nb_simple.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "from nikmod import nik\n",
    "import pandas as pd\n",
    "import scipy.special as special\n",
    "import tensorflow as tf\n",
    "import keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>utr</th>\n",
       "      <th>N45a</th>\n",
       "      <th>pPAS</th>\n",
       "      <th>N45b</th>\n",
       "      <th>dPAS</th>\n",
       "      <th>N45c</th>\n",
       "      <th>p</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>sim</td>\n",
       "      <td>GTTCGTTCCATTTGCCACTAACGTGTGCAATGTATCTCTCATGTG</td>\n",
       "      <td>AATACA</td>\n",
       "      <td>TTTGAAAGTATTCTTACACCTCCTAATTGTACCTGTTGGCAAATT</td>\n",
       "      <td>ATTAAA</td>\n",
       "      <td>ATTATTGTGAAAATTGCTGTGGGTCGTAGGTCTAGTAGTTTTGGG</td>\n",
       "      <td>941</td>\n",
       "      <td>575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>sim</td>\n",
       "      <td>ACAAGCGCTGGTCGGCGTAACTCGAAGTAATGATATCATCCCATG</td>\n",
       "      <td>AATAAC</td>\n",
       "      <td>TTGCAAGGTTATATTAGTGCCTCTTGGCTTCAATAAAATCATTTA</td>\n",
       "      <td>ACTACA</td>\n",
       "      <td>ACATTCATGTGTCCACGTGGTGCGTTGGCCAGTCTGTCTTTTACT</td>\n",
       "      <td>907</td>\n",
       "      <td>536</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sim</td>\n",
       "      <td>GCGCCCCTTGAATACGCGGGTCTAGTGTTCCGCCATGTATGCTAC</td>\n",
       "      <td>AATAAA</td>\n",
       "      <td>ATCCTCTTTTCCTACCTATTGTCTCTTCTTGTAAAGCATTTTGCA</td>\n",
       "      <td>AATAAA</td>\n",
       "      <td>TTGTTCGTTGTTGTGTAGGACCATATGCGGGGGTTAAGTGCGGCT</td>\n",
       "      <td>1153</td>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>sim</td>\n",
       "      <td>TCTCTGTGTCTGGTACATACAACCCTAAGCTGTAAACAGTGGTTC</td>\n",
       "      <td>AATAAA</td>\n",
       "      <td>TTTATTTACTGGCATCACCTCATCTATCCCTTTTTGTGGTGTGAG</td>\n",
       "      <td>ATTAAA</td>\n",
       "      <td>GGGTTTTACTCTAGGGGGCGAGTCCGTTATGTTTTACTCCCTACA</td>\n",
       "      <td>1077</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>sim</td>\n",
       "      <td>TGCAAGCGTTAGGGGTTCGATTCCTCTGTTAGAGAGTTGGTATTC</td>\n",
       "      <td>AATAAA</td>\n",
       "      <td>CACCGGCTCTATTTTTTTGTTTTATGATTCTTATGTCAAGTGGTT</td>\n",
       "      <td>ATCAAA</td>\n",
       "      <td>TTAAAGATCTTCTAGTGTATATAATCTCTACAACTATCTTATGGG</td>\n",
       "      <td>995</td>\n",
       "      <td>58</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   utr                                           N45a    pPAS  \\\n",
       "0  sim  GTTCGTTCCATTTGCCACTAACGTGTGCAATGTATCTCTCATGTG  AATACA   \n",
       "1  sim  ACAAGCGCTGGTCGGCGTAACTCGAAGTAATGATATCATCCCATG  AATAAC   \n",
       "2  sim  GCGCCCCTTGAATACGCGGGTCTAGTGTTCCGCCATGTATGCTAC  AATAAA   \n",
       "3  sim  TCTCTGTGTCTGGTACATACAACCCTAAGCTGTAAACAGTGGTTC  AATAAA   \n",
       "4  sim  TGCAAGCGTTAGGGGTTCGATTCCTCTGTTAGAGAGTTGGTATTC  AATAAA   \n",
       "\n",
       "                                            N45b    dPAS  \\\n",
       "0  TTTGAAAGTATTCTTACACCTCCTAATTGTACCTGTTGGCAAATT  ATTAAA   \n",
       "1  TTGCAAGGTTATATTAGTGCCTCTTGGCTTCAATAAAATCATTTA  ACTACA   \n",
       "2  ATCCTCTTTTCCTACCTATTGTCTCTTCTTGTAAAGCATTTTGCA  AATAAA   \n",
       "3  TTTATTTACTGGCATCACCTCATCTATCCCTTTTTGTGGTGTGAG  ATTAAA   \n",
       "4  CACCGGCTCTATTTTTTTGTTTTATGATTCTTATGTCAAGTGGTT  ATCAAA   \n",
       "\n",
       "                                            N45c     p    d  \n",
       "0  ATTATTGTGAAAATTGCTGTGGGTCGTAGGTCTAGTAGTTTTGGG   941  575  \n",
       "1  ACATTCATGTGTCCACGTGGTGCGTTGGCCAGTCTGTCTTTTACT   907  536  \n",
       "2  TTGTTCGTTGTTGTGTAGGACCATATGCGGGGGTTAAGTGCGGCT  1153   18  \n",
       "3  GGGTTTTACTCTAGGGGGCGAGTCCGTTATGTTTTACTCCCTACA  1077    9  \n",
       "4  TTAAAGATCTTCTAGTGTATATAATCTCTACAACTATCTTATGGG   995   58  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sim = pd.read_pickle('simple.pkl')\n",
