10_plot_ROC_multiclass.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import label_binarize\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn import svm\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib as mpl"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   variant  MCF10A score  CADD  SNAP2  PolyPhen  SIFT  WT_RSA (%)  \\\n",
      "0      K6E        0.3311  26.6     67     0.155  0.99         NaN   \n",
      "1      K6I        0.4544  27.3     64     0.244  1.00         NaN   \n",
      "2     R15S        0.3870  25.0     65     0.123  1.00        53.3   \n",
      "3     M35V        0.3606  25.0     44     0.914  1.00         0.0   \n",
      "4     P38H        0.3521  26.1     44     1.000  1.00         0.0   \n",
      "5     G44D        0.6232  26.0     65     0.968  1.00       117.5   \n",
      "6     A79T        0.5711  23.0    -95     0.005  0.41        86.3   \n",
      "7     D92N        0.2677  28.8     90     1.000  1.00        44.3   \n",
      "8     H93Q        0.3301  26.1     77     1.000  1.00        74.5   \n",
      "9     H93R        0.3930  27.0     83     0.998  0.97        74.5   \n",
      "10   I101F        0.8803  26.0      6     0.998  1.00         0.2   \n",
      "11   I101T        0.7138  29.0     44     0.998  1.00         0.2   \n",
      "\n",
      "    WT_DEPTH (_)  WT_OSP  MT_RSA (%)  MT_DEPTH (_)  MT_OSP  ground truth  \n",
      "0            NaN     NaN         NaN           NaN     NaN           NaN  \n",
      "1            NaN     NaN         NaN           NaN     NaN           NaN  \n",
      "2            3.6    0.32        52.3           3.6    0.37           NaN  \n",
      "3           10.8    0.53         2.7          10.3    0.50           NaN  \n",
      "4           10.5    0.60         0.0          11.3    0.62           NaN  \n",
      "5            3.4    0.14        98.9           3.3    0.12           NaN  \n",
      "6            3.1    0.19        80.2           3.2    0.19           2.0  \n",
      "7            3.7    0.32        44.6           3.9    0.34           NaN  \n",
      "8            3.4    0.22        78.0           3.5    0.21           NaN  \n",
      "9            3.4    0.22        79.1           3.4    0.16           NaN  \n",
      "10           8.7    0.49         0.8           8.8    0.56           NaN  \n",
      "11           8.7    0.49         2.8           8.4    0.44           NaN  \n"
     ]
    }
   ],
   "source": [
    "df = pd.read_csv('Datasets/PTEN_MCF10A_predictions_ref_set_multi_copy.csv',\n",
    "                index_col=False)\n",
    "print(df.head(12))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   variant  MCF10A score  CADD  SNAP2  PolyPhen  SIFT  WT_RSA (%)  \\\n",
      "0     A79T        0.5711  23.0    -95     0.005  0.41        86.3   \n",
      "1    N117S        0.4916  21.0    -92     0.096  0.27        25.3   \n",
      "2    H123Q        0.9934  25.0     82     0.998  0.98         0.0   \n",
      "3    H123Y        0.6383  25.0     80     0.996  1.00         0.0   \n",
      "4    C124S        0.8797  27.3     83     1.000  1.00         4.3   \n",
      "5    C124R        0.6986  28.0     98     0.999  1.00         4.3   \n",
      "6    A126D        0.5325  28.2     76     1.000  1.00        55.7   \n",
      "7    G127R        0.9753  29.0     91     1.000  1.00         0.0   \n",
      "8    G129R        0.8214  29.0     92     1.000  1.00        18.3   \n",
