PDL.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Practical Deep Learning for Genomic Prediction\n",
    "## A Keras based guide to implement deep learning\n",
    "\n",
    "### M Perez-Enciso & ML Zingaretti\n",
    "### miguel.perez@uab.es, laura.zingaretti@cragenomica.es\n",
    "\n",
    "### If you find this resource useful, please cite: \n",
    "### Perez-Enciso M, Zingaretti ML, 2019. \n",
    "### A Guide on Deep Learning for Genomic Prediction. \n",
    "### Submitted\n",
    "\n",
    "### install dependencies if needed (only once)\n",
    "#sudo python3 -m pip install tensorflow\n",
    "#sudo python3 -m pip install keras\n",
    "#sudo python3 -m pip install scikit-learn\n",
    "#sudo python3 -m pip install scipy\n",
    "#sudo python3 -m pip install matplotlib\n",
    "#sudo python3 -m pip install talos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# main modules needed\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "from sklearn import linear_model\n",
    "from sklearn.model_selection import train_test_split\n",
    "from matplotlib import pyplot as plt\n",
    "from scipy import stats\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.preprocessing import scale\n",
    "\n",
    "# keras items \n",
    "from keras import regularizers\n",
    "from keras.models import Sequential, load_model\n",
    "from keras.layers import Dense, Activation, Dropout\n",
    "from keras.layers import Flatten, Conv1D, MaxPooling1D, LSTM #CNNs\n",
    "from keras.activations import relu, elu, linear, softmax\n",
    "from keras.callbacks import EarlyStopping, Callback\n",
    "from keras.wrappers.scikit_learn import KerasRegressor\n",
    "from keras.optimizers import adam, Nadam, sgd\n",
    "from keras.losses import mean_squared_error, categorical_crossentropy, logcosh\n",
    "from keras.utils.np_utils import to_categorical\n",
    "\n",
    "# talos items (for hyperparameter search)\n",
    "import talos as ta\n",
    "import wrangle as wr\n",
    "from talos.metrics.keras_metrics import fmeasure_acc\n",
    "from talos.model.layers import hidden_layers\n",
    "from talos import live\n",
    "from talos.model import lr_normalizer, early_stopper, hidden_layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(479, 1279) (479,)\n",
      "(120, 1279) (120,)\n",
      "       min max mean sd\n",
      "Train: -2.41866172921982 3.27892080508434 -3.516032490523932e-06 1.0064524506262142\n",
      "Test: -2.22424623595239 2.5269045419802203 1.4034829691607817e-05 0.9779568655008886\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# DATA LOADING AND BASIC INSPECTION\n",
    "# We use wheat data from BGLR (https://cran.r-project.org/web/packages/BGLR/BGLR.pdf)\n",
    "'''\n",
    "Matrix Y contains the average grain yield, column 1: Grain yield for environment 1 and so on.\n",
    "Matrix X contains marker genotypes.\n",
    "'''\n",
    "\n",
    "# load the dataset as a pandas data frame\n",
    "X = pd.read_csv('DATA/wheat.X', header=None, sep='\\s+')\n",
    "Y = pd.read_csv('DATA/wheat.Y', header=None, sep='\\s+')\n",
    "\n",
    "# data partitioning into train and validation\n",
    "itrait=0 # first trait analyzed\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, Y[itrait], test_size=0.2)\n",
    "print(X_train.shape, y_train.shape)\n",
    "print(X_test.shape, y_test.shape)\n",
    "\n",
    "# print basic statistics: max, min, mean, sd\n",
    "print('       min max mean sd')\n",
    "print('Train:', y_train.min(), y_train.max(), y_train.mean(), np.sqrt(y_train.var()))\n",
    "print('Test:', y_test.min(), y_test.max(), y_test.mean(), np.sqrt(y_test.var()))\n",
    "\n",
    "# do basic histograms\n",
    "plt.title('Train / test data')\n",
    "plt.hist(y_train, label='Train')\n",
    "plt.hist(y_test, label='Test')\n",
    "plt.legend(loc='best')\n",
    "plt.show()\n",
    "\n",
    "# marker PCA, use whole X with diff color for train and test\n",
    "X = np.concatenate((X_train, X_test))\n",
    "pca = PCA(n_components=2)\n",
    "p = pca.fit(X).fit_transform(X)\n",
    "Ntrain=X_train.shape[0]\n",
    "plt.title('PCA decomposition')\n",
    "plt.scatter(p[0:Ntrain,0], p[0:Ntrain,1], label='Train')\n",
    "plt.scatter(p[Ntrain:,0], p[Ntrain:,1], label='Test', color='orange')\n",
    "plt.legend(loc='best')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# OPTIONAL: SNP preselection according to a simple GWAS\n",
    "pvals = []\n",
    "for i in range(X_train.shape[1]):\n",
    "    b, intercept, r_value, p_value, std_err = stats.linregress(X_train[i], y_train)\n",
    "    pvals.append(-np.log10(p_value))\n",
    "pvals = np.array(pvals)\n",
    "\n",
    "# plot GWAS\n",
    "plt.ylabel('-log10 P-value')\n",
    "plt.xlabel('SNP')\n",
    "plt.plot(pvals, marker='o')\n",
    "plt.show()\n",
    "\n",
    "# select N_best most associated SNPs\n",
    "#N_best = X_train.shape[1] #all SNPs\n",
    "N_best = 100\n",
    "snp_list = pvals.argsort()[-N_best:]\n",
    "\n",
    "# or select by min_P_value\n",
    "min_P_value = 2 # P = 0.01\n",
    "snp_list = np.nonzero(pvals>min_P_value)\n",
    "\n",
    "# finally slice X\n",
    "X_train = X_train[X_train.columns[snp_list]] \n",
    "X_test = X_test[X_test.columns[snp_list]] \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "MSE in prediction = Tensor(\"Mean:0\", shape=(), dtype=float64)\n",
      "\n",
      "Corr obs vs pred = 0.5386793603926693\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Standard penalized methods (lasso using scikit-learn)\n",
    "\n",
    "# alpha is the regularization parameter\n",
    "lasso = linear_model.Lasso(alpha=0.01)\n",
    "lasso.fit(X_train, y_train)\n",
    "y_hat = lasso.predict(X_test)\n",
    "\n",
    "# mean squared error\n",
    "mse = mean_squared_error(y_test, y_hat)\n",
    "print('\\nMSE in prediction =',mse)\n",
    "\n",
    "# correlation btw predicted and observed\n",
    "corr = np.corrcoef(y_test,y_hat)[0,1]\n",
    "print('\\nCorr obs vs pred =',corr)\n",
    "\n",
    "# plot observed vs. predicted targets\n",
    "plt.title('Lasso: Observed vs Predicted Y')\n",
    "plt.ylabel('Predicted')\n",
    "plt.xlabel('Observed')\n",
    "plt.scatter(y_test, y_hat, marker='o')\n",
    "plt.show()\n",
    "\n",
    "# Exercises\n",
    "# - Implement an internal crossvalidation to optimize alpha\n",
    "# - try different sizes of most associated SNPs\n",
    "# - implement ridge regression instead of lasso"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 64)                9920      \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 12,033\n",
      "Trainable params: 12,033\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/100\n",
      "479/479 [==============================] - 0s 284us/step - loss: 1.1291\n",
      "Epoch 2/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.9517\n",
      "Epoch 3/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.8980\n",
      "Epoch 4/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.8673\n",
      "Epoch 5/100\n",
