[DGL-LifeSci] Documentation (#1414)

* Update

* Update

* Update

apps/life_sci/README.md

 # DGL-LifeSci
 
+[Documentation](https://lifesci.dgl.ai/index.html) | [Discussion Forum](https://discuss.dgl.ai)
+
 ## Introduction
 
 Deep learning on graphs has been an arising trend in the past few years. There are a lot of graphs in 

apps/life_sci/docs/source/api/data.rst

 .. _apidata:
 
-dgllife.data
-============
+Datasets
+========
 
-TBD by Murphy
+.. contents:: Contents
+    :local:
 
-dgllife.data.alchemy
---------------------
+Molecular Property Prediction
+-----------------------------
 
-.. automodule:: dgllife.data.alchemy
-    :members:
+Tox21
+`````
 
+.. autoclass:: dgllife.data.Tox21
+    :members: task_pos_weights, __getitem__, __len__
+    :show-inheritance:
 
-dgllife.data.csv_dataset
-------------------------
+Alchemy for Quantum Chemistry
+`````````````````````````````
 
-.. automodule:: dgllife.data.csv_dataset
-    :members:
+.. autoclass:: dgllife.data.TencentAlchemyDataset
+    :members: set_mean_and_std, __getitem__, __len__
 
+Pubmed Aromaticity
+``````````````````
 
-dgllife.data.pdbbind
----------------------
+.. autoclass:: dgllife.data.PubChemBioAssayAromaticity
+    :members: __getitem__, __len__
+    :show-inheritance:
 
-.. automodule:: dgllife.data.pdbbind
-    :members:
+Adapting to New Datasets with CSV
+`````````````````````````````````
 
+.. autoclass:: dgllife.data.MoleculeCSVDataset
+    :members: __getitem__, __len__
 
-dgllife.data.pubchem_aromaticity
----------------------------------
+Reaction Prediction
+-------------------
 
-.. automodule:: dgllife.data.pubchem_aromaticity
-    :members:
+USPTO
+`````
 
+.. autoclass:: dgllife.data.USPTO
+    :members: __getitem__, __len__
+    :show-inheritance:
 
-dgllife.data.tox21
----------------------------------
+Adapting to New Datasets for Weisfeiler-Lehman Networks
+```````````````````````````````````````````````````````
 
-.. automodule:: dgllife.data.tox21
-    :members:
+.. autoclass:: dgllife.data.WLNReactionDataset
+    :members: __getitem__, __len__
 
+Protein-Ligand Binding Affinity Prediction
+------------------------------------------
 
-dgllife.data.uspto
----------------------------------
+PDBBind
+```````
 
-.. automodule:: dgllife.data.uspto
-    :members:
+.. autoclass:: dgllife.data.PDBBind
+    :members: __getitem__, __len__

apps/life_sci/docs/source/api/model.gnn.rst

 .. _apimodelgnn:
 
-dgllife.model.gnn
-==================
+Graph Neural Networks for Updating Node/Edge Representations
+============================================================
 
-TBD by Murphy
+All models based on graph neural networks start with updating node/edge representations.
+We introduce various GNN models implemented in DGL-LifeSci for representation update.
 
-dgllife.model.gnn.attentivefp
--------------------------------------------
+.. contents:: Contents
+    :local:
+
+AttentiveFP
+-----------
 .. automodule:: dgllife.model.gnn.attentivefp
     :members:
 
-
-dgllife.model.gnn.gat
-----------------------------
+GAT
+---
 .. automodule:: dgllife.model.gnn.gat
     :members:
 
-dgllife.model.gnn.gcn
-----------------------------
+GCN
+---
 .. automodule:: dgllife.model.gnn.gcn
     :members:
 
-dgllife.model.gnn.mgcn
-----------------------------
+MGCN
+----
 .. automodule:: dgllife.model.gnn.mgcn
     :members:
 
-dgllife.model.gnn.mpnn
-----------------------------
+MPNN
+----
 .. automodule:: dgllife.model.gnn.mpnn
     :members:
 
-dgllife.model.gnn.schnet
-----------------------------
+SchNet
+------
 .. automodule:: dgllife.model.gnn.schnet
     :members:
 
