eval_tox.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from dl_util import *\n",
    "from ml_util import *\n",
    "\n",
    "from eval_class_util import *\n",
    "from run_eval import split_fit_plot_predict"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Loading and Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('eval_datasets/toxicity.txt','\\t')\n",
    "tox = df[pd.notnull(df['nontoxic'])]\n",
    "tox_maccs_all  = loadNumpy('maccs_toxic_all')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "tox_notnull_indices = df[df['nontoxic'].notnull()].index.tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "tox_smiles= tox['Canonical_QSARr'].tolist()\n",
    "tox_nontoxic = tox['nontoxic'].tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "saveData(tox_smiles,'tox_smiles')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8981, 8994)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(tox_smiles), len(tox_maccs_all)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "tox_maccs = []\n",
    "for i in tox_notnull_indices: \n",
    "    tox_maccs += [tox_maccs_all[i]]\n",
    "    \n",
    "tox_maccs = np.array(tox_maccs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8981, 8981, 8981)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(tox_maccs), len(tox_nontoxic), len(tox_notnull_indices)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Converting SMILES to sequence data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(42, 940)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len_tox_smiles = []\n",
    "for SMILE in tox_smiles:\n",
    "    len_tox_smiles += [len(SMILE)]\n",
    "    \n",
    "max_len_tox = max(len_tox_smiles)\n",
    "\n",
    "tox_vocab_size = generate_onehot_encoding(tox_smiles,'vocab_size')\n",
    "tox_vocab_size, max_len_tox"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "tox_encodings = generate_onehot_encoding(tox_smiles)\n",
    "tox_sequences = sequence.pad_sequences(tox_encodings, maxlen=max_len_tox)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "tox_sequences = loadNumpy('tox_sequences')\n",
    "tox_maccs = loadNumpy('tox_maccs')\n",
    "tox_nontoxic = loadNumpy('tox_nontoxic')\n",
    "tox_vocab_size, max_len_tox = (42, 940)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([False, False, False, ..., False, False, False])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tox_nontoxic"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train and Test set preparation "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "X1_tox = tox_sequences\n",
    "X2_tox = tox_maccs \n",
    "Y_tox = tox_nontoxic\n",
    "# X1_train_tox, X1_test_tox, y_train_tox, y_test_tox = train_test_split(X1_tox, Y_tox, random_state=1024)\n",
    "# X2_train_tox, X2_test_tox, y_train_tox, y_test_tox = train_test_split(X2_tox, Y_tox, random_state=1024)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <font color=\"red\"> CNN only </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_3 (Embedding)      (None, 940, 32)           1344      \n",
      "_________________________________________________________________\n",
      "conv1d_5 (Conv1D)            (None, 940, 32)           3104      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_5 (MaxPooling1 (None, 470, 32)           0         \n",
      "_________________________________________________________________\n",
      "conv1d_6 (Conv1D)            (None, 470, 32)           3104      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_6 (MaxPooling1 (None, 235, 32)           0         \n",
      "_________________________________________________________________\n",
      "flatten_3 (Flatten)          (None, 7520)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 7521      \n",
      "=================================================================\n",
      "Total params: 15,073\n",
      "Trainable params: 15,073\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 6061 samples, validate on 674 samples\n",
      "Epoch 1/20\n",
      "6061/6061 [==============================] - 2s 402us/step - loss: 0.6609 - acc: 0.6040 - precision: 0.4087 - recall: 0.9947 - auc: 0.5422 - val_loss: 0.6474 - val_acc: 0.6602 - val_precision: 0.4242 - val_recall: 1.0000 - val_auc: 0.6148\n",
      "Epoch 2/20\n",
      "6061/6061 [==============================] - 2s 337us/step - loss: 0.6125 - acc: 0.6756 - precision: 0.4241 - recall: 1.0000 - auc: 0.6458 - val_loss: 0.6152 - val_acc: 0.6706 - val_precision: 0.4246 - val_recall: 1.0000 - val_auc: 0.6679\n",
      "Epoch 3/20\n",