      "479/479 [==============================] - 0s 74us/step - loss: 0.8204\n",
      "Epoch 6/100\n",
      "479/479 [==============================] - 0s 75us/step - loss: 0.8006\n",
      "Epoch 7/100\n",
      "479/479 [==============================] - 0s 94us/step - loss: 0.7967\n",
      "Epoch 8/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.7676\n",
      "Epoch 9/100\n",
      "479/479 [==============================] - 0s 40us/step - loss: 0.7522\n",
      "Epoch 10/100\n",
      "479/479 [==============================] - 0s 100us/step - loss: 0.7279\n",
      "Epoch 11/100\n",
      "479/479 [==============================] - 0s 62us/step - loss: 0.7459\n",
      "Epoch 12/100\n",
      "479/479 [==============================] - 0s 41us/step - loss: 0.7285\n",
      "Epoch 13/100\n",
      "479/479 [==============================] - 0s 45us/step - loss: 0.7054\n",
      "Epoch 14/100\n",
      "479/479 [==============================] - 0s 66us/step - loss: 0.6851\n",
      "Epoch 15/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.6966\n",
      "Epoch 16/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.7511\n",
      "Epoch 17/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.6935\n",
      "Epoch 18/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.6718\n",
      "Epoch 19/100\n",
      "479/479 [==============================] - 0s 76us/step - loss: 0.6877\n",
      "Epoch 20/100\n",
      "479/479 [==============================] - 0s 72us/step - loss: 0.6566\n",
      "Epoch 21/100\n",
      "479/479 [==============================] - 0s 72us/step - loss: 0.6994\n",
      "Epoch 22/100\n",
      "479/479 [==============================] - 0s 72us/step - loss: 0.6422\n",
      "Epoch 23/100\n",
      "479/479 [==============================] - 0s 72us/step - loss: 0.6357\n",
      "Epoch 24/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.6317\n",
      "Epoch 25/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.6719\n",
      "Epoch 26/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.6367\n",
      "Epoch 27/100\n",
      "479/479 [==============================] - 0s 69us/step - loss: 0.6111\n",
      "Epoch 28/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.6309\n",
      "Epoch 29/100\n",
      "479/479 [==============================] - 0s 88us/step - loss: 0.5843\n",
      "Epoch 30/100\n",
      "479/479 [==============================] - 0s 116us/step - loss: 0.5783\n",
      "Epoch 31/100\n",
      "479/479 [==============================] - 0s 34us/step - loss: 0.6240\n",
      "Epoch 32/100\n",
      "479/479 [==============================] - 0s 101us/step - loss: 0.6004\n",
      "Epoch 33/100\n",
      "479/479 [==============================] - 0s 62us/step - loss: 0.5834\n",
      "Epoch 34/100\n",
      "479/479 [==============================] - 0s 42us/step - loss: 0.5725\n",
      "Epoch 35/100\n",
      "479/479 [==============================] - 0s 56us/step - loss: 0.5681\n",
      "Epoch 36/100\n",
      "479/479 [==============================] - 0s 67us/step - loss: 0.5826\n",
      "Epoch 37/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.5484\n",
      "Epoch 38/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.5348\n",
      "Epoch 39/100\n",
      "479/479 [==============================] - 0s 72us/step - loss: 0.5406\n",
      "Epoch 40/100\n",
      "479/479 [==============================] - 0s 69us/step - loss: 0.5414\n",
      "Epoch 41/100\n",
      "479/479 [==============================] - 0s 69us/step - loss: 0.5860\n",
      "Epoch 42/100\n",
      "479/479 [==============================] - 0s 68us/step - loss: 0.5449\n",
      "Epoch 43/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.5596\n",
      "Epoch 44/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.5183\n",
      "Epoch 45/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.5378\n",
      "Epoch 46/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.5392\n",
      "Epoch 47/100\n",
      "479/479 [==============================] - 0s 74us/step - loss: 0.5297\n",
      "Epoch 48/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.5330\n",
      "Epoch 49/100\n",
      "479/479 [==============================] - 0s 69us/step - loss: 0.5046\n",
      "Epoch 50/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.5118\n",
      "Epoch 51/100\n",
      "479/479 [==============================] - 0s 69us/step - loss: 0.5042\n",
      "Epoch 52/100\n",
      "479/479 [==============================] - 0s 97us/step - loss: 0.5005\n",
      "Epoch 53/100\n",
      "479/479 [==============================] - 0s 69us/step - loss: 0.4627\n",
      "Epoch 54/100\n",
      "479/479 [==============================] - 0s 50us/step - loss: 0.4690\n",
      "Epoch 55/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.4874\n",
      "Epoch 56/100\n",
      "479/479 [==============================] - 0s 81us/step - loss: 0.4485\n",
      "Epoch 57/100\n",
      "479/479 [==============================] - 0s 35us/step - loss: 0.4491\n",
      "Epoch 58/100\n",
      "479/479 [==============================] - 0s 36us/step - loss: 0.4824\n",
      "Epoch 59/100\n",
      "479/479 [==============================] - 0s 54us/step - loss: 0.4364\n",
      "Epoch 60/100\n",
      "479/479 [==============================] - 0s 68us/step - loss: 0.5103\n",
      "Epoch 61/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.4170\n",
      "Epoch 62/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.5138\n",
      "Epoch 63/100\n",
      "479/479 [==============================] - 0s 74us/step - loss: 0.4379\n",
      "Epoch 64/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.4351\n",
      "Epoch 65/100\n",
      "479/479 [==============================] - 0s 75us/step - loss: 0.5690\n",
      "Epoch 66/100\n",
      "479/479 [==============================] - 0s 68us/step - loss: 0.4452\n",
      "Epoch 67/100\n",
      "479/479 [==============================] - 0s 72us/step - loss: 0.4547\n",
      "Epoch 68/100\n",
      "479/479 [==============================] - 0s 69us/step - loss: 0.5203\n",
      "Epoch 69/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.4277\n",
      "Epoch 70/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.4831\n",
      "Epoch 71/100\n",
      "479/479 [==============================] - 0s 68us/step - loss: 0.3705\n",
      "Epoch 72/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.4012\n",
      "Epoch 73/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.4112\n",
      "Epoch 74/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.4003\n",
      "Epoch 75/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.3503\n",
      "Epoch 76/100\n",
      "479/479 [==============================] - 0s 92us/step - loss: 0.4301\n",
      "Epoch 77/100\n",
      "479/479 [==============================] - 0s 79us/step - loss: 0.3678\n",
      "Epoch 78/100\n",
      "479/479 [==============================] - 0s 58us/step - loss: 0.4804\n",
      "Epoch 79/100\n",