-dgllife.model.gnn.wln
-----------------------------
+WLN
+---
 .. automodule:: dgllife.model.gnn.wln
     :members:

apps/life_sci/docs/source/api/model.pretrain.rst

+.. _apimodelpretrain:
+
+Pre-trained Models
+==================
+
+We provide multiple pre-trained models for users to use without the need of training from scratch.
+
+Example Usage
+-------------
+
+Property Prediction
+```````````````````
+
+.. code-block:: python
+
+    from dgllife.data import Tox21
+    from dgllife.model import load_pretrained
+    from dgllife.utils import smiles_to_bigraph, CanonicalAtomFeaturizer
+
+    dataset = Tox21(smiles_to_bigraph, CanonicalAtomFeaturizer())
+    model = load_pretrained('GCN_Tox21') # Pretrained model loaded
+    model.eval()
+
+    smiles, g, label, mask = dataset[0]
+    feats = g.ndata.pop('h')
+    label_pred = model(g, feats)
+    print(smiles)                   # CCOc1ccc2nc(S(N)(=O)=O)sc2c1
+    print(label_pred[:, mask != 0]) # Mask non-existing labels
+    # tensor([[ 1.4190, -0.1820,  1.2974,  1.4416,  0.6914,
+    # 2.0957,  0.5919,  0.7715, 1.7273,  0.2070]])
+
+Generative Models
+
+.. code-block:: python
+
+    from dgllife.model import load_pretrained
+
+    model = load_pretrained('DGMG_ZINC_canonical')
+    model.eval()
+    smiles = []
+    for i in range(4):
+        smiles.append(model(rdkit_mol=True))
+
+    print(smiles)
+    # ['CC1CCC2C(CCC3C2C(NC2=CC(Cl)=CC=C2N)S3(=O)=O)O1',
+    # 'O=C1SC2N=CN=C(NC(SC3=CC=CC=N3)C1=CC=CO)C=2C1=CCCC1',
+    # 'CC1C=CC(=CC=1)C(=O)NN=C(C)C1=CC=CC2=CC=CC=C21',
+    # 'CCN(CC1=CC=CC=C1F)CC1CCCN(C)C1']
+
+If you are running the code block above in Jupyter notebook, you can also visualize the molecules generated with
+
+.. code-block:: python
+
+    from IPython.display import SVG
+    from rdkit import Chem
+    from rdkit.Chem import Draw
+
+    mols = [Chem.MolFromSmiles(s) for s in smiles]
+    SVG(Draw.MolsToGridImage(mols, molsPerRow=4, subImgSize=(180, 150), useSVG=True))
+
+.. image:: https://data.dgl.ai/dgllife/dgmg/dgmg_model_zoo_example2.png
+
+API
+---
+
+.. autofunction:: dgllife.model.load_pretrained

apps/life_sci/docs/source/api/model.readout.rst

 .. _apimodelreadout:
 
-dgllife.model.readout
-========================
+Readout for Computing Graph Representations
+===========================================
 
-TBD by Murphy
+After updating node/edge representations with graph neural networks (GNNs), a common operation is to compute
+graph representations out of updated node/edge representations. For example, we need to compute molecular
+representations out of atom/bond representations in molecular property prediction. We call the various modules
+for computing graph-level representations **readout** as in Neural Message Passing for Quantum Chemistry and this
+section lists the readout modules implemented in DGL-LifeSci.
 
-dgllife.model.readout.attentivefp_readout
-------------------------------------------
+.. contents:: Contents
+    :local:
+
+AttentiveFP Readout
+-------------------
 .. automodule:: dgllife.model.readout.attentivefp_readout
     :members:
 
-
-
-dgllife.model.readout.mlp_readout
-------------------------------------------
+MLP Readout
+-----------
 .. automodule:: dgllife.model.readout.mlp_readout
     :members:
 
-
-
-dgllife.model.readout.weighted_sum_and_max
---------------------------------------------
+Weighted Sum and Max Readout
+----------------------------
 .. automodule:: dgllife.model.readout.weighted_sum_and_max
     :members:

apps/life_sci/docs/source/api/model.rst

-.. _apimodelgnn:
-
-dgllife.model
-==================
-
-TBD by Murphy
-
-.. autofunction:: dgllife.model.load_pretrained

apps/life_sci/docs/source/api/model.zoo.rst

 .. _apimodelzoo:
 
-dgllife.model.model_zoo
-=============================
+Model Zoo
+=========
 
-TBD by Murphy
+This section introduces complete models for various downstream tasks.
 