      "9    G129E        0.8841  32.0     95     1.000  1.00        18.3   \n",
      "10   R130Q        0.5361  34.0     96     0.998  0.98         3.1   \n",
      "11   R130P        0.9134  34.0     93     0.999  1.00         3.1   \n",
      "12   R130L        0.4728  34.0     92     0.999  1.00         3.1   \n",
      "13   T131I        0.4788  26.0     80     0.999  0.98         0.0   \n",
      "14   G132D        0.6304  31.0     92     1.000  0.97         0.0   \n",
      "15   Y155C        0.5901  33.0     22     0.999  1.00         0.0   \n",
      "16   T167N        0.5664  26.0     45     0.955  1.00        32.8   \n",
      "17   Y176C        0.5603  23.0    -47     0.000  0.81        52.3   \n",
      "18   Y180H        0.1171  23.0    -78     0.000  0.47        54.0   \n",
      "19   M198I        0.1821  22.0    -85     0.001  0.00         0.0   \n",
      "20   N228S        0.1243  18.0    -78     0.007  0.26       108.4   \n",
      "21   P246L        0.6162  22.0     35     0.116  0.94        45.3   \n",
      "22   N262S        0.1021  23.0    -63     0.007  0.79        86.7   \n",
      "23   D268E        0.1080  18.0    -89     0.007  0.00        55.5   \n",
      "\n",
      "    WT_DEPTH (_)  WT_OSP  MT_RSA (%)  MT_DEPTH (_)  MT_OSP  ground truth  \n",
      "0            3.1    0.19        80.2           3.2    0.19           2.0  \n",
      "1            4.0    0.51        10.1           4.0    0.53           1.0  \n",
      "2           11.0    0.60         0.0          11.4    0.64           1.0  \n",
      "3           11.0    0.60         0.0          12.1    0.58           1.0  \n",
      "4            6.7    0.52         4.3           7.3    0.54           2.0  \n",
      "5            6.7    0.52         8.1           5.8    0.48           2.0  \n",
      "6            3.1    0.23        68.7           3.5    0.21           1.0  \n",
      "7            5.2    0.49         0.0           8.3    0.74           2.0  \n",
      "8            6.1    0.44        23.3           4.7    0.37           2.0  \n",
      "9            6.1    0.44         4.5           7.3    0.61           2.0  \n",
      "10           6.5    0.49        11.6           6.8    0.49           1.0  \n",
      "11           6.5    0.49        15.6           6.3    0.44           2.0  \n",
      "12           6.5    0.49         8.0           6.9    0.47           1.0  \n",
      "13           9.1    0.59         0.0           9.1    0.64           1.0  \n",
      "14          11.0    0.64         0.0          10.4    0.72           1.0  \n",
      "15           8.7    0.51         4.0           9.1    0.39           1.0  \n",
      "16           3.7    0.29        34.3           3.8    0.28           1.0  \n",
      "17           3.6    0.33        75.7           3.3    0.28           0.0  \n",
      "18           3.8    0.35        54.7           3.6    0.33           0.0  \n",
      "19          11.3    0.55         0.0          11.0    0.52           0.0  \n",
      "20           3.3    0.15       105.2           3.0    0.15           0.0  \n",
      "21           3.4    0.27        55.1           3.7    0.28           1.0  \n",
      "22           3.4    0.20        85.1           3.3    0.19           0.0  \n",
      "23           3.6    0.30        51.0           3.5    0.28           0.0  \n"
     ]
    }
   ],
   "source": [
    "df = df.dropna().reset_index(drop=True)\n",
    "print(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Binarize the output\n",