    "sim.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY0AAAD8CAYAAACLrvgBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7ba9407150>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sim['n'] = sim['p'] + sim['d']; sim.query('n >= 10')\n",
    "sim['a'] = sim['N45b'].apply(lambda x: x.count('A'))#; sim.query('a >= 12')\n",
    "sim['p'] = sim['p'].apply(lambda x: 0.99 if x == 0 else x)\n",
    "sim['d'] = sim['d'].apply(lambda x: 0.99 if x == 0 else x)\n",
    "sim['logit'] = special.logit(sim['p']/(sim['p']+sim['d']))*-1 \n",
    "sim.sort_values('n',ascending=False,inplace=True)\n",
    "sim.reset_index(drop=True,inplace=True)\n",
    "sim['logit'].hist(bins=50,grid=False);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "seqs = sim['N45a']+sim['pPAS']+sim['N45b']+sim['dPAS']+sim['N45c']\n",
    "# seqs = sim['N45a']+sim['N45b']+sim['N45c']\n",
    "seqs = nik.one_hot_encoder(seqs)\n",
    "vals = sim['logit']; del sim"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train:            1,627,149\n",
      "test:               180,794\n"
     ]
    }
   ],
   "source": [
    "seqs_train, seqs_test, vals_train, vals_test = nik.seq_data_splitter(seqs,vals,split=.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn import preprocessing\n",
    "vals_train = preprocessing.StandardScaler().fit_transform(vals_train.values.reshape(-1,1))\n",
    "vals_test = preprocessing.StandardScaler().fit_transform(vals_test.values.reshape(-1,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv1d_1 (Conv1D)            (None, 147, 100)          3300      \n",
      "_________________________________________________________________\n",
      "conv1d_2 (Conv1D)            (None, 147, 100)          80100     \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 14700)             0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 100)               1470100   \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 100)               0         \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 100)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 101       \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 1)                 0         \n",
      "=================================================================\n",
      "Total params: 1,553,601\n",
      "Trainable params: 1,553,601\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "simple_cnn = nik.cnn_constructor(seqs_train,nf1=100,nf2=100,ks1=8,ks2=8,dense=100); simple_cnn.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/6\n",
      "1627149/1627149 [==============================] - 206s - loss: 0.3727   \n",
      "Epoch 2/6\n",
      "1627149/1627149 [==============================] - 219s - loss: 0.3398   \n",
      "Epoch 3/6\n",
      "1627149/1627149 [==============================] - 222s - loss: 0.3310   \n",
      "Epoch 4/6\n",
      "1627149/1627149 [==============================] - 222s - loss: 0.3263   \n",
      "Epoch 5/6\n",
      "1627149/1627149 [==============================] - 222s - loss: 0.3230   \n",
      "Epoch 6/6\n",
      "1627149/1627149 [==============================] - 222s - loss: 0.3204   \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f7b7e9742d0>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adam = keras.optimizers.Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-08)\n",
    "simple_cnn.compile(loss='mean_squared_error', optimizer=adam)\n",
    "simple_cnn.fit(seqs_train, vals_train, batch_size=100, epochs=6, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "simple_cnn.save('simple_cnn.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQEAAAENCAYAAAAPLtCGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npedMVQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7b859b5910>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "vals_test = vals_test.flatten()\n",
    "pred = simple_cnn.predict(seqs_test).flatten()\n",
    "nik.pred_scatter(vals_test,pred,title='SIMPLE.CNN',r2=True)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}