-dgllife.model.model_zoo.attentivefp_predictor
------------------------------------------------
+.. contents:: Contents
+    :local:
+
+Building Blocks
+---------------
+
+MLP Predictor
+`````````````
+.. automodule:: dgllife.model.model_zoo.mlp_predictor
+    :members:
+
+Molecular Property Prediction
+-----------------------------
+
+AttentiveFP Predictor
+`````````````````````
 .. automodule:: dgllife.model.model_zoo.attentivefp_predictor
     :members:
 
-dgllife.model.model_zoo.gat_predictor
--------------------------------------------
+GAT Predictor
+`````````````
 .. automodule:: dgllife.model.model_zoo.gat_predictor
     :members:
 
-dgllife.model.model_zoo.gcn_predictor
--------------------------------------------
+GCN Predictor
+`````````````
 .. automodule:: dgllife.model.model_zoo.gcn_predictor
     :members:
 
-dgllife.model.model_zoo.mgcn_predictor
--------------------------------------------
+MGCN Predictor
+``````````````
 .. automodule:: dgllife.model.model_zoo.mgcn_predictor
     :members:
 
-dgllife.model.model_zoo.mlp_predictor
--------------------------------------------
-.. automodule:: dgllife.model.model_zoo.mlp_predictor
-    :members:
-
-dgllife.model.model_zoo.mpnn_predictor
--------------------------------------------
+MPNN Predictor
+``````````````
 .. automodule:: dgllife.model.model_zoo.mpnn_predictor
     :members:
 
-dgllife.model.model_zoo.schnet_predictor
--------------------------------------------
+SchNet Predictor
+````````````````
 .. automodule:: dgllife.model.model_zoo.schnet_predictor
     :members:
 
-dgllife.model.model_zoo.wln_reaction_center
--------------------------------------------
-.. automodule:: dgllife.model.model_zoo.wln_reaction_center
+Generative Models
+-----------------
+
+DGMG
+````
+.. automodule:: dgllife.model.model_zoo.dgmg
     :members:
 
-dgllife.model.model_zoo.acnn
--------------------------------------------
-.. automodule:: dgllife.model.model_zoo.acnn
+JTNN
+````
+.. autoclass:: dgllife.model.model_zoo.jtnn.DGLJTNNVAE
     :members:
 
-dgllife.model.model_zoo.dgmg
--------------------------------------------
-.. automodule:: dgllife.model.model_zoo.dgmg
+Reaction Prediction
+
+WLN for Reaction Center Prediction
+``````````````````````````````````
+.. automodule:: dgllife.model.model_zoo.wln_reaction_center
     :members:
 
-dgllife.model.model_zoo.jtnn
--------------------------------------------
-.. autoclass:: dgllife.model.model_zoo.jtnn.DGLJTNNVAE
+Protein-Ligand Binding Affinity Prediction
+
+ACNN
+````
+.. automodule:: dgllife.model.model_zoo.acnn
     :members:
\ No newline at end of file

apps/life_sci/docs/source/api/utils.complexes.rst

+.. _apiutilscomplexes:
+
+Utils for protein-ligand complexes
+==================================
+
+Utilities in DGL-LifeSci for working with protein-ligand complexes.
+
+.. autosummary::
+    :toctree: ../generated/
+
+    dgllife.utils.ACNN_graph_construction_and_featurization