    "from sklearn.preprocessing import label_binarize\n",
    "\n",
    "y = label_binarize(df['ground truth'], classes=[0, 1, 2])\n",
    "n_classes = y.shape[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X = df.iloc[:, 7:-1]\n",
    "X_MCF10A = X.assign(MCF10A=df['MCF10A score'])\n",
    "X_CADD = X.assign(CADD=df['CADD'])\n",
    "X_SNAP2 = X.assign(SNAP2=df['SNAP2'])\n",
    "X_SIFT = X.assign(SIFT=df['SIFT'])\n",
    "X_PolyPhen = X.assign(PolyPhen=df['PolyPhen'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    WT_DEPTH (_)  WT_OSP  MT_RSA (%)  MT_DEPTH (_)  MT_OSP  MCF10A\n",
      "0            3.1    0.19        80.2           3.2    0.19  0.5711\n",
      "1            4.0    0.51        10.1           4.0    0.53  0.4916\n",
      "2           11.0    0.60         0.0          11.4    0.64  0.9934\n",
      "3           11.0    0.60         0.0          12.1    0.58  0.6383\n",
      "4            6.7    0.52         4.3           7.3    0.54  0.8797\n",
      "5            6.7    0.52         8.1           5.8    0.48  0.6986\n",
      "6            3.1    0.23        68.7           3.5    0.21  0.5325\n",
      "7            5.2    0.49         0.0           8.3    0.74  0.9753\n",
      "8            6.1    0.44        23.3           4.7    0.37  0.8214\n",
      "9            6.1    0.44         4.5           7.3    0.61  0.8841\n",
      "10           6.5    0.49        11.6           6.8    0.49  0.5361\n",
      "11           6.5    0.49        15.6           6.3    0.44  0.9134\n",
      "12           6.5    0.49         8.0           6.9    0.47  0.4728\n",
      "13           9.1    0.59         0.0           9.1    0.64  0.4788\n",
      "14          11.0    0.64         0.0          10.4    0.72  0.6304\n",
      "15           8.7    0.51         4.0           9.1    0.39  0.5901\n",
      "16           3.7    0.29        34.3           3.8    0.28  0.5664\n",
      "17           3.6    0.33        75.7           3.3    0.28  0.5603\n",
      "18           3.8    0.35        54.7           3.6    0.33  0.1171\n",
      "19          11.3    0.55         0.0          11.0    0.52  0.1821\n",
      "20           3.3    0.15       105.2           3.0    0.15  0.1243\n",
      "21           3.4    0.27        55.1           3.7    0.28  0.6162\n",
      "22           3.4    0.20        85.1           3.3    0.19  0.1021\n",
      "23           3.6    0.30        51.0           3.5    0.28  0.1080\n"
     ]
    }
   ],
   "source": [
    "print(X_MCF10A)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# using simple classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Learn to predict each class against the other\n",
    "# using Support Vector Classification, kernal type is linear\n",
    "\n",
    "def classification(dset):\n",
    "    X_train, X_test, y_train, y_test = train_test_split(\n",
    "                            dset, y, test_size=0.33,\n",
    "                                            random_state=0)\n",
    "    \n",
    "    classifier = OneVsRestClassifier(svm.SVC(kernel='linear', \n",
    "                                         probability=True, random_state=0))\n",
    "    \n",
    "    y_score = classifier.fit(X_train, y_train).decision_function(\n",
    "                                                    X_test)\n",
    "    return y_score, y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Compute micro-average ROC curve and area\n",
    "def miu_avg_roc(test, score):\n",