      "479/479 [==============================] - 0s 94us/step - loss: 0.4139\n",
      "Epoch 80/100\n",
      "479/479 [==============================] - 0s 80us/step - loss: 0.3658\n",
      "Epoch 81/100\n",
      "479/479 [==============================] - 0s 41us/step - loss: 0.3375\n",
      "Epoch 82/100\n",
      "479/479 [==============================] - 0s 60us/step - loss: 0.3307\n",
      "Epoch 83/100\n",
      "479/479 [==============================] - 0s 75us/step - loss: 0.3651\n",
      "Epoch 84/100\n",
      "479/479 [==============================] - 0s 77us/step - loss: 0.3706\n",
      "Epoch 85/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.4457\n",
      "Epoch 86/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.3537\n",
      "Epoch 87/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.3819\n",
      "Epoch 88/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.4450\n",
      "Epoch 89/100\n",
      "479/479 [==============================] - 0s 75us/step - loss: 0.3452\n",
      "Epoch 90/100\n",
      "479/479 [==============================] - 0s 73us/step - loss: 0.3179\n",
      "Epoch 91/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.4349\n",
      "Epoch 92/100\n",
      "479/479 [==============================] - 0s 70us/step - loss: 0.2990\n",
      "Epoch 93/100\n",
      "479/479 [==============================] - 0s 72us/step - loss: 0.3086\n",
      "Epoch 94/100\n",
      "479/479 [==============================] - 0s 75us/step - loss: 0.3345\n",
      "Epoch 95/100\n",
      "479/479 [==============================] - 0s 74us/step - loss: 0.2877\n",
      "Epoch 96/100\n",
      "479/479 [==============================] - 0s 72us/step - loss: 0.2836\n",
      "Epoch 97/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.3687\n",
      "Epoch 98/100\n",
      "479/479 [==============================] - 0s 71us/step - loss: 0.3465\n",
      "Epoch 99/100\n",
      "479/479 [==============================] - 0s 82us/step - loss: 0.2905\n",
      "Epoch 100/100\n",
      "479/479 [==============================] - 0s 101us/step - loss: 0.3216\n",
      "120/120 [==============================] - 0s 557us/step\n",
      "\n",
      "MSE in prediction = 0.9173638224601746\n",
      "\n",
      "Corr obs vs pred = 0.5604678787782341\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Implements a standard fully connected network (MLP) for a quantitative target\n",
    "\n",
    "# no. of SNPs in data\n",
    "nSNP=X_train.shape[1] \n",
    "\n",
    "# Instantiate\n",
    "model = Sequential()\n",
    "\n",
    "# Add first layer\n",
    "model.add(Dense(64, input_dim=nSNP))\n",
    "model.add(Activation('relu'))\n",
    "# Add second layer\n",
    "model.add(Dense(32))\n",
    "model.add(Activation('softplus'))\n",
    "# Last, output layer\n",
    "model.add(Dense(1))\n",
    "\n",
    "# Model Compiling (https://keras.io/models/sequential/) \n",
    "# compile(optimizer, loss=None, metrics=None, loss_weights=None, sample_weight_mode=None, weighted_metrics=None, target_tensors=None)\n",
    "# Stochastic Gradient Descent (‘sgd’) as optimization algorithm\n",
    "# Mean Squared Error as loss, ie, quantitative variable, regression\n",
    "model.compile(loss='mean_squared_error', optimizer='sgd')\n",
    "\n",
    "# list some properties\n",
    "model.summary()\n",
    "\n",
    "# training\n",
    "# fit(x=None, y=None, batch_size=None, epochs=1, verbose=1, callbacks=None, validation_split=0.0, validation_data=None, shuffle=True, class_weight=None, sample_weight=None, initial_epoch=0, steps_per_epoch=None, validation_steps=None, validation_freq=1)\n",
    "model.fit(X_train, y_train, epochs=100)\n",
    "\n",
    "# cross-validation: get predicted target values\n",
    "y_hat = model.predict(X_test, batch_size=128)\n",
    "\n",
    "mse_prediction = model.evaluate(X_test, y_test, batch_size=128)\n",
    "print('\\nMSE in prediction =',mse_prediction)\n",
    "\n",
    "# correlation btw predicted and observed\n",
    "corr = np.corrcoef(y_test,y_hat[:,0])[0,1]\n",
    "print('\\nCorr obs vs pred =',corr)\n",
    "\n",
    "# plot observed vs. predicted targets\n",
    "plt.title('MLP: Observed vs Predicted Y')\n",
    "plt.ylabel('Predicted')\n",
    "plt.xlabel('Observed')\n",
    "plt.scatter(y_test, y_hat, marker='o')\n",
    "plt.show()\n",
    "\n",
    "# Exercises\n",
    "# - Check predictions across environments (Y[0] is first environment, etc)\n",
    "# - Try to improve model with other activation functions and|or no. of neurons∫ "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 431 samples, validate on 48 samples\n",
      "Epoch 1/100\n",
      "431/431 [==============================] - 0s 571us/step - loss: 11.0440 - val_loss: 7.6817\n",
      "Epoch 2/100\n",
      "431/431 [==============================] - 0s 94us/step - loss: 8.5581 - val_loss: 7.2182\n",
      "Epoch 3/100\n",
      "431/431 [==============================] - 0s 117us/step - loss: 7.8863 - val_loss: 6.8191\n",
      "Epoch 4/100\n",
      "431/431 [==============================] - 0s 40us/step - loss: 7.6269 - val_loss: 6.5095\n",
      "Epoch 5/100\n",
      "431/431 [==============================] - 0s 157us/step - loss: 7.1305 - val_loss: 6.3397\n",
      "Epoch 6/100\n",
      "431/431 [==============================] - 0s 45us/step - loss: 6.8776 - val_loss: 6.0632\n",
      "Epoch 7/100\n",
      "431/431 [==============================] - 0s 60us/step - loss: 6.7768 - val_loss: 5.8693\n",
      "Epoch 8/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 6.4442 - val_loss: 5.7010\n",
      "Epoch 9/100\n",
      "431/431 [==============================] - 0s 82us/step - loss: 6.2948 - val_loss: 5.5251\n",
      "Epoch 10/100\n",
      "431/431 [==============================] - 0s 90us/step - loss: 6.1346 - val_loss: 5.4480\n",
      "Epoch 11/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 5.8571 - val_loss: 5.4379\n",
      "Epoch 12/100\n",
      "431/431 [==============================] - 0s 73us/step - loss: 5.7276 - val_loss: 5.2250\n",
      "Epoch 13/100\n",
      "431/431 [==============================] - 0s 73us/step - loss: 5.7092 - val_loss: 5.0674\n",
      "Epoch 14/100\n",
      "431/431 [==============================] - 0s 71us/step - loss: 5.5282 - val_loss: 4.9650\n",
      "Epoch 15/100\n",
      "431/431 [==============================] - 0s 72us/step - loss: 5.3624 - val_loss: 4.8545\n",
      "Epoch 16/100\n",
      "431/431 [==============================] - 0s 72us/step - loss: 5.1924 - val_loss: 4.7781\n",
      "Epoch 17/100\n",
      "431/431 [==============================] - 0s 73us/step - loss: 5.0882 - val_loss: 4.7191\n",
      "Epoch 18/100\n",
      "431/431 [==============================] - 0s 73us/step - loss: 4.9639 - val_loss: 4.6269\n",
      "Epoch 19/100\n",
      "431/431 [==============================] - 0s 66us/step - loss: 4.9083 - val_loss: 4.5289\n",
      "Epoch 20/100\n",
      "431/431 [==============================] - 0s 73us/step - loss: 4.7830 - val_loss: 4.4576\n",