apps/life_sci/docs/source/api/utils.mols.rst

+.. _apiutilsmols:
+
+Utils for Molecules
+===================
+
+Utilities in DGL-LifeSci for working with molecules.
+
+RDKit Utils
+-----------
+
+RDKit utils for loading molecules and accessing their information.
+
+.. autosummary::
+    :toctree: ../generated/
+
+    dgllife.utils.get_mol_3d_coordinates
+    dgllife.utils.load_molecule
+    dgllife.utils.multiprocess_load_molecules
+
+Graph Construction
+------------------
+
+The modeling of graph neural networks starts with constructing appropriate graph topologies. We provide
+three common graph constructions:
+
+* ``bigraph``: Bi-directed graphs corresponding exactly to molecular graphs
+* ``complete_graph``: Graphs with all pairs of atoms connected
+* ``nearest_neighbor_graph``: Graphs where each atom is connected to its closest (k) atoms based on molecule coordinates
+
+.. autosummary::
+    :toctree: ../generated/
+
+    dgllife.utils.mol_to_graph
+    dgllife.utils.smiles_to_bigraph
+    dgllife.utils.mol_to_bigraph
+    dgllife.utils.smiles_to_complete_graph
+    dgllife.utils.mol_to_complete_graph
+    dgllife.utils.k_nearest_neighbors
+    dgllife.utils.mol_to_nearest_neighbor_graph
+    dgllife.utils.smiles_to_nearest_neighbor_graph
+
+Featurization for Molecules
+---------------------------
+
+To apply graph neural networks, we need to prepare node and edge features for molecules. Intuitively,
+they can be developed based on various descriptors (features) of atoms/bonds/molecules. Particularly, we can
+work with numerical descriptors directly or use ``one_hot_encoding`` for categorical descriptors. When using
+multiple descriptors together, we can simply concatenate them with ``ConcatFeaturizer``.
+
+General Utils
+```````````
+
+.. autosummary::
+    :toctree: ../generated/
+
+    dgllife.utils.one_hot_encoding
+    dgllife.utils.ConcatFeaturizer
+
+Featurization for Nodes
+```````````````````````
+
+We consider the following atom descriptors:
+
+* type/atomic number
+* degree (excluding neighboring hydrogen atoms)
+* total degree (including neighboring hydrogen atoms)
+* explicit valence
+* implicit valence
+* hybridization
+* total number of neighboring hydrogen atoms
+* formal charge
+* number of radical electrons
+* aromatic atom
+* ring membership
+* chirality
+* mass
+
+We can employ their numerical values directly or with one-hot encoding.
+
+.. autosummary::
+    :toctree: ../generated/
+
+    dgllife.utils.atom_type_one_hot
+    dgllife.utils.atomic_number_one_hot
+    dgllife.utils.atomic_number
+    dgllife.utils.atom_degree_one_hot
+    dgllife.utils.atom_degree
+    dgllife.utils.atom_total_degree_one_hot
+    dgllife.utils.atom_total_degree
+    dgllife.utils.atom_explicit_valence_one_hot
+    dgllife.utils.atom_explicit_valence
+    dgllife.utils.atom_implicit_valence_one_hot
+    dgllife.utils.atom_implicit_valence
+    dgllife.utils.atom_hybridization_one_hot
+    dgllife.utils.atom_total_num_H_one_hot
+    dgllife.utils.atom_total_num_H
+    dgllife.utils.atom_formal_charge_one_hot
+    dgllife.utils.atom_formal_charge
+    dgllife.utils.atom_num_radical_electrons_one_hot
+    dgllife.utils.atom_num_radical_electrons
+    dgllife.utils.atom_is_aromatic_one_hot
+    dgllife.utils.atom_is_aromatic
+    dgllife.utils.atom_is_in_ring_one_hot
+    dgllife.utils.atom_is_in_ring
+    dgllife.utils.atom_chiral_tag_one_hot
+    dgllife.utils.atom_mass
+
+For using featurization methods like above in creating node features:
+
+.. autosummary::
+    :toctree: ../generated/
+
+    dgllife.utils.BaseAtomFeaturizer
+    dgllife.utils.BaseAtomFeaturizer.feat_size
+    dgllife.utils.CanonicalAtomFeaturizer
+    dgllife.utils.CanonicalAtomFeaturizer.feat_size
+
+Featurization for Edges
+```````````````````````
+
+We consider the following bond descriptors:
+
+* type
+* conjugated bond
+* ring membership
+* stereo configuration
+
+.. autosummary::
+    :toctree: ../generated/
+
+    dgllife.utils.bond_type_one_hot
+    dgllife.utils.bond_is_conjugated_one_hot
+    dgllife.utils.bond_is_conjugated
+    dgllife.utils.bond_is_in_ring_one_hot
+    dgllife.utils.bond_is_in_ring
+    dgllife.utils.bond_stereo_one_hot
+
+For using featurization methods like above in creating edge features:
+
+.. autosummary::
+    :toctree: ../generated/
+
+    dgllife.utils.BaseBondFeaturizer
+    dgllife.utils.BaseBondFeaturizer.feat_size
+    dgllife.utils.CanonicalBondFeaturizer
+    dgllife.utils.CanonicalBondFeaturizer.feat_size