    "    fpr, tpr, _ = roc_curve(test.ravel(), score.ravel())\n",
    "    # .ravel() returns a contiguous flattened array\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    return fpr, tpr, roc_auc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_roc(fpr, tpr, roc_auc, name):\n",
    "    plt.plot(fpr, tpr, label='AUC for {0} is {1:0.3f}'.format(name, roc_auc))\n",
    "    plt.plot([0,1], [0,1], color='navy', linestyle='--')\n",
    "    \n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.ylabel('True Positive Rate')\n",
    "#    plt.xlim([0.0, 1.0])\n",
    "#    plt.ylim([0.0, 1.05])\n",
    "#    plt.title('micro-avg ROC curves')\n",
    "    plt.legend(loc=\"lower right\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_score_MCF10A, y_test = classification(X_MCF10A)\n",
    "fpr_MCF10A, tpr_MCF10A, roc_auc_MCF10A = miu_avg_roc(y_test, y_score_MCF10A)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[-1.71602587 -0.82642106  0.61914366]\n",
      " [-1.66004299 -0.74496142  0.25188996]\n",
      " [ 0.29198662 -0.77767726 -2.58442396]\n",
      " [-0.24652636  0.56080859 -0.94774179]\n",
      " [ 0.96839954 -0.54886044 -3.55512062]\n",
      " [-2.34103447 -1.85145074  1.75898232]\n",
      " [-0.97689676 -0.00726119 -0.37236216]\n",
      " [-0.7421661  -1.24596644 -0.75997125]]\n",
      "\n",
      " [[0 0 1]\n",
      " [0 1 0]\n",
      " [1 0 0]\n",
      " [0 1 0]\n",
      " [1 0 0]\n",
      " [0 1 0]\n",
      " [0 1 0]\n",
      " [1 0 0]]\n"
     ]
    }
   ],
   "source": [
    "print(y_score_MCF10A)\n",
    "print('\\n', y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.     0.0625 0.0625 0.125  0.125  0.3125 0.3125 0.8125 0.8125 1.    ] \n",
      " [0.    0.    0.5   0.5   0.625 0.625 0.875 0.875 1.    1.   ]\n"
     ]
    }
   ],
   "source": [
    "print(fpr_MCF10A, '\\n', tpr_MCF10A)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_score_CADD, y_test = classification(X_CADD)\n",
    "fpr_CADD, tpr_CADD, roc_auc_CADD = miu_avg_roc(y_test, y_score_CADD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_score_SNAP2, y_test = classification(X_SNAP2)\n",
    "fpr_SNAP2, tpr_SNAP2, roc_auc_SNAP2 = miu_avg_roc(y_test, y_score_SNAP2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_score_SIFT, y_test = classification(X_SIFT)\n",
    "fpr_SIFT, tpr_SIFT, roc_auc_SIFT = miu_avg_roc(y_test, y_score_SIFT)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_score_PolyPhen, y_test = classification(X_PolyPhen)\n",
    "fpr_PolyPhen, tpr_PolyPhen, roc_auc_PolyPhen = miu_avg_roc(y_test, y_score_PolyPhen)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_score_baseline, y_test = classification(X)\n",
    "fpr_baseline, tpr_baseline, roc_auc_baseline = miu_avg_roc(y_test, y_score_baseline)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEFQIpxAEQRBUCKcQBEEQVPgfeno280YA590AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plot_roc(fpr_MCF10A, tpr_MCF10A, roc_auc_MCF10A, 'MCF10A')\n",
    "plot_roc(fpr_CADD, tpr_CADD, roc_auc_CADD, 'CADD')\n",
    "plot_roc(fpr_SNAP2, tpr_SNAP2, roc_auc_CADD, 'SNAP2')\n",
    "plot_roc(fpr_SIFT, tpr_SIFT, roc_auc_SIFT, 'SIFT')\n",
    "plot_roc(fpr_PolyPhen, tpr_PolyPhen, roc_auc_PolyPhen, 'PolyPhen2')\n",
    "plot_roc(fpr_baseline, tpr_baseline, roc_auc_baseline, 'baseline')\n",
    "\n",
    "mpl.rcParams['pdf.fonttype'] = 42\n",
    "plt.savefig('Output PDFs/micro-avg roc.pdf', transparent=True, dpi=200)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# calculate macro-average roc\n",