      "6061/6061 [==============================] - 2s 332us/step - loss: 0.5941 - acc: 0.6926 - precision: 0.4243 - recall: 1.0000 - auc: 0.6818 - val_loss: 0.6005 - val_acc: 0.6795 - val_precision: 0.4247 - val_recall: 1.0000 - val_auc: 0.6927\n",
      "Epoch 4/20\n",
      "6061/6061 [==============================] - 2s 339us/step - loss: 0.5818 - acc: 0.6963 - precision: 0.4243 - recall: 1.0000 - auc: 0.7008 - val_loss: 0.5901 - val_acc: 0.6929 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.7081\n",
      "Epoch 5/20\n",
      "6061/6061 [==============================] - 2s 334us/step - loss: 0.5620 - acc: 0.7157 - precision: 0.4249 - recall: 1.0000 - auc: 0.7163 - val_loss: 0.5804 - val_acc: 0.7047 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.7221\n",
      "Epoch 6/20\n",
      "6061/6061 [==============================] - 2s 335us/step - loss: 0.5487 - acc: 0.7273 - precision: 0.4248 - recall: 1.0000 - auc: 0.7281 - val_loss: 0.5854 - val_acc: 0.6914 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.7336\n",
      "Epoch 7/20\n",
      "6061/6061 [==============================] - 2s 338us/step - loss: 0.5371 - acc: 0.7304 - precision: 0.4250 - recall: 1.0000 - auc: 0.7384 - val_loss: 0.5783 - val_acc: 0.7077 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.7429\n",
      "Epoch 8/20\n",
      "6061/6061 [==============================] - 2s 332us/step - loss: 0.5252 - acc: 0.7360 - precision: 0.4250 - recall: 1.0000 - auc: 0.7475 - val_loss: 0.5664 - val_acc: 0.7122 - val_precision: 0.4249 - val_recall: 1.0000 - val_auc: 0.7515\n",
      "Epoch 9/20\n",
      "6061/6061 [==============================] - 2s 339us/step - loss: 0.5168 - acc: 0.7428 - precision: 0.4248 - recall: 1.0000 - auc: 0.7557 - val_loss: 0.5663 - val_acc: 0.7062 - val_precision: 0.4249 - val_recall: 1.0000 - val_auc: 0.7590\n",
      "Epoch 10/20\n",
      "6061/6061 [==============================] - 2s 340us/step - loss: 0.5048 - acc: 0.7484 - precision: 0.4246 - recall: 1.0000 - auc: 0.7623 - val_loss: 0.5614 - val_acc: 0.7166 - val_precision: 0.4249 - val_recall: 1.0000 - val_auc: 0.7659\n",
      "Epoch 11/20\n",
      "6061/6061 [==============================] - 2s 336us/step - loss: 0.4998 - acc: 0.7555 - precision: 0.4250 - recall: 1.0000 - auc: 0.7690 - val_loss: 0.5716 - val_acc: 0.7226 - val_precision: 0.4249 - val_recall: 1.0000 - val_auc: 0.7719\n",
      "Epoch 12/20\n",
      "6061/6061 [==============================] - 2s 335us/step - loss: 0.4911 - acc: 0.7593 - precision: 0.4251 - recall: 1.0000 - auc: 0.7749 - val_loss: 0.5726 - val_acc: 0.7181 - val_precision: 0.4249 - val_recall: 1.0000 - val_auc: 0.7774\n",
      "2246/2246 [==============================] - 0s 134us/step\n",
      "('Test AUC:', 0.7773314716554813)\n",
      "('History saved successfully', 'history_tox_cnn_model_dropout_0_epochs_20_batch_32_lr_0.001.pkl')\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "split_fit_plot_predict(cnn_model,  tox_sequences, tox_maccs, tox_nontoxic, tox_vocab_size, max_len_tox, \"tox\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### MLP only "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_4 (Dense)              (None, 512)               86016     \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 64)                16448     \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 1)                 65        \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 1)                 2         \n",
      "=================================================================\n",
      "Total params: 233,859\n",
      "Trainable params: 233,859\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 6061 samples, validate on 674 samples\n",
      "Epoch 1/20\n",
      "6061/6061 [==============================] - 2s 332us/step - loss: 0.5818 - acc: 0.7009 - precision: 0.4103 - recall: 0.9947 - auc: 0.6812 - val_loss: 0.5652 - val_acc: 0.7092 - val_precision: 0.4242 - val_recall: 1.0000 - val_auc: 0.7528\n",
      "Epoch 2/20\n",
      "6061/6061 [==============================] - 2s 263us/step - loss: 0.5136 - acc: 0.7481 - precision: 0.4245 - recall: 1.0000 - auc: 0.7717 - val_loss: 0.5245 - val_acc: 0.7418 - val_precision: 0.4246 - val_recall: 1.0000 - val_auc: 0.7860\n",
      "Epoch 3/20\n",
      "6061/6061 [==============================] - 2s 265us/step - loss: 0.4781 - acc: 0.7741 - precision: 0.4259 - recall: 1.0000 - auc: 0.7966 - val_loss: 0.5231 - val_acc: 0.7315 - val_precision: 0.4247 - val_recall: 1.0000 - val_auc: 0.8067\n",
      "Epoch 4/20\n",