      "Epoch 21/100\n",
      "431/431 [==============================] - 0s 71us/step - loss: 4.7135 - val_loss: 4.3799\n",
      "Epoch 22/100\n",
      "431/431 [==============================] - 0s 74us/step - loss: 4.6520 - val_loss: 4.3036\n",
      "Epoch 23/100\n",
      "431/431 [==============================] - 0s 65us/step - loss: 4.5344 - val_loss: 4.2915\n",
      "Epoch 24/100\n",
      "431/431 [==============================] - 0s 69us/step - loss: 4.4435 - val_loss: 4.1834\n",
      "Epoch 25/100\n",
      "431/431 [==============================] - 0s 89us/step - loss: 4.3826 - val_loss: 4.1359\n",
      "Epoch 26/100\n",
      "431/431 [==============================] - 0s 92us/step - loss: 4.2479 - val_loss: 4.0967\n",
      "Epoch 27/100\n",
      "431/431 [==============================] - 0s 96us/step - loss: 4.2589 - val_loss: 4.0096\n",
      "Epoch 28/100\n",
      "431/431 [==============================] - 0s 89us/step - loss: 4.2020 - val_loss: 3.9871\n",
      "Epoch 29/100\n",
      "431/431 [==============================] - 0s 80us/step - loss: 4.0695 - val_loss: 3.9397\n",
      "Epoch 30/100\n",
      "431/431 [==============================] - 0s 76us/step - loss: 4.0043 - val_loss: 3.9270\n",
      "Epoch 31/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 3.9488 - val_loss: 3.8423\n",
      "Epoch 32/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 3.8676 - val_loss: 3.7508\n",
      "Epoch 33/100\n",
      "431/431 [==============================] - 0s 81us/step - loss: 3.8966 - val_loss: 3.6896\n",
      "Epoch 34/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 3.7652 - val_loss: 3.6933\n",
      "Epoch 35/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 3.7430 - val_loss: 3.6277\n",
      "Epoch 36/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 3.6713 - val_loss: 3.5917\n",
      "Epoch 37/100\n",
      "431/431 [==============================] - 0s 118us/step - loss: 3.6784 - val_loss: 3.5325\n",
      "Epoch 38/100\n",
      "431/431 [==============================] - 0s 121us/step - loss: 3.5909 - val_loss: 3.5037\n",
      "Epoch 39/100\n",
      "431/431 [==============================] - 0s 42us/step - loss: 3.5891 - val_loss: 3.5677\n",
      "Epoch 40/100\n",
      "431/431 [==============================] - 0s 125us/step - loss: 3.5360 - val_loss: 3.6414\n",
      "Epoch 41/100\n",
      "431/431 [==============================] - 0s 52us/step - loss: 3.5143 - val_loss: 3.4114\n",
      "Epoch 42/100\n",
      "431/431 [==============================] - 0s 45us/step - loss: 3.3898 - val_loss: 3.4284\n",
      "Epoch 43/100\n",
      "431/431 [==============================] - 0s 66us/step - loss: 3.3611 - val_loss: 3.2980\n",
      "Epoch 44/100\n",
      "431/431 [==============================] - 0s 85us/step - loss: 3.3673 - val_loss: 3.4129\n",
      "Epoch 45/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 3.3664 - val_loss: 3.2575\n",
      "Epoch 46/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 3.3121 - val_loss: 3.2620\n",
      "Epoch 47/100\n",
      "431/431 [==============================] - 0s 85us/step - loss: 3.2210 - val_loss: 3.1797\n",
      "Epoch 48/100\n",
      "431/431 [==============================] - 0s 88us/step - loss: 3.2683 - val_loss: 3.1511\n",
      "Epoch 49/100\n",
      "431/431 [==============================] - 0s 83us/step - loss: 3.2160 - val_loss: 3.1684\n",
      "Epoch 50/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 3.1497 - val_loss: 3.0795\n",
      "Epoch 51/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 3.1071 - val_loss: 3.0942\n",
      "Epoch 52/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 3.0970 - val_loss: 3.0540\n",
      "Epoch 53/100\n",
      "431/431 [==============================] - 0s 85us/step - loss: 3.1362 - val_loss: 3.0146\n",
      "Epoch 54/100\n",
      "431/431 [==============================] - 0s 85us/step - loss: 3.0630 - val_loss: 2.9608\n",
      "Epoch 55/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 3.0542 - val_loss: 2.9991\n",
      "Epoch 56/100\n",
      "431/431 [==============================] - 0s 88us/step - loss: 3.0351 - val_loss: 3.0401\n",
      "Epoch 57/100\n",
      "431/431 [==============================] - 0s 88us/step - loss: 3.0364 - val_loss: 2.9701\n",
      "Epoch 58/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 3.0590 - val_loss: 2.9980\n",
      "Epoch 59/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 3.0695 - val_loss: 2.8628\n",
      "Epoch 60/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 3.0483 - val_loss: 2.8448\n",
      "Epoch 61/100\n",
      "431/431 [==============================] - 0s 82us/step - loss: 2.9791 - val_loss: 2.8475\n",
      "Epoch 62/100\n",
      "431/431 [==============================] - 0s 83us/step - loss: 2.9838 - val_loss: 2.8617\n",
      "Epoch 63/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 3.0160 - val_loss: 2.8182\n",
      "Epoch 64/100\n",
      "431/431 [==============================] - 0s 82us/step - loss: 2.8975 - val_loss: 2.7476\n",
      "Epoch 65/100\n",
      "431/431 [==============================] - 0s 88us/step - loss: 2.9452 - val_loss: 2.8076\n",
      "Epoch 66/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 3.0048 - val_loss: 2.7508\n",
      "Epoch 67/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 2.9002 - val_loss: 2.7697\n",
      "Epoch 68/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 2.9718 - val_loss: 2.6585\n",
      "Epoch 69/100\n",
      "431/431 [==============================] - 0s 85us/step - loss: 2.9096 - val_loss: 2.6595\n",
      "Epoch 70/100\n",
      "431/431 [==============================] - 0s 89us/step - loss: 2.9658 - val_loss: 2.6597\n",
      "Epoch 71/100\n",
      "431/431 [==============================] - 0s 115us/step - loss: 2.9624 - val_loss: 2.6192\n",
      "Epoch 72/100\n",
      "431/431 [==============================] - 0s 122us/step - loss: 2.9054 - val_loss: 2.6157\n",
      "Epoch 73/100\n",
      "431/431 [==============================] - 0s 42us/step - loss: 2.9597 - val_loss: 2.5975\n",
      "Epoch 74/100\n",
      "431/431 [==============================] - 0s 121us/step - loss: 2.9343 - val_loss: 2.5937\n",
      "Epoch 75/100\n",
      "431/431 [==============================] - 0s 50us/step - loss: 2.8918 - val_loss: 2.5586\n",
      "Epoch 76/100\n",
      "431/431 [==============================] - 0s 47us/step - loss: 2.9047 - val_loss: 2.6015\n",
      "Epoch 77/100\n",
      "431/431 [==============================] - 0s 74us/step - loss: 2.8992 - val_loss: 2.5392\n",
      "Epoch 78/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 2.8734 - val_loss: 2.5298\n",
      "Epoch 79/100\n",
      "431/431 [==============================] - 0s 85us/step - loss: 2.8491 - val_loss: 2.4837\n",
      "Epoch 80/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "431/431 [==============================] - 0s 84us/step - loss: 2.8155 - val_loss: 2.5002\n",
      "Epoch 81/100\n",