apps/life_sci/docs/source/api/utils.pipeline.rst

+.. _apiutilspipeline:
+
+Model Development Pipeline
+==========================
+
+.. contents:: Contents
+    :local:
+
+Model Evaluation
+----------------
+
+A utility class for evaluating model performance on (multi-label) supervised learning.
+
+.. autoclass:: dgllife.utils.Meter
+    :members: update, compute_metric
+
+Early Stopping
+--------------
+
+Early stopping is a standard practice for preventing models from overfitting and we provide a utility
+class for handling it.
+
+.. autoclass:: dgllife.utils.EarlyStopping
+    :members:

apps/life_sci/docs/source/api/utils.rst

-.. _apiutils:
-
-dgllife.utils
-==================
-
-TBD by Murphy
-
-dgllife.utils.complex_to_graph
--------------------------------------------
-.. automodule:: dgllife.utils.complex_to_graph
-    :members:
-
-
-dgllife.utils.early_stop
--------------------------------------------
-.. automodule:: dgllife.utils.early_stop
-    :members:
-
-
-dgllife.utils.eval
--------------------------------------------
-.. automodule:: dgllife.utils.eval
-    :members:
-
-
-dgllife.utils.featurizers
--------------------------------------------
-.. automodule:: dgllife.utils.featurizers
-    :members:
-
-
-dgllife.utils.mol_to_graph
--------------------------------------------
-.. automodule:: dgllife.utils.mol_to_graph
-    :members:
-
-
-dgllife.utils.rdkit_utils
--------------------------------------------
-.. automodule:: dgllife.utils.rdkit_utils
-    :members:
-
-
-dgllife.utils.splitters
--------------------------------------------
-.. automodule:: dgllife.utils.splitters
-    :members:

apps/life_sci/docs/source/api/utils.splitters.rst

+.. _apiutilssplitters:
+
+Splitting Datasets
+==================
+
+We provide multiple splitting methods for datasets.
+
+.. contents:: Contents
+    :local:
+
+ConsecutiveSplitter
+-------------------
+
+.. autoclass:: dgllife.utils.ConsecutiveSplitter
+    :members: train_val_test_split, k_fold_split
+
+RandomSplitter
+--------------
+
+.. autoclass:: dgllife.utils.RandomSplitter
+    :members: train_val_test_split, k_fold_split
+
+MolecularWeightSplitter
+-----------------------
+
+.. autoclass:: dgllife.utils.MolecularWeightSplitter
+    :members: train_val_test_split, k_fold_split
+
+ScaffoldSplitter
+----------------
+
+.. autoclass:: dgllife.utils.ScaffoldSplitter
+    :members: train_val_test_split, k_fold_split
+
+SingleTaskStratifiedSplitter
+----------------------------
+
+.. autoclass:: dgllife.utils.SingleTaskStratifiedSplitter
+    :members: train_val_test_split, k_fold_split

apps/life_sci/docs/source/conf.py

@@ -157,7 +157,7 @@ man_pages = [
 #  dir menu entry, description, category)
 texinfo_documents = [
     (master_doc, 'dgllife', 'DGL-LifeSci Documentation',
-     author, 'dgllife', 'Application library for XXXXXXXXXXXXXXXX.',
+     author, 'dgllife', 'Application library for life science.',
      'Miscellaneous'),
 ]
 

apps/life_sci/docs/source/get_started.rst

-Get Started
-===========

apps/life_sci/docs/source/index.rst

-DGL-LifeSci: A GNN Package for Chemistry and Molecular Applications
+DGL-LifeSci: Bringing Graph Neural Networks to Chemistry and Biology
 ===========================================================================================
 
-Blahlah ...
+DGL-LifeSci is a python package for applying graph neural networks to various tasks in chemistry
+and biology, on top of PyTorch and DGL. It provides:
 
-Get Started
-------------
+* Various utilities for data processing, training and evaluation.
+* Efficient and flexible model implementations.
+* Pre-trained models for use without training from scratch.
 