      "6061/6061 [==============================] - 2s 266us/step - loss: 0.4431 - acc: 0.7969 - precision: 0.4256 - recall: 1.0000 - auc: 0.8163 - val_loss: 0.5288 - val_acc: 0.7344 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.8229\n",
      "Epoch 5/20\n",
      "6061/6061 [==============================] - 2s 264us/step - loss: 0.4063 - acc: 0.8141 - precision: 0.4244 - recall: 1.0000 - auc: 0.8305 - val_loss: 0.5252 - val_acc: 0.7463 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.8373\n",
      "2246/2246 [==============================] - 0s 114us/step\n",
      "('Test AUC:', 0.836472649289791)\n",
      "('History saved successfully', 'history_tox_mlp_model_dropout_0_epochs_20_batch_32_lr_0.001.pkl')\n",
      "Stats saved in model/stats_tox_mlp_model_dropout_0_epochs_20_batch_32_lr_0.001\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "split_fit_plot_predict(mlp_model,  tox_sequences, tox_maccs, tox_nontoxic, tox_vocab_size, max_len_tox, \"tox\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### CNN-MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_4 (Embedding)      (None, 940, 32)           1344      \n",
      "_________________________________________________________________\n",
      "conv1d_7 (Conv1D)            (None, 940, 32)           3104      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_7 (MaxPooling1 (None, 470, 32)           0         \n",
      "_________________________________________________________________\n",
      "conv1d_8 (Conv1D)            (None, 470, 32)           3104      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_8 (MaxPooling1 (None, 235, 32)           0         \n",
      "_________________________________________________________________\n",
      "flatten_4 (Flatten)          (None, 7520)              0         \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 64)                481344    \n",
      "=================================================================\n",
      "Total params: 488,896\n",
      "Trainable params: 488,896\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_10 (Dense)             (None, 512)               86016     \n",
      "_________________________________________________________________\n",
      "dense_11 (Dense)             (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 64)                16448     \n",
      "=================================================================\n",
      "Total params: 233,792\n",
      "Trainable params: 233,792\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "merge_1 (Merge)              (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 1)                 65        \n",
      "=================================================================\n",
      "Total params: 731,009\n",
      "Trainable params: 731,009\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 6061 samples, validate on 674 samples\n",
      "Epoch 1/20\n",
      "6061/6061 [==============================] - 3s 520us/step - loss: 0.5646 - acc: 0.7109 - precision: 0.4197 - recall: 0.9947 - auc: 0.7162 - val_loss: 0.5284 - val_acc: 0.7374 - val_precision: 0.4242 - val_recall: 1.0000 - val_auc: 0.7703\n",
      "Epoch 2/20\n",
      "6061/6061 [==============================] - 3s 413us/step - loss: 0.4997 - acc: 0.7552 - precision: 0.4243 - recall: 1.0000 - auc: 0.7899 - val_loss: 0.5378 - val_acc: 0.7315 - val_precision: 0.4246 - val_recall: 1.0000 - val_auc: 0.8004\n",
      "Epoch 3/20\n",
      "6061/6061 [==============================] - 3s 421us/step - loss: 0.4559 - acc: 0.7855 - precision: 0.4240 - recall: 1.0000 - auc: 0.8140 - val_loss: 0.5144 - val_acc: 0.7374 - val_precision: 0.4247 - val_recall: 1.0000 - val_auc: 0.8212\n",
      "Epoch 4/20\n",
      "6061/6061 [==============================] - 3s 416us/step - loss: 0.4129 - acc: 0.8111 - precision: 0.4244 - recall: 1.0000 - auc: 0.8314 - val_loss: 0.5885 - val_acc: 0.7270 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.8383\n",
      "Epoch 5/20\n",
      "6061/6061 [==============================] - 3s 415us/step - loss: 0.3575 - acc: 0.8416 - precision: 0.4248 - recall: 1.0000 - auc: 0.8474 - val_loss: 0.5320 - val_acc: 0.7596 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.8545\n",
      "2246/2246 [==============================] - 0s 147us/step\n",
      "('Test AUC:', 0.8526414599359194)\n",
      "('History saved successfully', 'history_tox_merged_cnn_mlp_model_dropout_0_epochs_20_batch_32_lr_0.001.pkl')\n",
