Commit a1323d95 authored by Bharath Ramsundar's avatar Bharath Ramsundar Committed by GitHub
Browse files

Merge pull request #709 from rbharath/ipython_notebooks 

Fixing Protein Ligand Notebook
parents 5ec720a7 b0a8f372

deepchem/hyper/__init__.py

+16 −17
Original line number Diff line number Diff line
@@ -12,6 +12,7 @@ from operator import mul 
from deepchem.utils.evaluate import Evaluator

from deepchem.utils.save import log





class HyperparamOpt(object):

  """

  Provides simple hyperparameter search capabilities.
@@ -23,8 +24,13 @@ class HyperparamOpt(object): 


  # TODO(rbharath): This function is complicated and monolithic. Is there a nice

  # way to refactor this?

  def hyperparam_search(self, params_dict, train_dataset, valid_dataset,

                        output_transformers, metric, use_max=True,

  def hyperparam_search(self,

                        params_dict,

                        train_dataset,

                        valid_dataset,

                        output_transformers,

                        metric,

                        use_max=True,

                        logdir=None):

    """Perform hyperparams search according to params_dict.
@@ -40,8 +46,6 @@ class HyperparamOpt(object): 


    number_combinations = reduce(mul, [len(vals) for vals in hyperparam_vals])



    valid_csv_out = tempfile.NamedTemporaryFile()

    valid_stats_out = tempfile.NamedTemporaryFile()

    if use_max:

      best_validation_score = -np.inf

    else:
@@ -49,10 +53,10 @@ class HyperparamOpt(object): 
    best_hyperparams = None

    best_model, best_model_dir = None, None

    all_scores = {}

    for ind, hyperparameter_tuple in enumerate(itertools.product(*hyperparam_vals)):

    for ind, hyperparameter_tuple in enumerate(

        itertools.product(*hyperparam_vals)):

      model_params = {}

      log("Fitting model %d/%d" % (ind+1, number_combinations),

          self.verbose)

      log("Fitting model %d/%d" % (ind + 1, number_combinations), self.verbose)

      for hyperparam, hyperparam_val in zip(hyperparams, hyperparameter_tuple):

        model_params[hyperparam] = hyperparam_val

      log("hyperparameters: %s" % str(model_params), self.verbose)
@@ -75,8 +79,7 @@ class HyperparamOpt(object): 
      model.save()



      evaluator = Evaluator(model, valid_dataset, output_transformers)

      multitask_scores = evaluator.compute_model_performance(

          [metric], valid_csv_out.name, valid_stats_out.name)

      multitask_scores = evaluator.compute_model_performance([metric])

      valid_score = multitask_scores[metric.name]

      all_scores[str(hyperparameter_tuple)] = valid_score
@@ -92,8 +95,8 @@ class HyperparamOpt(object): 
        shutil.rmtree(model_dir)



      log("Model %d/%d, Metric %s, Validation set %s: %f" %

          (ind+1, number_combinations, metric.name, ind, valid_score),

          self.verbose)

          (ind + 1, number_combinations, metric.name, ind,

           valid_score), self.verbose)

      log("\tbest_validation_score so far: %f" % best_validation_score,

          self.verbose)

    if best_model is None:
@@ -101,14 +104,10 @@ class HyperparamOpt(object): 
      # arbitrarily return last model

      best_model, best_hyperparams = model, hyperparameter_tuple

      return best_model, best_hyperparams, all_scores

    train_csv_out = tempfile.NamedTemporaryFile()

    train_stats_out = tempfile.NamedTemporaryFile()

    train_evaluator = Evaluator(best_model, train_dataset, output_transformers)

    multitask_scores = train_evaluator.compute_model_performance(

        [metric], train_csv_out.name, train_stats_out.name)

    multitask_scores = train_evaluator.compute_model_performance([metric])

    train_score = multitask_scores[metric.name]

    log("Best hyperparameters: %s" % str(best_hyperparams),

        self.verbose)

    log("Best hyperparameters: %s" % str(best_hyperparams), self.verbose)

    log("train_score: %f" % train_score, self.verbose)

    log("validation_score: %f" % best_validation_score, self.verbose)

    return best_model, best_hyperparams, all_scores

examples/notebooks/Multitask Networks on MUV.ipynb

+12 −205
Original line number Diff line number Diff line 
%% Cell type:markdown id: tags:



This notebook walks through the creation of multitask models on MUV. The goal is to demonstrate that multitask methods outperform singletask methods on MUV.