      "431/431 [==============================] - 0s 85us/step - loss: 2.8824 - val_loss: 2.5182\n",
      "Epoch 82/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 2.8435 - val_loss: 2.4696\n",
      "Epoch 83/100\n",
      "431/431 [==============================] - 0s 85us/step - loss: 2.8467 - val_loss: 2.4470\n",
      "Epoch 84/100\n",
      "431/431 [==============================] - 0s 84us/step - loss: 2.8754 - val_loss: 2.4348\n",
      "Epoch 85/100\n",
      "431/431 [==============================] - 0s 81us/step - loss: 2.8254 - val_loss: 2.4632\n",
      "Epoch 86/100\n",
      "431/431 [==============================] - 0s 82us/step - loss: 2.9043 - val_loss: 2.5083\n",
      "Epoch 87/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 3.0144 - val_loss: 2.5575\n",
      "Epoch 88/100\n",
      "431/431 [==============================] - 0s 88us/step - loss: 3.0285 - val_loss: 2.6051\n",
      "Epoch 89/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 2.9687 - val_loss: 2.4088\n",
      "Epoch 90/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 2.8980 - val_loss: 2.7042\n",
      "Epoch 91/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 3.0094 - val_loss: 2.4989\n",
      "Epoch 92/100\n",
      "431/431 [==============================] - 0s 90us/step - loss: 2.8322 - val_loss: 2.4077\n",
      "Epoch 93/100\n",
      "431/431 [==============================] - 0s 92us/step - loss: 2.9071 - val_loss: 2.4650\n",
      "Epoch 94/100\n",
      "431/431 [==============================] - 0s 122us/step - loss: 2.8406 - val_loss: 2.3177\n",
      "Epoch 95/100\n",
      "431/431 [==============================] - 0s 40us/step - loss: 2.8359 - val_loss: 2.3583\n",
      "Epoch 96/100\n",
      "431/431 [==============================] - 0s 143us/step - loss: 2.8903 - val_loss: 2.5085\n",
      "Epoch 97/100\n",
      "431/431 [==============================] - 0s 46us/step - loss: 3.0092 - val_loss: 2.3883\n",
      "Epoch 98/100\n",
      "431/431 [==============================] - 0s 66us/step - loss: 2.9693 - val_loss: 2.5783\n",
      "Epoch 99/100\n",
      "431/431 [==============================] - 0s 87us/step - loss: 3.1067 - val_loss: 2.2820\n",
      "Epoch 100/100\n",
      "431/431 [==============================] - 0s 86us/step - loss: 2.7897 - val_loss: 2.3433\n",
      "120/120 [==============================] - 0s 16us/step\n",
      "\n",
      "MSE in prediction = 6.269443035125732\n"
     ]
    }
   ],
   "source": [
    "# Controlling overfit: regularization, dropout and early stopping\n",
    "\n",
    "# deletes current model\n",
    "del model\n",
    "\n",
    "model = Sequential()\n",
    "\n",
    "# Add l1 & l2 regularization in first layer\n",
    "model.add(Dense(64, input_dim=nSNP,\n",
    "                kernel_regularizer=regularizers.l2(0.01),\n",
    "                activity_regularizer=regularizers.l1(0.01)))\n",
    "model.add(Activation('relu'))\n",
    "# Add second layer\n",
    "model.add(Dense(32))\n",
    "model.add(Activation('softplus'))\n",
    "## Adding dropout to second layer\n",
    "model.add(Dropout(0.2))\n",
    "# Last, output layer\n",
    "model.add(Dense(1))\n",
    "\n",
    "# Model Compiling (https://keras.io/models/sequential/) \n",
    "model.compile(loss='mean_squared_error', optimizer='sgd')\n",
    "\n",
    "# Split the train set into proper train & validation\n",
    "X_train0, X_val, y_train0, y_val = train_test_split(X_train, y_train, test_size=0.1)\n",
    "nEpochs=100\n",
    "\n",
    "# Early stopping\n",
    "early_stopper = EarlyStopping(monitor='val_loss', patience=10, min_delta=0.01)\n",
    "model.fit(X_train0, y_train0, epochs=nEpochs, verbose=1, validation_data=(X_val, y_val), callbacks=[early_stopper])\n",
    "\n",
    "# cross-validation\n",
    "mse_prediction = model.evaluate(X_test, y_test, batch_size=128)\n",
    "print('\\nMSE in prediction =',mse_prediction)\n",
    "\n",
    "## In this case neither l1 nor l2 regularization helps"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Talos is a powerful tool to keras hyperparameter optimization and model evaluation. For regression models, it includes root_mean_squared_error as an evaluation metric, but you can also use a custom function as the train/test correlation. \n",
    "**Warning**: you should include the string 'acc' in the name in order to  pick epochs will be well recorded."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Defining pearson correlation as custom metric to model optimization in talos! \n",
    "# warning! you have to use acc in the metric name!\n",
    "\n",
    "from keras import backend as K\n",
    "\n",
    "def acc_pearson_r(y_true, y_pred):\n",
    "    x = y_true\n",
    "    y = y_pred\n",
    "    mx = K.mean(x, axis=0)\n",
    "    my = K.mean(y, axis=0)\n",
    "    xm, ym = x - mx, y - my\n",
    "    r_num = K.sum(xm * ym)\n",
    "    x_square_sum = K.sum(xm * xm)\n",
    "    y_square_sum = K.sum(ym * ym)\n",
    "    r_den = K.sqrt(x_square_sum * y_square_sum)\n",
    "    r = r_num / r_den\n",
    "    return K.mean(r)    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# HYPERPARAMETER OPTIMIZATION USING TALOS: Simple example\n",
    "# https://autonomio.github.io/docs_talos/\n",
    "# 1. Hyperparameter ranges and Model definition\n",
    "\n",
    "# model definition\n",
    "def baby_model(x, y, x_val, y_val, params):    \n",
    "    # replace the hyperparameter inputs with references to params dictionary \n",
    "    model = Sequential()\n",
    "    model.add(Dense(params['first_neuron'], input_dim=x.shape[1],\n",
    "                activation=params['activation']))\n",
    "    #last neuron\n",
    "    model.add(Dense(1, activation=None))\n",
    "    model.compile(loss=mean_squared_error, optimizer='sgd', metrics=[acc_pearson_r])\n",
    "    \n",
    "    # make sure history object is returned by model.fit()\n",
    "    out = model.fit(x, y,\n",
    "                    epochs=50,\n",
    "                    validation_data=[x_val, y_val],\n",
    "                    batch_size=params['batch_size'],\n",
    "                    verbose=0,callbacks=[live()])\n",
    "    \n",
    "    # modify the output model\n",
    "    return out, model\n",
    "\n",
    "# dictionary with hyperparameters and range values allowed\n",
    "p = {\n",
    "    'first_neuron': [12, 48],\n",
    "    'activation': [relu, elu],\n",
    "    'batch_size': [10, 30]\n",
    "}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 8/8 [02:17<00:00, 17.65s/it]\n"
     ]
    }
   ],
   "source": [
    "# HYPERPARAMETER OPTIMIZATION USING TALOS: Simple example\n",
    "# 2. Search, run this grid should take between 1-2 minutes\n",
    "\n",
    "# Split the train set into proper train & validation\n",
    "X_train0, X_val, y_train0, y_val = train_test_split(X_train, y_train, test_size=0.1)\n",
    "X_train0=np.asarray(X_train0)\n",
    "X_val=np.asarray(X_val)\n",
    "y_train0=np.asarray(y_train0)\n",
    "y_val=np.asarray(y_val)\n",
    "\n",