    "def mac_avg_roc(y_test, y_score):\n",
    "    from scipy import interp\n",
    "    \n",
    "    # first ompute ROC curves and ROC areas for each class\n",
    "    fpr = dict()\n",
    "    tpr = dict()\n",
    "    roc_auc = dict()\n",
    "\n",
    "    for i in range(n_classes):\n",
    "        fpr[i], tpr[i], _ = roc_curve(y_test[:, i], y_score[:, i])\n",
    "        roc_auc[i] = auc(fpr[i], tpr[i])\n",
    "    \n",
    "    # aggregate all false positive rates\n",
    "    all_fpr = np.unique(np.concatenate([fpr[i] for i in range(n_classes)]))\n",
    "\n",
    "    # Then interpolate all ROC curves at these points    \n",
    "    mean_tpr = np.zeros_like(all_fpr)\n",
    "    for i in range(n_classes):\n",
    "        mean_tpr += interp(all_fpr, fpr[i], tpr[i])\n",
    "\n",
    "    # Finally average it and compute AUC        \n",
    "    mean_tpr /= n_classes\n",
    "    \n",
    "    fpr_macro = np.insert(all_fpr, 0, 0)\n",
    "    tpr_macro = np.insert(mean_tpr, 0, 0)\n",
    "    \n",
    "    return fpr_macro, tpr_macro"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "fpr_MCF10A_macro, tpr_MCF10A_macro = mac_avg_roc(y_test, y_score_MCF10A)\n",
    "roc_auc_macro_MCF10A = auc(fpr_MCF10A_macro, tpr_MCF10A_macro)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.         0.         0.14285714 0.2        0.25       1.        ] \n",
      "\n",
      "[0.         0.38888889 0.72222222 0.83333333 0.91666667 1.        ]\n"
     ]
    }
   ],
   "source": [
    "print(fpr_MCF10A_macro, '\\n')\n",
    "print(tpr_MCF10A_macro)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "all_fpr_CADD, mean_tpr_CADD = mac_avg_roc(y_test, y_score_CADD)\n",
    "roc_auc_macro_CADD = auc(all_fpr_CADD, mean_tpr_CADD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "all_fpr_SNAP2, mean_tpr_SNAP2 = mac_avg_roc(y_test, y_score_SNAP2)\n",
    "roc_auc_macro_SNAP2 = auc(all_fpr_SNAP2, mean_tpr_SNAP2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "all_fpr_SIFT, mean_tpr_SIFT = mac_avg_roc(y_test, y_score_SIFT)\n",
    "roc_auc_macro_SIFT = auc(all_fpr_SIFT, mean_tpr_SIFT)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "all_fpr_PolyPhen, mean_tpr_PolyPhen = mac_avg_roc(y_test, y_score_PolyPhen)\n",
    "roc_auc_macro_PolyPhen = auc(all_fpr_PolyPhen, mean_tpr_PolyPhen)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "fpr_baseline_macro, tpr_baseline_macro = mac_avg_roc(y_test, y_score_baseline)\n",
    "roc_auc_macro_baseline = auc(fpr_baseline_macro, fpr_baseline_macro)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrF4+AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plot_roc(fpr_MCF10A_macro, tpr_MCF10A_macro, roc_auc_macro_MCF10A, 'MCF10A')\n",
    "plot_roc(all_fpr_CADD, mean_tpr_CADD, roc_auc_macro_CADD, 'CADD')\n",
    "plot_roc(all_fpr_SNAP2, mean_tpr_SNAP2, roc_auc_macro_SNAP2, 'SNAP2')\n",
    "plot_roc(all_fpr_SIFT, mean_tpr_SIFT, roc_auc_macro_SIFT, 'SIFT')\n",
    "plot_roc(all_fpr_PolyPhen, mean_tpr_PolyPhen, roc_auc_macro_PolyPhen, 'PolyPhen')\n",
    "plot_roc(fpr_baseline_macro, tpr_baseline_macro, roc_auc_macro_baseline, 'baseline')\n",
    "\n",
    "mpl.rcParams['pdf.fonttype'] = 42\n",
    "plt.savefig('Output PDFs/macro-avg roc multi.pdf', transparent=True, dpi=200)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0 0 1]\n",
      " [0 1 0]\n",
      " [0 1 0]\n",
      " [0 1 0]\n",
      " [0 0 1]\n",
      " [0 0 1]\n",
      " [0 1 0]\n",
      " [0 0 1]\n",
      " [0 0 1]\n",