%% Cell type:code id: tags:



``` python

%reload_ext autoreload

%autoreload 2

%pdb off

reload = True

```



%% Output



    Automatic pdb calling has been turned OFF



%% Cell type:code id: tags:



``` python

import deepchem as dc



dataset_file= "../../datasets/muv.csv.gz"

dataset = dc.utils.save.load_from_disk(dataset_file)

print("Columns of dataset: %s" % str(dataset.columns.values))

print("Number of examples in dataset: %s" % str(dataset.shape[0]))

```



%% Output



    Columns of dataset: ['MUV-466' 'MUV-548' 'MUV-600' 'MUV-644' 'MUV-652' 'MUV-689' 'MUV-692'

     'MUV-712' 'MUV-713' 'MUV-733' 'MUV-737' 'MUV-810' 'MUV-832' 'MUV-846'

     'MUV-852' 'MUV-858' 'MUV-859' 'mol_id' 'smiles']

    Number of examples in dataset: 93127



%% Cell type:markdown id: tags:



Now, let's visualize some compounds from our dataset



%% Cell type:code id: tags:



``` python

from rdkit import Chem

from rdkit.Chem import Draw

from itertools import islice

from IPython.display import Image, display, HTML



def display_images(filenames):

    """Helper to pretty-print images."""

    imagesList=''.join(

        ["<img style='width: 140px; margin: 0px; float: left; border: 1px solid black;' src='%s' />"

         % str(s) for s in sorted(filenames)])

    display(HTML(imagesList))



def mols_to_pngs(mols, basename="test"):

    """Helper to write RDKit mols to png files."""

    filenames = []

    for i, mol in enumerate(mols):

        filename = "%s%d.png" % (basename, i)

        Draw.MolToFile(mol, filename)

        filenames.append(filename)

    return filenames



num_to_display = 12

molecules = []

for _, data in islice(dataset.iterrows(), num_to_display):

    molecules.append(Chem.MolFromSmiles(data["smiles"]))

display_images(mols_to_pngs(molecules))

```



%% Output





%% Cell type:code id: tags:



``` python

MUV_tasks = ['MUV-692', 'MUV-689', 'MUV-846', 'MUV-859', 'MUV-644',

             'MUV-548', 'MUV-852', 'MUV-600', 'MUV-810', 'MUV-712',

             'MUV-737', 'MUV-858', 'MUV-713', 'MUV-733', 'MUV-652',

             'MUV-466', 'MUV-832']



featurizer = dc.feat.CircularFingerprint(size=1024)

loader = dc.data.CSVLoader(

      tasks=MUV_tasks, smiles_field="smiles",

      featurizer=featurizer)

dataset = loader.featurize(dataset_file)

```



%% Output



    Loading raw samples now.

    shard_size: 8192

    About to start loading CSV from ../../datasets/muv.csv.gz

    Loading shard 1 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 0 took 22.680 s

    Loading shard 2 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 1 took 31.196 s

    Loading shard 3 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 2 took 29.828 s

    Loading shard 4 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 3 took 16.055 s

    Loading shard 5 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 4 took 21.150 s

    Loading shard 6 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 5 took 16.878 s

    Loading shard 7 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 6 took 16.032 s

    Loading shard 8 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 7 took 16.518 s

    Loading shard 9 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 8 took 17.401 s

    Loading shard 10 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 9 took 21.326 s

    Loading shard 11 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    Featurizing sample 4000

    Featurizing sample 5000

    Featurizing sample 6000

    Featurizing sample 7000

    Featurizing sample 8000

    TIMING: featurizing shard 10 took 23.441 s

    Loading shard 12 of size 8192.

    Featurizing sample 0

    Featurizing sample 1000

    Featurizing sample 2000

    Featurizing sample 3000

    TIMING: featurizing shard 11 took 8.318 s

    TIMING: dataset construction took 246.032 s

    Loading dataset from disk.



%% Cell type:code id: tags:



``` python

splitter = dc.splits.RandomSplitter(dataset_file)

train_dataset, valid_dataset, test_dataset = splitter.train_valid_test_split(

    dataset)

#NOTE THE RENAMING:

valid_dataset, test_dataset = test_dataset, valid_dataset

```



%% Output



    Computing train/valid/test indices

    TIMING: dataset construction took 6.272 s

    Loading dataset from disk.