    "# COOL! this shows real time plots\n",
    "t_Init = ta.Scan(x=X_train0,\n",
    "                 y=y_train0,\n",
    "                 x_val=X_val,\n",
    "                 y_val=y_val,\n",
    "                 model=baby_model, \n",
    "                 params=p, \n",
    "                 grid_downsample=None,   # grid search\n",
    "                 dataset_name='baby_model',\n",
    "                 print_params=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  round_epochs            val_loss  val_acc_pearson_r      loss  \\\n",
      "0           50  0.9465713128447533           0.382998  0.582150   \n",
      "1           50  0.8263635188341141           0.529416  0.286149   \n",
      "2           50  0.8942086026072502           0.423034  0.535030   \n",
      "3           50  0.8275855059425036           0.512533  0.409994   \n",
      "4           50  0.7703300900757313           0.557218  0.533220   \n",
      "5           50  0.8903041034936905           0.438549  0.388465   \n",
      "6           50  0.8233412603537241           0.544002  0.210641   \n",
      "7           50  0.8689843788743019           0.474178  0.490671   \n",
      "\n",
      "        acc_pearson_r first_neuron                         activation  \\\n",
      "0  0.6531129757378882           12   <function elu at 0x7f04dbb64400>   \n",
      "1                 nan           12  <function relu at 0x7f04dbb64620>   \n",
      "2  0.7168528407189796           48   <function elu at 0x7f04dbb64400>   \n",
      "3                 nan           48   <function elu at 0x7f04dbb64400>   \n",
      "4  0.7013518201779323           12  <function relu at 0x7f04dbb64620>   \n",
      "5   0.798336085338327           48  <function relu at 0x7f04dbb64620>   \n",
      "6                 nan           48  <function relu at 0x7f04dbb64620>   \n",
      "7                 nan           12   <function elu at 0x7f04dbb64400>   \n",
      "\n",
      "  batch_size  \n",
      "0         30  \n",
      "1         10  \n",
      "2         30  \n",
      "3         10  \n",
      "4         30  \n",
      "5         30  \n",
      "6         10  \n",
      "7         10  \n",
      "\n",
      "Best prediction combination:\n",
      "  round_epochs            val_loss  val_acc_pearson_r     loss  \\\n",
      "4           50  0.7703300900757313           0.557218  0.53322   \n",
      "\n",
      "        acc_pearson_r first_neuron                         activation  \\\n",
      "4  0.7013518201779323           12  <function relu at 0x7f04dbb64620>   \n",
      "\n",
      "  batch_size  \n",
      "4         30   \n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# HYPERPARAMETER OPTIMIZATION USING TALOS: Simple example\n",
    "# 3. Inspect results\n",
    "Data=pd.DataFrame(t_Init.data)\n",
    "Data[\"loss\"] = pd.to_numeric(Data[\"loss\"])\n",
    "Data[\"val_acc_pearson_r\"] = pd.to_numeric(Data[\"val_acc_pearson_r\"])\n",
    "print(Data)\n",
    "\n",
    "# Minimum loss set\n",
    "i=Data['val_acc_pearson_r'].argmax()\n",
    "print('\\nBest prediction combination:')\n",
    "print(Data[i:(i+1)],'\\n')\n",
    "\n",
    "# Visualize some parameters combinations \n",
    "x = sns.boxplot(y=\"val_acc_pearson_r\",x=\"first_neuron\",hue=\"activation\",data=Data).set_title('First neuron x activation')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hyperparameter tunning: 'the real world model'. Talos uses a dictionary for keras hyperparameter optimization. You ha ve to declare the hyperparameters and their boundaries in a python dictionary. \n",
    "Warning! Keras optimizer, losses, activation, initializer need to be loaded! \n",
    "Note that talos Scan includes an argument called grid_downsampling, which allows a random search. Just to illustrate how to use talos, we've chosen 0.01 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 23/23 [12:49<00:00, 33.02s/it]\n"
     ]
    }
   ],
   "source": [
    "# HYPERPARAMETER OPTIMIZATION USING TALOS: more complex example\n",
    "# We do a random search here (by downsampling among all possible grid values)\n",
    "\n",
    "# model definition\n",
    "def grown_model(x, y, x_val, y_val, params):\n",
    "    # next we can build the model exactly like we would normally do it\n",
    "    model = Sequential()\n",
    "    model.add(Dense(params['first_neuron'], input_dim=x.shape[1],\n",
    "                    activation=params['activation'],\n",
    "                    kernel_initializer=params['kernel_initializer']))\n",
    "    \n",
    "    model.add(Dropout(params['dropout']))\n",
    "    \n",
    "    # if we want to also test for number of layers and shapes, that's possible\n",
    "    hidden_layers(model, params, 1)\n",
    "\n",
    "    # then we finish again with completely standard Keras way\n",
    "    model.add(Dense(1, activation=params['last_activation'],\n",
    "                    kernel_initializer='normal'))\n",
    "    \n",
    "    model.compile(loss=params['losses'],\n",
    "        optimizer=params['optimizer'](lr=lr_normalizer(params['lr'],params['optimizer'])),\n",
    "            metrics=[acc_pearson_r])\n",
    "                  \n",
    "   \n",
    "    out = model.fit(x, y, validation_data=[x_val, y_val], verbose=0,batch_size=params['batch_size'],\n",
    "                        epochs=params['epochs'],callbacks=[live()])\n",
    "\n",
    "    # finally we have to make sure that history object and model are returned\n",
    "    return out, model\n",
    "\n",
    "# hyperparameters\n",
    "p = {'first_neuron':[32,64],\n",
    "     'lr':[0.2,0.5],\n",
    "     'batch_size': [30,50],\n",
    "     'hidden_layers':[1,2,3,4],\n",
    "     'epochs': [100],\n",
    "     'dropout': [0, 0.01, 0.1, 0.5],\n",
    "     'optimizer': [Adam,sgd,Nadam],\n",
    "     'losses': [mean_squared_error],\n",
    "     'activation':[relu, elu,linear],\n",
    "     'last_activation': [None],\n",
    "     'kernel_initializer':[\"uniform\",\"normal\"]}\n",
    "\n",
    "# Split the train set into proper train & validation\n",
    "X_train0, X_val, y_train0, y_val = train_test_split(X_train, y_train, test_size=0.1)\n",
    "X_train0=np.asarray(X_train0)\n",
    "X_val=np.asarray(X_val)\n",
    "y_train0=np.asarray(y_train0)\n",
    "y_val=np.asarray(y_val)\n",
    "\n",
    "# Example with multiclass target\n",
    "# grid_downsample number of combinations to be checked 10% in this example\n",
    "tcomp = ta.Scan(x=X_train0,\n",
    "                y=y_train0,\n",
    "                x_val=X_val,\n",
    "                y_val=y_val,\n",
    "                model=grown_model, \n",
    "                params=p, \n",
    "                grid_downsample=0.01,\n",
    "                print_params=True,\n",
    "                dataset_name='grown_model')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# HYPERPARAMETER OPTIMIZATION USING TALOS: more complex example\n",
    "# Some plotting\n",
    "Data=pd.DataFrame(tcomp.data)\n",
    "Data[\"val_loss\"] = pd.to_numeric(Data[\"val_loss\"])\n",