      " [0 0 1]\n",
      " [0 1 0]\n",
      " [0 0 1]\n",
      " [0 1 0]\n",
      " [0 1 0]\n",
      " [0 1 0]\n",
      " [0 1 0]\n",
      " [0 1 0]\n",
      " [1 0 0]\n",
      " [1 0 0]\n",
      " [1 0 0]\n",
      " [1 0 0]\n",
      " [0 1 0]\n",
      " [1 0 0]\n",
      " [1 0 0]]\n"
     ]
    }
   ],
   "source": [
    "print(y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# experimental codes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# either run this classifier (SVC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Learn to predict each class against the other\n",
    "# using C-Support Vector Classification, kernal type is linear\n",
    "from sklearn import svm\n",
    "\n",
    "X_train_MCF10A, X_test_MCF10A, y_train, y_test = train_test_split(\n",
    "    X_MCF10A, y, test_size=.33, random_state=0)\n",
    "\n",
    "classifier = OneVsRestClassifier(svm.SVC(kernel='linear', \n",
    "                                         probability=True, random_state=0))\n",
    "\n",
    "y_score_MCF10A = classifier.fit(X_train_MCF10A, y_train).decision_function(X_test_MCF10A)\n",
    "#y_score_CADD = classifier.decision_function(X_CADD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# or run this one (Decision Tree)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Learn to predict each class against the other\n",
    "# using Decision Tree classifier\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "\n",
    "classifier = OneVsRestClassifier(DecisionTreeClassifier(random_state=0))\n",
    "\n",
    "y_score_MCF10A = classifier.fit(X_train_MCF10A, y_train).predict_proba(X_test_MCF10A)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Compute micro-average ROC curve and area\n",
    "fpr_MCF10A, tpr_MCF10A, _ = roc_curve(y_test.ravel(), y_score_MCF10A.ravel())\n",
    "roc_auc_MCF10A = auc(fpr_MCF10A, tpr_MCF10A)\n",
    "\n",
    "#fpr_CADD, tpr_CADD, _ = roc_curve(y_test.ravel(), y_score_CADD.ravel())\n",
    "#roc_auc_CADD = auc(fpr_CADD, tpr_CADD)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJD9xnFMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.plot(fpr_MCF10A, tpr_MCF10A,\n",
    "         label='AUC MCF10A (area = {0:0.2f})'.format(roc_auc_MCF10A),\n",
    "         color='deeppink')\n",
    "#ax.plot(fpr_CADD, tpr_CADD,\n",
    "#         label='AUC CADD (area = {0:0.2f})'.format(roc_auc_CADD),\n",
    "#         color='orange')\n",
    "\n",
    "ax.plot([0, 1], [0, 1], color='navy', linestyle='--')\n",
    "\n",
    "ax.set_xlabel('False Positive Rate')\n",
    "ax.set_ylabel('True Positive Rate')\n",
    "ax.set_title('micro-avg ROC curves')\n",
    "legend = ax.legend(loc=\"lower right\")\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# exercise using the ex from scikit-learn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# shuffle and split training and test sets\n",
    "\n",
    "X_train_MCF10A, X_test_MCF10A, y_train, y_test = train_test_split(\n",
    "    X_MCF10A, y, test_size=.33)\n",
    "\n",
    "# Learn to predict each class against the other\n",
    "classifier = OneVsRestClassifier(svm.SVC(kernel='linear', probability=True,))\n",
    "\n",
    "y_score_MCF10A = classifier.fit(X_train_MCF10A, y_train).decision_function(X_test_MCF10A)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Compute ROC curve and ROC area for each class\n",
    "fpr = dict()\n",
    "tpr = dict()\n",
    "roc_auc = dict()\n",
    "\n",
    "for i in range(n_classes):\n",
