first pass at examples travis build, chembl in comments

      language: c

      python: '3.7'

      os: windows

    - name: DocTest Examples

      language: python

      python: '3.7'

      sudo: required

      dist: xenial

      env: DOCTEST_EXAMPLES=true

cache: pip

install:

  - if [[ "$TRAVIS_OS_NAME" != "windows" ]]; then

  - conda activate deepchem

  - pip install -e .

script:

  - if [[ "$DOCTEST_EXAMPLES" == "true" ]]; then

      cd docs && pip install -r requirements.txt;

      PYTHONWARNINGS= sphinx-build -M doctest ./ _build examples.rst

      exit

    fi

  - bash devtools/run_yapf.sh

  - bash devtools/run_flake8.sh

  - mypy -p deepchem

Examples

========

SAMPL models

============

Before jumping in to examples, we'll import our libraries and ensure our `doctests <https://www.sphinx-doc.org/en/master/usage/extensions/doctest.html>`_ are reproducible:

Some examples of training models on the SAMPL(FreeSolv) dataset included in :code:`dc.molnet`.

We'll be using its ``smiles`` field to train models to predict its experimentally measured solvation energy (``expt``).

First, we'll load our libraries:

.. doctest::

.. doctest:: *

    >>> import numpy as np

    >>> import tensorflow as tf

    >>> import deepchem as dc

   >>> from deepchem.molnet import load_sampl

    >>> 

   >>> # for reproducibility 

   >>> np.random.seed(123)

   >>> tf.random.set_seed(123)

    >>> # Run before every test for reproducibility

    >>> def seed_all():

    ...     np.random.seed(123)

    ...     tf.random.set_seed(123)

    >>>

.. testsetup:: *

.. doctest:: 

    import numpy as np

    import tensorflow as tf

    import deepchem as dc

   >>> # Load SAMPL dataset

   >>> SAMPL_tasks, SAMPL_datasets, transformers = load_sampl()

    # Run before every test for reproducibility

    def seed_all():

        np.random.seed(123)

        tf.random.set_seed(123)

SAMPL (FreeSolv)

----------------

Examples of training models on the SAMPL(FreeSolv) dataset included in `MoleculeNet <./moleculenet.html>`_.

We'll be using its :code:`smiles` field to train models to predict its experimentally measured solvation energy (:code:`expt`).

MultitaskRegressor

^^^^^^^^^^^^^^^^^^

First, we'll load the dataset with :func:`load_sampl() <deepchem.molnet.load_sampl>` and fit a :class:`MultitaskRegressor <deepchem.models.MultitaskRegressor>`:

.. doctest:: sampl

    >>> seed_all()

    >>> # Load SAMPL dataset with default 'index' splitting

    >>> SAMPL_tasks, SAMPL_datasets, transformers = dc.molnet.load_sampl()

    >>> SAMPL_tasks

    ['expt']

    >>> train_dataset, valid_dataset, test_dataset = SAMPL_datasets

    >>>

   >>> # We'll train a multitask regressor (fully connected network)

   >>> metric = dc.metrics.Metric(dc.metrics.pearson_r2_score, np.mean)

    >>> # We want to know the pearson R squared score, averaged across tasks

    >>> avg_pearson_r2 = dc.metrics.Metric(dc.metrics.pearson_r2_score, np.mean)

    >>> 

    >>> # We'll train a multitask regressor (fully connected network)

    >>> model = dc.models.MultitaskRegressor(

    ...     len(SAMPL_tasks),

    ...     n_features = 1024,

    >>> model.fit(train_dataset)

    0.1726440668106079

    >>> 

   >>> # We now evaluate our fitted model on our train and test sets

   >>> model.evaluate(train_dataset, [metric], transformers)

    >>> # We now evaluate our fitted model on our training and validation sets

    >>> model.evaluate(train_dataset, [avg_pearson_r2], transformers)

    {'mean-pearson_r2_score': 0.9244964295814636}

   >>> model.evaluate(valid_dataset, [metric], transformers)

    >>> model.evaluate(valid_dataset, [avg_pearson_r2], transformers)

    {'mean-pearson_r2_score': 0.7532658569385681}

For a :code:`GraphConvModel` we'll need to reload with the appropriate featurizer:

.. doctest:: 

   >>> # for reproducibility 

   >>> np.random.seed(123)

   >>> tf.random.set_seed(123)

GraphConvModel

^^^^^^^^^^^^^^

The default `featurizer <./featurizers.html>`_ for SAMPL is :code:`ECFP`, short for

`"Extended-connectivity fingerprints." <./featurizers.html#circularfingerprint>`_

For a :class:`GraphConvModel <deepchem.models.GraphConvModel>`, we'll reload our datasets with :code:`featurizer='GraphConv'`:

.. doctest:: sampl

    >>> seed_all()

    >>> # Load SAMPL dataset

   >>> SAMPL_tasks, SAMPL_datasets, transformers = load_sampl(

    >>> SAMPL_tasks, SAMPL_datasets, transformers = dc.molnet.load_sampl(

    ...     featurizer='GraphConv')

    >>> train_dataset, valid_dataset, test_dataset = SAMPL_datasets

    >>>

    >>> model.fit(train_dataset, nb_epoch=20)

    0.05753047466278076

    >>> 

   >>> model.evaluate(train_dataset, [metric], transformers)

    >>> # We now evaluate our fitted model on our training and validation sets

    >>> model.evaluate(train_dataset, [avg_pearson_r2], transformers)

    {'mean-pearson_r2_score': 0.5772751202910659}

   >>> model.evaluate(valid_dataset, [metric], transformers)

    >>> model.evaluate(valid_dataset, [avg_pearson_r2], transformers)

    {'mean-pearson_r2_score': 0.36771456280565507}

..

  ChEMBL

  -------

  Examples of training models on `ChEMBL <https://www.ebi.ac.uk/chembl/>` dataset included in `MoleculeNet <./moleculenet.html>`_.

    ChEMBL is a manually curated database of bioactive molecules with drug-like properties.

    It brings together chemical, bioactivity and genomic data to aid the translation of genomic information into effective new drugs.

  MultitaskRegressor

  ^^^^^^^^^^^^^^^^^^

  .. doctest:: chembl

      >>> seed_all()

      >>> # Load ChEMBL 5thresh dataset with random splitting

      >>> chembl_tasks, datasets, transformers = dc.molnet.load_chembl(

      ...     shard_size=2000, featurizer="ECFP", set="5thresh", split="random")

      >>> train_dataset, valid_dataset, test_dataset = datasets

      >>> len(chembl_tasks)

      691

      >>> f'Compound train/valid/test split: {len(train_dataset)}/{len(valid_dataset)}/{len(test_dataset)}'

      'Compound train/valid/test split: 19096/2387/2388'

      >>>

      >>> # We want to know the pearson R squared score, averaged across tasks

      >>> avg_pearson_r2 = dc.metrics.Metric(dc.metrics.pearson_r2_score, np.mean)

      >>>

      >>> # Create our model

      >>> n_layers = 3

      >>> model = dc.models.MultitaskRegressor(

      ...     len(chembl_tasks),

      ...     train_dataset.get_data_shape()[0],

      ...     layer_sizes=[1000] * n_layers,

      ...     dropouts=[.25] * n_layers,

      ...     weight_init_stddevs=[.02] * n_layers,

      ...     bias_init_consts=[1.] * n_layers,

      ...     learning_rate=.0003,

      ...     weight_decay_penalty=.0001,

      ...     batch_size=100,

      ...     verbosity="high")

      >>>

      >>> model.fit(train_dataset, nb_epoch=10) # orig. 20

      0.04922508895397186

      >>> # We now evaluate our fitted model on our training, validation, and test sets

      >>> model.evaluate(train_dataset, [avg_pearson_r2], transformers)

      {'mean-pearson_r2_score': nan}

      >>> model.evaluate(valid_dataset, [avg_pearson_r2], transformers)

      {'mean-pearson_r2_score': nan}

      >>> model.evaluate(test_dataset, [avg_pearson_r2], transformers)

      {'mean-pearson_r2_score': nan}

  GraphConvModel

  ^^^^^^^^^^^^^^

  .. doctest:: chembl

      >>> # Load ChEMBL dataset

      >>> chembl_tasks, datasets, transformers = dc.molnet.load_chembl(

      ...    shard_size=2000, featurizer="GraphConv", set="5thresh", split="random")

      >>> train_dataset, valid_dataset, test_dataset = datasets

      >>> 

      >>> # pearson R squared score, averaged across tasks

      >>> avg_pearson_r2 = dc.metrics.Metric(dc.metrics.pearson_r2_score, np.mean)

      >>>

      >>> model = dc.models.GraphConvModel(

      ...    len(chembl_tasks), batch_size=128, mode='regression')

      >>>

      >>> # Fit trained model

      >>> model.fit(train_dataset, nb_epoch=20)

      None

      >>>

      >>> # We now evaluate our fitted model on our training, validation, and test sets

      >>> model.evaluate(train_dataset, [avg_pearson_r2], transformers)

      {'mean-pearson_r2_score': nan}

      >>> model.evaluate(valid_dataset, [avg_pearson_r2], transformers) and False

      {'mean-pearson_r2_score': nan}

      >>> model.evaluate(test_dataset, [avg_pearson_r2], transformers) and False

      {'mean-pearson_r2_score': nan}