    "Data[\"acc_pearson_r\"] = pd.to_numeric(Data[\"acc_pearson_r\"])\n",
    "Data[\"val_acc_pearson_r\"] = pd.to_numeric(Data[\"val_acc_pearson_r\"])\n",
    "#print Data\n",
    "#print(Data)\n",
    "#write Data\n",
    "Data.to_csv(\"mlp_real_world.csv\",index=False)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Best loss combination:\n",
      "  round_epochs  val_loss  val_acc_pearson_r                loss  \\\n",
      "4          100  0.629735           0.623303  0.0781390678543779   \n",
      "\n",
      "   acc_pearson_r first_neuron   lr batch_size hidden_layers epochs dropout  \\\n",
      "4       0.963721           64  0.5         30             4    100     0.1   \n",
      "\n",
      "                          optimizer  \\\n",
      "4  <class 'keras.optimizers.Nadam'>   \n",
      "\n",
      "                                            losses  \\\n",
      "4  <function mean_squared_error at 0x7f04dbb38378>   \n",
      "\n",
      "                          activation last_activation kernel_initializer  \n",
      "4  <function relu at 0x7f04dbb64620>            None             normal   \n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x475.2 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Maximum correlation\n",
    "i=Data['val_acc_pearson_r'].argmax()\n",
    "print('\\nBest loss combination:')\n",
    "print(Data[i:(i+1)],'\\n')\n",
    "# talos have a function called reporting, which has implemented some plots \n",
    "# reporting to see the best model, draw an histogram\n",
    "r = ta.Reporting('mlp_real_world.csv')\n",
    "r.plot_hist('val_acc_pearson_r')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.6233031004667282\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x648 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x475.2 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(r.high('val_acc_pearson_r'))\n",
    "# we can check the correlation between val accuracy and other parameters into the model\n",
    "r.plot_corr('val_acc_pearson_r')\n",
    "# we can see the distribution of the val_acc parameter chosen \n",
    "r.plot_kde('val_acc_pearson_r')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x576 with 8 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "# You also can use seaborn to check other combination of parameters\n",
    "sns.set_context(\"paper\", rc={\"font.size\":15,\"axes.titlesize\":15,\"axes.labelsize\":15})   \n",
    "g = sns.FacetGrid(tcomp.data, col=\"hidden_layers\",row=\"first_neuron\", margin_titles=True,height=4)\n",
    "g.map(sns.boxplot,\"lr\",\"val_acc_pearson_r\", palette=sns.cubehelix_palette(8),saturation=.5);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.7279666910087476,\n",
       " 0.6151571066733459,\n",
       " 0.8163882441795121,\n",
       " 0.6710367840631976,\n",
       " 0.5476321105866963]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Model evaluation \n",
    "# talos has a function to evaluate the model from Scan object \n",
    "\n",
    "from talos import Evaluate\n",
    "\n",
    "# create the evaluate object\n",
    "e = Evaluate(tcomp)\n",
    "\n",
    "# perform the evaluation\n",
    "e.evaluate(X_val, y_val,metric=\"val_acc_pearson_r\" ,mode=\"regression\")\n",
    "# model regression uses the mean absolute error by default. Custom metrics are not allowing in evaluate\n",
    "#Warning: note we use the same data than in scan validation, however, it should be used a fully different \n",
    "# dataset (the validation set)!  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, use Deploy() to save and transfer the results of your \"star\" model "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Deploy package the_real_word_experiment have been saved.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<talos.commands.deploy.Deploy at 0x7f048c66cfd0>"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from talos import Deploy\n",
    "\n",
    "Deploy(tcomp, 'the_real_word_experiment',metric='val_acc_pearson_r',asc=False)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "N / class\n",
      "Test  [103, 324, 52]\n",
      "Train  [28, 75, 16]\n",
      "All  [131, 399, 68]\n",
      "120/120 [==============================] - 0s 326us/step\n",
      "\n",
      "MSE in prediction = 2.0637567043304443\n",
      "\n",
      "Probabilities matrix\n",
      " [[3.75487030e-01 5.92163086e-01 3.23316455e-02 1.82374915e-05]\n",
      " [1.21018790e-01 7.91246057e-01 8.76956880e-02 3.94919152e-05]\n",
      " [6.22581318e-02 6.69291258e-01 2.68378854e-01 7.17674266e-05]\n",
      " [8.53250967e-04 5.75047374e-01 4.24076319e-01 2.30197638e-05]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# MLP example with multiclass target\n",
    "\n",
    "# Let us round class vector and make a few classes, all positive numbers\n",
    "# just a trick to convert to classes\n",
    "yi_train=[int(round(x-np.min(y_train))/2) for x in y_train]\n",
    "yi_test=[int(round(x-np.min(y_test))/2) for x in y_test]\n",
    "\n",
    "# N obs / clas; this may result in some very rare classes so consider merging\n",
    "print('N / class')\n",
    "print('Test ',[yi_train.count(i) for i in range(max(yi_train+yi_test))])\n",
    "print('Train ',[yi_test.count(i) for i in range(max(yi_train+yi_test))])\n",
    "print('All ',[(yi_test+yi_train).count(i) for i in range(max(yi_train+yi_test))])\n",
    "# WARNING: make sure all clases in test are in train!!\n",
    "\n",
    "# convert to classes, i need to make all classes equivalent\n",
    "n_train=len(yi_train)\n",
    "itemp = to_categorical(yi_train+yi_test)\n",
    "i_train = itemp[:n_train,:]\n",
    "i_test = itemp[n_train:,:]\n",
    "\n",
    "# no. of SNPs in data\n",
    "nSNP=X_train.shape[1]\n",
    "nClasses=i_train.shape[1]\n",
    "\n",
    "# Instantiate\n",
    "model = Sequential()\n",
    "\n",
    "# Add first layer\n",
    "model.add(Dense(64, input_dim=nSNP))\n",
    "model.add(Activation('relu'))\n",
    "# Add second layer\n",
    "model.add(Dense(32))\n",
    "model.add(Activation('softplus'))\n",
    "# Last, output layer\n",
    "model.add(Dense(nClasses, activation='softmax'))\n",
    "\n",
    "# Model Compiling \n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam')\n",
    "\n",
    "# training\n",
    "model.fit(X_train, i_train, epochs=100, verbose=0)\n",
    "\n",
    "# cross-validation: get predicted target values\n",
    "i_hat = model.predict(X_test, batch_size=128)\n",
    "\n",
    "mse_prediction = model.evaluate(X_test, i_test, batch_size=128)\n",
    "print('\\nMSE in prediction =',mse_prediction)\n",
    "\n",
    "# do a heatplot, obs vs expected class distribution\n",
    "# collect all results by class\n",
    "# compute average prediction by class\n",
    "heat = np.zeros([nClasses,nClasses])\n",
    "for i in range(nClasses):\n",
    "    iclass = np.nonzero(i_test[:,i]>0) # samples of i-th class\n",
    "    for j in range(nClasses):\n",
    "        heat[i,j] = np.mean(i_hat[iclass,j])\n",
    "\n",
    "# plot observed vs. predicted targets\n",