    TIMING: dataset construction took 3.243 s

    Loading dataset from disk.

    TIMING: dataset construction took 3.498 s

    Loading dataset from disk.



%% Cell type:code id: tags:



``` python

import numpy as np

import numpy.random



params_dict = {"activation": ["relu"],

               "momentum": [.9],

               "batch_size": [50],

               "init": ["glorot_uniform"],

               "data_shape": [train_dataset.get_data_shape()],

               "learning_rate": [1e-3],

               "decay": [1e-6],

               "nb_epoch": [1],

               "nesterov": [False],

               "dropouts": [(.5,)],

               "nb_layers": [1],

               "batchnorm": [False],

               "layer_sizes": [(1000,)],

               "weight_init_stddevs": [(.1,)],

               "bias_init_consts": [(1.,)],

               "penalty": [0.],

              }





n_features = train_dataset.get_data_shape()[0]

def model_builder(model_params, model_dir):

  model = dc.models.TensorflowMultiTaskClassifier(

    len(MUV_tasks), n_features, **model_params)

  return model



metric = dc.metrics.Metric(dc.metrics.roc_auc_score, np.mean)

optimizer = dc.hyper.HyperparamOpt(model_builder)

best_dnn, best_hyperparams, all_results = optimizer.hyperparam_search(

    params_dict, train_dataset, valid_dataset, [], metric)

```
 


%% Output



    Fitting model 1/1

    hyperparameters: {'learning_rate': 0.001, 'layer_sizes': (1000,), 'data_shape': (1024,), 'dropouts': (0.5,), 'activation': 'relu', 'decay': 1e-06, 'batch_size': 50, 'penalty': 0.0, 'nesterov': False, 'init': 'glorot_uniform', 'bias_init_consts': (1.0,), 'weight_init_stddevs': (0.1,), 'batchnorm': False, 'nb_layers': 1, 'nb_epoch': 1, 'momentum': 0.9}

    Training for 1 epochs

    On batch 0

    On batch 50

    On batch 100

    On batch 150

    On batch 200

    On batch 250

    On batch 300

    On batch 350

    On batch 400

    On batch 450

    On batch 500

    On batch 550

    On batch 600

    On batch 650

    On batch 700

    On batch 750

    On batch 800

    On batch 850

    On batch 900

    On batch 950

    On batch 1000

    On batch 1050

    On batch 1100

    On batch 1150

    On batch 1200

    On batch 1250

    On batch 1300

    On batch 1350

    On batch 1400

    On batch 1450

    Ending epoch 0: Average loss 0.0517954

    On batch 0

    TIMING: model fitting took 42.847 s True

    computed_metrics: [0.65548567435359884, 0.3938885157824043, 0.91541865214431595, 0.39478957915831664, 0.68465248721524885, 0.74978870858688307, 0.80529733424470273, 0.77885952712100137, 0.79980310400741261, 0.88896680691912111, 0.5185931899641576, 0.5150129017124091, 0.75240054869684503, 0.36638813096862211, 0.58238095238095244, 0.61817558299039777, 0.68414343983684567]

    Model 1/1, Metric mean-roc_auc_score, Validation set 0: 0.653179

    	best_validation_score so far: 0.653179

    computed_metrics: [0.9154553963735379, 0.93463278293773655, 0.98457975954502508, 0.93589327852250226, 0.97265932420872547, 0.9838091436190004, 0.97946596833011079, 0.92610806949042246, 0.88254565136882901, 0.94433074824070085, 0.93797633429015803, 0.94311234732601956, 0.93923263973993476, 0.94550404459919302, 0.90309495615889945, 0.87384089962146405, 0.99468966106036016]

    Best hyperparameters: (1e-06, (1024,), 1, 'relu', (1000,), 50, 0.0, False, 'glorot_uniform', (1.0,), (0.1,), 1, False, (0.5,), 0.001, 0.9)

    train_score: 0.940996

    validation_score: 0.653179

examples/notebooks/protein_ligand_complex_notebook.ipynb

+252 −631

File changed.

Preview size limit exceeded, changes collapsed.

CRA Git | Maintained and supported by SUSTech CRA and CCSE