    "    fpr[i], tpr[i], _ = roc_curve(y_test[:, i], y_score_MCF10A[:, i])\n",
    "    roc_auc[i] = auc(fpr[i], tpr[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.         0.16666667 0.16666667 1.        ]\n",
      "[0.  0.  0.2 0.2 0.6 0.6 1. ]\n",
      "[0. 0. 0. 1. 1.]\n"
     ]
    }
   ],
   "source": [
    "print(fpr[0])\n",
    "print(fpr[1])\n",
    "print(fpr[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Compute micro-average ROC curve and ROC area\n",
    "fpr[\"micro\"], tpr[\"micro\"], _ = roc_curve(y_test.ravel(), y_score_MCF10A.ravel())\n",
    "roc_auc[\"micro\"] = auc(fpr[\"micro\"], tpr[\"micro\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Compute macro-average ROC curve and ROC area\n",
    "# First aggregate all false positive rates\n",
    "all_fpr = np.unique(np.concatenate([fpr[i] for i in range(n_classes)]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.        , 0.16666667, 0.2       , 0.6       , 1.        ])"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_fpr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Then interpolate all ROC curves at these points\n",
    "from scipy import interp\n",
    "\n",
    "mean_tpr = np.zeros_like(all_fpr)\n",
    "for i in range(n_classes):\n",
    "    mean_tpr += interp(all_fpr, fpr[i], tpr[i]) #'+=' is i_add. a=2. a+=3 returns 5.\n",
    "\n",
    "# Finally average it and compute AUC\n",
    "mean_tpr /= n_classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "all_fpr:  [0.         0.16666667 0.2        0.6        1.        ] \n",
      "\n",
      "mean_tpr [0.33333333 0.66666667 0.77777778 0.88888889 1.        ]\n"
     ]
    }
   ],
   "source": [
    "type(mean_tpr)\n",
    "print('all_fpr: ', all_fpr, '\\n')\n",
    "print('mean_tpr', mean_tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "fpr[\"macro\"] = np.insert(all_fpr, 0, 0)\n",
    "tpr[\"macro\"] = np.insert(mean_tpr, 0, 0)\n",
    "roc_auc[\"macro\"] = auc(fpr[\"macro\"], tpr[\"macro\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot all the classes\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.plot(fpr[0], tpr[0],\n",
    "         label='not pathogenic (area = {0:0.3f}).'.format(roc_auc[0]))\n",
    "ax.plot(fpr[1], tpr[1],\n",
    "         label='likely pathogenic (area = {0:0.3f}).'.format(roc_auc[1]))\n",
    "ax.plot(fpr[2], tpr[2],\n",
    "         label='pathogenic (area = {0:0.3f}).'.format(roc_auc[2]))\n",
    "ax.plot(fpr[\"micro\"], tpr[\"micro\"],\n",
    "         label='micro-average (area = {0:0.2f})'.format(roc_auc[\"micro\"]),\n",
    "         color='deeppink', linestyle=':')\n",
    "ax.plot(fpr[\"macro\"], tpr[\"macro\"],\n",
    "         label='macro-average (area = {0:0.2f})'.format(roc_auc[\"macro\"]),\n",
    "         color='navy', linestyle=':')\n",
    "\n",
    "ax.plot([0, 1], [0, 1], color='navy', linestyle='--')\n",
    "\n",
    "ax.set_xlabel('False Positive Rate')\n",
    "ax.set_ylabel('True Positive Rate')\n",
    "ax.set_title('ROC curves for MCF10A')\n",
    "legend = ax.legend(loc=\"lower right\")\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# trying cross-validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import StratifiedKFold\n",
    "\n",
    "# Run classifier with cross-validation and plot ROC curves\n",
    "cv = StratifiedKFold(n_splits=6)\n",
    "classifier = svm.SVC(kernel='linear', probability=True,\n",
    "                     random_state=0)"
   ]
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}