    "print('\\nProbabilities matrix\\n',heat)\n",
    "plot = sns.heatmap(heat, cmap=\"Blues\")\n",
    "plot.set(xlabel='Observed class', ylabel='Predicted class')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv1d_1 (Conv1D)            (None, 46, 32)            128       \n",
      "_________________________________________________________________\n",
      "max_pooling1d_1 (MaxPooling1 (None, 23, 32)            0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 736)               0         \n",
      "_________________________________________________________________\n",
      "dense_17 (Dense)             (None, 64)                47168     \n",
      "_________________________________________________________________\n",
      "activation_14 (Activation)   (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_18 (Dense)             (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "activation_15 (Activation)   (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dense_19 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 49,409\n",
      "Trainable params: 49,409\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "120/120 [==============================] - 0s 2ms/step\n",
      "\n",
      "MSE in prediction = 1.224562168121338\n",
      "\n",
      "Corr obs vs pred = 0.2684987969727777\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#--> CNN example\n",
    "\n",
    "nSNP=X_train.shape[1] \n",
    "nStride=3  # stride between convolutions\n",
    "nFilter=32 # no. of convolutions\n",
    "\n",
    "# Instantiate\n",
    "model_cnn = Sequential()\n",
    "\n",
    "#WARNING!!! I need this to match dimensions \n",
    "#https://stackoverflow.com/questions/43396572/dimension-of-shape-in-conv1d\n",
    "X2_train = np.expand_dims(X_train, axis=2) \n",
    "X2_test = np.expand_dims(X_test, axis=2) \n",
    "\n",
    "# add convolutional layer\n",
    "model_cnn.add(Conv1D(nFilter, kernel_size=3, strides=nStride, input_shape=(nSNP,1)))\n",
    "# add pooling layer: takes maximum of two consecutive values\n",
    "model_cnn.add(MaxPooling1D(pool_size=2))\n",
    "# Solutions above are linearized to accommodate a standard layer\n",
    "model_cnn.add(Flatten())\n",
    "model_cnn.add(Dense(64))\n",
    "model_cnn.add(Activation('relu'))\n",
    "model_cnn.add(Dense(32))\n",
    "model_cnn.add(Activation('softplus'))\n",
    "model_cnn.add(Dense(1))\n",
    "\n",
    "# Model Compiling (https://keras.io/models/sequential/) \n",
    "model_cnn.compile(loss='mean_squared_error', optimizer='sgd')\n",
    "\n",
    "# list some properties\n",
    "model_cnn.summary()\n",
    "\n",
    "# training\n",
    "model_cnn.fit(X2_train, y_train, epochs=200, verbose=0)\n",
    "\n",
    "# cross-validation\n",
    "mse_prediction = model_cnn.evaluate(X2_test, y_test, batch_size=128)\n",
    "print('\\nMSE in prediction =',mse_prediction)\n",
    "\n",
    "# get predicted target values\n",
    "y_hat = model_cnn.predict(X2_test, batch_size=128)\n",
    "\n",
    "# correlation btw predicted and observed\n",
    "corr = np.corrcoef(y_test,y_hat[:,0])[0,1]\n",
    "print('\\nCorr obs vs pred =',corr)\n",
    "\n",
    "# plot observed vs. predicted targets\n",
    "plt.title('MLP: Observed vs Predicted Y')\n",
    "plt.ylabel('Predicted')\n",
    "plt.xlabel('Observed')\n",
    "plt.scatter(y_test, y_hat, marker='o')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(479, 138)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lstm_49 (LSTM)               (None, None, 32)          4352      \n",
      "_________________________________________________________________\n",
      "dropout_46 (Dropout)         (None, None, 32)          0         \n",
      "_________________________________________________________________\n",
      "lstm_50 (LSTM)               (None, None, 64)          24832     \n",
      "_________________________________________________________________\n",
      "dropout_47 (Dropout)         (None, None, 64)          0         \n",
      "_________________________________________________________________\n",
      "lstm_51 (LSTM)               (None, 64)                33024     \n",
      "_________________________________________________________________\n",
      "dropout_48 (Dropout)         (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_29 (Dense)             (None, 1)                 65        \n",
      "_________________________________________________________________\n",
      "activation_23 (Activation)   (None, 1)                 0         \n",
      "=================================================================\n",
      "Total params: 62,273\n",
      "Trainable params: 62,273\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f18bc15d240>"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#--> RNN example\n",
    "\n",
    "nSNP=X_train.shape[1] \n",
    "\n",
    "#-->data shape\n",
    "X2_train = np.expand_dims(X_train, axis=2) \n",
    "X2_test = np.expand_dims(X_test, axis=2) \n",
    "\n",
    "# Instantiate\n",
    "model_cnn = Sequential()\n",
    "\n",
    "# Instantiate\n",
    "model = Sequential()\n",
    "model.add(LSTM(32,return_sequences=True,input_shape=(None,1), activation=\"tanh\"))\n",
    "model.add(Dropout(0.1))\n",
    "model.add(LSTM(64, return_sequences=True, activation=\"tanh\"))\n",
    "model.add(Dropout(0.1))\n",
    "model.add(LSTM(64, activation=\"tanh\"))\n",
    "model.add(Dropout(0.1))\n",
    "model.add(Dense(units=1))\n",
    "model.add(Activation(\"tanh\"))\n",
    "model.compile(optimizer=\"adam\",loss=mean_squared_error, metrics=['mae'])\n",
    "model.summary()\n",
    "\n",
    "model.fit(X2_train, y_train, epochs=200, verbose=0)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "120/120 [==============================] - 1s 9ms/step\n",
      "\n",
      "MSE in prediction = [1.1969413757324219, 0.8699095845222473]\n",
      "\n",
      "Corr obs vs pred = 0.2293501665943067\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# cross-validation\n",
    "mse_prediction = model.evaluate(X2_test, y_test, batch_size=128)\n",
    "print('\\nMSE in prediction =',mse_prediction)\n",
    "\n",
    "# get predicted target values\n",
    "y_hat = model.predict(X2_test, batch_size=128)\n",
    "\n",
    "# correlation btw predicted and observed\n",
    "corr = np.corrcoef(y_test,y_hat[:,0])[0,1]\n",
    "print('\\nCorr obs vs pred =',corr)\n",
    "\n",
    "# plot observed vs. predicted targets\n",
    "plt.title('MLP: Observed vs Predicted Y')\n",
    "plt.ylabel('Predicted')\n",
    "plt.xlabel('Observed')\n",
    "plt.scatter(y_test, y_hat, marker='o')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save and reuse model\n",
    "from keras.models import load_model\n",
    "\n",
    "# creates a HDF5 file 'my_model.h5'\n",
    "model.save('my_model.h5')  \n",
    "\n",
    "# deletes the existing model\n",
    "del model\n",
    "\n",
    "# loads a compiled model, identical to the previous one\n",
    "model = load_model('my_model.h5')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}