-You could borrow some from the README page.
+We cover various applications in our
+`examples <https://github.com/dmlc/dgl/tree/master/apps/life_sci/examples>`_, including:
 
-API Reference
----------------
+* `Molecular property prediction <https://github.com/dmlc/dgl/tree/master/apps/life_sci/examples/property_prediction>`_
+* `Generative models <https://github.com/dmlc/dgl/tree/master/apps/life_sci/examples/generative_models>`_
+* `Protein-ligand binding affinity prediction <https://github.com/dmlc/dgl/tree/master/apps/life_sci/examples/binding_affinity_prediction>`_
+* `Reaction prediction <https://github.com/dmlc/dgl/tree/master/apps/life_sci/examples/reaction_prediction>`_
 
-The highest level breakdown. What are the APIs for?
+Get Started
+------------
 
+Follow the :doc:`instructions<install/index>` to install DGL.
 
 .. toctree::
    :maxdepth: 1
-    :caption: Get Started
+   :caption: Installation
    :hidden:
    :glob:
 
-    get_started
+   install/index
 
 .. toctree::
     :maxdepth: 2
@@ -28,12 +35,20 @@ The highest level breakdown. What are the APIs for?
     :hidden:
     :glob:
 
+    api/utils.mols
+    api/utils.splitters
+    api/utils.pipeline
+    api/utils.complexes
     api/data
-    api/model
+    api/model.pretrain
     api/model.gnn
-    api/model.zoo
     api/model.readout
-    api/utils
+    api/model.zoo
+
+Free software
+-------------
+DGL-LifeSci is free software; you can redistribute it and/or modify it under the terms
+of the Apache License 2.0. We welcome contributions. Join us on `GitHub <https://github.com/dmlc/dgl/tree/master/apps/life_sci>`_.
 
 Index
 -----

apps/life_sci/docs/source/install/index.rst

+Install DGL-LifeSci
+===================
+
+This topic explains how to install DGL-LifeSci. We recommend installing DGL-LifeSci by using ``conda`` or ``pip``.
+
+System requirements
+-------------------
+DGL-LifeSci works with the following operating systems:
+
+* Ubuntu 16.04
+* macOS X
+* Windows 10
+
+DGL-LifeSci requires:
+
+* Python 3.6 or later
+* `DGL 0.4.3 or later <https://www.dgl.ai/pages/start.html>`_
+* `PyTorch 1.2.0 or later <https://pytorch.org/>`_
+
+If you have just installed DGL, the first time you use it, a message will pop up as follows:
+
+.. code:: bash
+
+    DGL does not detect a valid backend option. Which backend would you like to work with?
+    Backend choice (pytorch, mxnet or tensorflow):
+
+and you need to enter ``pytorch``.
+
+Additionally, we require **RDKit 2018.09.3** for cheminformatics. We recommend installing it with
+
+.. code:: bash
+
+    conda install -c conda-forge rdkit==2018.09.3
+
+Other verions of RDKit are not tested.
+
+Install from conda
+----------------------
+If ``conda`` is not yet installed, get either `miniconda <https://conda.io/miniconda.html>`_ or
+the full `anaconda <https://www.anaconda.com/download/>`_.
+
+.. code:: bash
+
+    conda install -c dglteam dgllife
+
+Install from pip
+----------------
+
+.. code:: bash
+
+    pip install dgllife
+
+.. _install-from-source:
+
+Install from source
+-------------------
+
+To use the latest experimental features,
+
+.. code:: bash
+
+    git clone https://github.com/dmlc/dgl.git
+    cd apps/life_sci/python
+    python setup.py install

apps/life_sci/python/dgllife/data/alchemy.py

@@ -259,30 +259,32 @@ class TencentAlchemyDataset(object):
             SMILES for the ith datapoint
         DGLGraph
             DGLGraph for the ith datapoint
-        Tensor of dtype float32
-            Labels of the datapoint for all tasks
+        Tensor of dtype float32 and shape (T)
+            Labels of the datapoint for all tasks.
         """
         return self.smiles[item], self.graphs[item], self.labels[item]
 
     def __len__(self):
-        """Length of the dataset
+        """Size for the dataset.
 
         Returns
         -------
         int
-            Length of Dataset
+            Size for the dataset.
         """
         return len(self.graphs)
 
     def set_mean_and_std(self, mean=None, std=None):
         """Set mean and std or compute from labels for future normalization.
 