      "Stats saved in model/stats_tox_merged_cnn_mlp_model_dropout_0_epochs_20_batch_32_lr_0.001\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "split_fit_plot_predict(merged_cnn_mlp_model,  tox_sequences, tox_maccs, tox_nontoxic, tox_vocab_size, max_len_tox,\"tox\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /usr/local/lib/python2.7/dist-packages/tensorflow/python/util/deprecation.py:497: calling conv1d (from tensorflow.python.ops.nn_ops) with data_format=NHWC is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "`NHWC` for data_format is deprecated, use `NWC` instead\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "merge_1 (Merge)              (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 1)                 65        \n",
      "=================================================================\n",
      "Total params: 271,393\n",
      "Trainable params: 271,393\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 6061 samples, validate on 674 samples\n",
      "Epoch 1/20\n",
      "6061/6061 [==============================] - 110s 18ms/step - loss: 0.5615 - acc: 0.7129 - precision: 0.4276 - recall: 0.9947 - auc: 0.7370 - val_loss: 0.5376 - val_acc: 0.7374 - val_precision: 0.4242 - val_recall: 1.0000 - val_auc: 0.7732\n",
      "Epoch 2/20\n",
      "6061/6061 [==============================] - 109s 18ms/step - loss: 0.5011 - acc: 0.7547 - precision: 0.4245 - recall: 1.0000 - auc: 0.7924 - val_loss: 0.5116 - val_acc: 0.7567 - val_precision: 0.4246 - val_recall: 1.0000 - val_auc: 0.8014\n",
      "Epoch 3/20\n",
      "6061/6061 [==============================] - 109s 18ms/step - loss: 0.4548 - acc: 0.7830 - precision: 0.4243 - recall: 1.0000 - auc: 0.8137 - val_loss: 0.5022 - val_acc: 0.7656 - val_precision: 0.4247 - val_recall: 1.0000 - val_auc: 0.8224\n",
      "Epoch 4/20\n",
      "6061/6061 [==============================] - 109s 18ms/step - loss: 0.4137 - acc: 0.8098 - precision: 0.4244 - recall: 1.0000 - auc: 0.8325 - val_loss: 0.5039 - val_acc: 0.7596 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.8396\n",
      "Epoch 5/20\n",
      "6061/6061 [==============================] - 109s 18ms/step - loss: 0.3785 - acc: 0.8291 - precision: 0.4247 - recall: 1.0000 - auc: 0.8478 - val_loss: 0.5535 - val_acc: 0.7463 - val_precision: 0.4248 - val_recall: 1.0000 - val_auc: 0.8535\n",
      "2246/2246 [==============================] - 8s 3ms/step\n",
      "('Test AUC:', 0.8517247180480142)\n"
     ]
    },
    {
     "ename": "NameError",
     "evalue": "global name 'in_jupyter' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-10-e205e8b90dee>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0msplit_fit_plot_predict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmerged_cnn_rnn_mlp_model\u001b[0m\u001b[0;34m,\u001b[0m  \u001b[0mtox_sequences\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtox_maccs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtox_nontoxic\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtox_vocab_size\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmax_len_tox\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\"tox\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/home/apx748/CEP/eval_class_util.py\u001b[0m in \u001b[0;36msplit_fit_plot_predict\u001b[0;34m(model_arch, X1, X2, Y, vocab, max_len, prefix, dropout, optimizer, lr, epochs, batch_size)\u001b[0m\n\u001b[1;32m    327\u001b[0m         \u001b[0;32mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Test AUC:\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mauc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    328\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 329\u001b[0;31m     \u001b[0;32mif\u001b[0m \u001b[0min_jupyter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    330\u001b[0m         \u001b[0mplot_history\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhistory\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmetric\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    331\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: global name 'in_jupyter' is not defined"
     ]
    }
   ],
   "source": [
    "split_fit_plot_predict(merged_cnn_rnn_mlp_model,  tox_sequences, tox_maccs, tox_nontoxic, tox_vocab_size, max_len_tox,\"tox\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('tox_sequences', 'saved successfully in', 'data')\n",
      "('tox_maccs', 'saved successfully in', 'data')\n",
      "('tox_nontoxic', 'saved successfully in', 'data')\n"
     ]
    }
   ],
   "source": [
    "saveNumpy(tox_sequences, 'tox_sequences')\n",
    "saveNumpy(tox_maccs, 'tox_maccs' )\n",
    "saveNumpy(tox_nontoxic, 'tox_nontoxic')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}