType(name='-ExcessHSA', metadata="")
    ExcessCRX = SynergyType(name='-ExcessCRX', metadata="")
    pGamma_scrambled = SynergyType(name='pGamma_scrambled', metadata="")

    synergy_types = [ pGamma, MedianExcess, NumExcess, LS3x3, DBSumPos, DBSumNeg, ExcessHSA, ExcessCRX, pGamma_scrambled ]

    # data
    compound_df = skc.read_sdf('compounds.sdf')
    compound_df['name'] = compound_df.Name
    compound_df.drop('Name', axis=1, inplace=True)
    compound_df['id'] = compound_df.index
    compound_df.drop('NCGC_ID', axis=1, inplace=True)
    compound_df.set_index('id', inplace=True)
    compound_df['pIC50'] = compound_df['pIC50'].apply(float)
    compound_df['IC50'] = compound_df.IC50.apply(float)

    combination_df = pd.read_csv('combinations.csv')
    combination_df.set_index('id', inplace=True)

    synergy_map = SynergyMap(compound_df=compound_df,
        combination_df=combination_df,
        representation_types=representation_types,
        reduction_types=reduction_types,
        activity_types=activity_types,
        synergy_types=synergy_types,
        metadata="Malaria NCATS dataset")

    return synergy_map
Ejemplo n.º 5
0
    def to_dict(self):
        """produce a dict representation of the type

        Args:
            None

        Returns:
            dict: A dictionary of the synergy type's name and metadata.
        """
        return {"name": self.name, "metadata": self.metadata}


morg2 = RepresentationType(
    name="morg2",
    representation_func=skchemize(morg, radius=2, nBits=2048),
    metadata="""Hashed Circular fingerprint generated by the Morgan algorithm, """
    """implemented in <a href="http://www.rdkit.org">RDKit</a>. <br/>"""
    """Parameters used: Radius = 2, Bit length = 2048""",
)

targets = RepresentationType(
    name="targets",
    representation_func=PIDGIN(),
    metadata="""Bayes affinity fingerprint for 1080 human targets, produced """
    """using the <a href="https://github.com/lhm30/PIDGIN">PIDGIN (Prediction of targets IncluDinG INactives)</a>"""
    """Target Prediction algorithm, implemented in <a href="https://github.com/richlewis42/scikit-chem">scikit-chem</a>.""",
)

random = RepresentationType(
    name="random",
Ejemplo n.º 6
0
def make_map():

    morg2 = RepresentationType(
        name='morg2',
        representation_func=skchemize(morg, radius=2, nBits=2048),
        metadata=
        """Hashed Circular fingerprint generated by the Morgan algorithm, """
        """implemented in <a href="http://www.rdkit.org">RDKit</a>. <br/>"""
        """Parameters used: Radius = 2, Bit length = 2048""")

    targets = RepresentationType(
        name='targets',
        representation_func=PIDGIN(),
        metadata=
        """Bayes affinity fingerprint for 1080 human targets, produced """
        """using the <a href="https://github.com/lhm30/PIDGIN">PIDGIN (Prediction of targets IncluDinG INactives)</a>"""
        """Target Prediction algorithm, implemented in <a href="https://github.com/richlewis42/scikit-chem">scikit-chem</a>."""
    )

    random = RepresentationType(
        name='random',
        representation_func=lambda m: pd.Series(np.random.random(100)),
        metadata="""Uniformly distributed random feature vector of length 100"""
        """implemented using <a href="http://www.numpy.org">numpy</a> <a href="http://docs.scipy.org/doc/numpy/reference/generated/numpy.random.random.html#numpy.random.random">random</a> module"""
    )

    representation_types = [morg2, targets, random]

    # reduction types

    pca = ReductionMethod(
        name='PCA',
        model=PCA(n_components=2),
        metadata=
        """<a href="http://en.wikipedia.org/wiki/Principal_component_analysis">Principal component analysis</a> implemented in <a href="http://scikit-learn.org/stable/" target="_blank">scikit-learn</a>\n"""
        """<br/>Default parameters used.""")

    mds = ReductionMethod(
        name='MDS',
        model=MDS(),
        metadata=
        """<a href="http://en.wikipedia.org/wiki/Multidimensional_scaling" target="_blank">Multidimensional Scaling</a> implemented in <a href="http://scikit-learn.org/stable/" target="_blank">scikit-learn</a>"""
        """<br/>Default parameters used.""")

    tsne = ReductionMethod(
        name='t-SNE',
        model=TSNE(perplexity=10),
        metadata=
        """<a href="http://lvdmaaten.github.io/tsne/">Student's t-distributed stochastic neighbour embedding</a>, """
        """implemented according to <a href="http://lvdmaaten.github.io/publications/papers/JMLR_2008.pdf">van der Maartin et al. 2008</a>\n"""
        """<br/>Parameters used: Perplexity = 10, theta=0""")

    reduction_types = [pca, mds, tsne]

    # activity types

    pIC50 = ActivityType(
        name='pIC50',
        metadata=
        """<a href="http://en.wikipedia.org/wiki/IC50">IC50</a>, the concentation of"""
        """compound required for 50% inhibition of growth of Malarial cells""")

    activity_types = [pIC50]
    # synergy types

    MedianExcess = SynergyType(name='MedianExcess', metadata="")
    NumExcess = SynergyType(name='NumExcess', metadata="")
    LS3x3 = SynergyType(name='LS3x3', metadata="")
    DBSumPos = SynergyType(name='DBSumPos', metadata="")
    DBSumNeg = SynergyType(name='DBSumNeg', metadata="")
    pGamma = SynergyType(name='pGamma', metadata="")
    ExcessHSA = SynergyType(name='-ExcessHSA', metadata="")
    ExcessCRX = SynergyType(name='-ExcessCRX', metadata="")
    pGamma_scrambled = SynergyType(name='pGamma_scrambled', metadata="")

    synergy_types = [
        pGamma, MedianExcess, NumExcess, LS3x3, DBSumPos, DBSumNeg, ExcessHSA,
        ExcessCRX, pGamma_scrambled
    ]

    # data
    compound_df = skc.read_sdf('compounds.sdf')
    compound_df['name'] = compound_df.Name
    compound_df.drop('Name', axis=1, inplace=True)
    compound_df['id'] = compound_df.index
    compound_df.drop('NCGC_ID', axis=1, inplace=True)
    compound_df.set_index('id', inplace=True)
    compound_df['pIC50'] = compound_df['pIC50'].apply(float)
    compound_df['IC50'] = compound_df.IC50.apply(float)

    combination_df = pd.read_csv('combinations.csv')
    combination_df.set_index('id', inplace=True)

    synergy_map = SynergyMap(compound_df=compound_df,
                             combination_df=combination_df,
                             representation_types=representation_types,
                             reduction_types=reduction_types,
                             activity_types=activity_types,
                             synergy_types=synergy_types,
                             metadata="Malaria NCATS dataset")

    return synergy_map