+        The mean and std can be fetched later with ``self.mean`` and ``self.std``.
+
         Parameters
         ----------
-        mean : int or float
-            Default to be None.
-        std : int or float
-            Default to be None.
+        mean : float32 tensor of shape (T)
+            Mean of labels for all tasks.
+        std : float32 tensor of shape (T)
+            Std of labels for all tasks.
         """
         labels = np.array([i.numpy() for i in self.labels])
         if mean is None:

apps/life_sci/python/dgllife/data/csv_dataset.py

@@ -10,10 +10,9 @@ __all__ = ['MoleculeCSVDataset']
 class MoleculeCSVDataset(object):
     """MoleculeCSVDataset
 
-    This is a general class for loading molecular data from pandas.DataFrame.
+    This is a general class for loading molecular data from :class:`pandas.DataFrame`.
 
-    In data pre-processing, we set non-existing labels to be 0,
-    and returning mask with 1 where label exists.
+    In data pre-processing, we construct a binary mask indicating the existence of labels.
 
     All molecules are converted into DGLGraphs. After the first-time construction, the
     DGLGraphs can be saved for reloading so that we do not need to reconstruct them every time.
@@ -22,8 +21,7 @@ class MoleculeCSVDataset(object):
     ----------
     df: pandas.DataFrame
         Dataframe including smiles and labels. Can be loaded by pandas.read_csv(file_path).
-        One column includes smiles and other columns for labels.
-        Column names other than smiles column would be considered as task names.
+        One column includes smiles and some other columns include labels.
     smiles_to_graph: callable, str -> DGLGraph
         A function turning a SMILES into a DGLGraph.
     node_featurizer : callable, rdkit.Chem.rdchem.Mol -> dict
@@ -33,7 +31,7 @@ class MoleculeCSVDataset(object):
         Featurization for edges like bonds in a molecule, which can be used to update
         edata for a DGLGraph.
     smiles_column: str
-        Column name that including smiles.
+        Column name that including smiles in ``df``.
     cache_file_path: str
         Path to store the preprocessed DGLGraphs. For example, this can be ``'dglgraph.bin'``.
     task_names : list of str or None
@@ -118,19 +116,19 @@ class MoleculeCSVDataset(object):
             SMILES for the ith datapoint
         DGLGraph
             DGLGraph for the ith datapoint
-        Tensor of dtype float32
+        Tensor of dtype float32 and shape (T)
             Labels of the datapoint for all tasks
-        Tensor of dtype float32
+        Tensor of dtype float32 and shape (T)
             Binary masks indicating the existence of labels for all tasks
         """
         return self.smiles[item], self.graphs[item], self.labels[item], self.mask[item]
 
     def __len__(self):
-        """Length of the dataset
+        """Size for the dataset
 
         Returns
         -------
         int
-            Length of Dataset
+            Size for the dataset
         """
         return len(self.smiles)

apps/life_sci/python/dgllife/data/pdbbind.py

@@ -57,9 +57,10 @@ class PDBBind(object):
         Whether we need to extract molecular conformation from proteins and ligands.
         Default to True.
     construct_graph_and_featurize : callable
-        Construct a DGLHeteroGraph for the use of GNNs. Mapping self.ligand_mols[i],
-        self.protein_mols[i], self.ligand_coordinates[i] and self.protein_coordinates[i]
-        to a DGLHeteroGraph. Default to :func:`ACNN_graph_construction_and_featurization`.
+        Construct a DGLHeteroGraph for the use of GNNs. Mapping ``self.ligand_mols[i]``,
+        ``self.protein_mols[i]``, ``self.ligand_coordinates[i]`` and
+        ``self.protein_coordinates[i]`` to a DGLHeteroGraph.
+        Default to :func:`dgllife.utils.ACNN_graph_construction_and_featurization`.
     zero_padding : bool
         Whether to perform zero padding. While DGL does not necessarily require zero padding,
         pooling operations for variable length inputs can introduce stochastic behaviour, which

apps/life_sci/python/dgllife/data/pubchem_aromaticity.py

@@ -12,7 +12,7 @@ class PubChemBioAssayAromaticity(MoleculeCSVDataset):
     """Subset of PubChem BioAssay Dataset for aromaticity prediction.
 
     The dataset was constructed in `Pushing the Boundaries of Molecular Representation for Drug
-    Discovery with the Graph Attention Mechanism.
+    Discovery with the Graph Attention Mechanism
     <https://www.ncbi.nlm.nih.gov/pubmed/31408336>`__ and is accompanied by the task of predicting
     the number of aromatic atoms in molecules.