xena_scraping.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/anne/anaconda/envs/psc/lib/python3.7/site-packages/xenaPython/__init__.py:110: FutureWarning: Possible nested set at position 7\n",
      "  re.sub(r\"^[^[]+[[]([^]]*)[]].*$\", r\"\\1\", query, flags=re.DOTALL))\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns #for plotting\n",
    "\n",
    "#for data fetching\n",
    "import xenaPython as xena\n",
    "\n",
    "# for survival analysis\n",
    "import lifelines\n",
    "from lifelines import CoxPHFitter\n",
    "from lifelines import KaplanMeierFitter"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fetch the data on TCGA (using XenaPython)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_codes(host, dataset, fields, data):\n",
    "    \"get codes for enumerations\"\n",
    "    codes = xena.field_codes(host, dataset, fields)\n",
    "    codes_idx = dict([(x['name'], x['code'].split('\\t')) for x in codes if x['code'] is not None])\n",
    "    for i in range(len(fields)):\n",
    "        if fields[i] in codes_idx:\n",
    "            data[i] = [None if v == 'NaN' else codes_idx[fields[i]][int(v)] for v in data[i]]\n",
    "    return data\n",
    "\n",
    "def get_fields(host, dataset, samples, fields):\n",
    "    \"get field values\"\n",
    "    data = xena.dataset_fetch(host, dataset, samples, fields)\n",
    "    return data\n",
    "\n",
    "def get_fields_and_codes(host, dataset, samples, fields):\n",
    "    \"get fields and resolve codes\"\n",
    "    return get_codes( host, dataset, fields, get_fields( host, dataset, samples, fields))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Gene expression data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "hub = \"https://gdc.xenahubs.net\"\n",
    "dataset = \"TCGA-BRCA/Xena_Matrices/TCGA-BRCA.htseq_fpkm-uq.tsv\" #database url was found on the hub website\n",
    "samples = xena.dataset_samples (hub, dataset, None)\n",
    "#wave genes\n",
    "genes =[\"NCKAP1\",\"CYFIP2\", \"NCKAP1L\", \"WASF2\", \"ABI3\", \"WASF3\", \"ABI1\", \"ABI2\", \"CYFIP1\", \"WASF1\", \"BRK1\"] #enter the genes of interest\n",
    "#Arp2-3 genes\n",
    "#genes = [\"ACTR2\", \"ACTR3\", \"ARPC1A\", \"ARPC2\", \"ARPC3\", \"ARPC4\", \"ARPC5\", \"ACTR3B\", \"ARPC1B\", \"ARPC5L\"]\n",
    "#Arpin\n",
    "#genes = [\"Arpin\"]\n",
    "fields = [\"sampleID\"]\n",
    "expressions_0 = xena.dataset_gene_probe_avg(hub, dataset, samples, genes)\n",
    "\n",
    "values = get_fields_and_codes(hub, dataset, samples, fields) # list of lists\n",
    "\n",
    "genes_dict_for_df = dict(zip(fields, values)) #dict where we add the gene expression and the sample ids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "for dico in expressions_0:\n",
    "    label = \"expression\" + dico[\"gene\"]\n",
    "    scores = dico[\"scores\"][0]\n",
    "    assert len(scores) > 0\n",
    "    genes_dict_for_df[label] = scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Survival data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = \"TCGA-BRCA/Xena_Matrices/TCGA-BRCA.survival.tsv\"\n",
    "fields = ['_TIME_TO_EVENT', \"_EVENT\", \"sampleID\"] #event is one if the patient has died or 0 if he was censored (lost...)\n",
    "#TODO\n",
    "# As in pancan, there are normal samples in tcga which should probably be removed. _sample_type will\n",
    "# identify normals. _study will identify tcga vs. gtex vs. target.\n",
    "values = get_fields_and_codes(hub, dataset, samples, fields) # list of lists\n",
    "survival_dict_for_df = dict(zip(fields, values)) # index by phenotype"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "survival_dict_for_df[\"survival\"] = survival_dict_for_df.pop(\"_TIME_TO_EVENT\") \n",
    "survival_dict_for_df[\"event\"] = survival_dict_for_df.pop(\"_EVENT\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Phenotype data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = \"TCGA-BRCA/Xena_Matrices/TCGA-BRCA.GDC_phenotype.tsv\"\n",
    "fields = ['sample_type.samples', \"sampleID\"] \n",
    "values = get_fields_and_codes(hub, dataset, samples, fields) # list of lists\n",
    "phenotype_dict_for_df = dict(zip(fields, values)) # index by phenotype"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "phenotype_dict_for_df[\"sample_type\"] = phenotype_dict_for_df.pop('sample_type.samples')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note : I don't know if it's useful to remove the patients whose tumor is not a \"Primary Tumor\", like advised by Xena. The goal is to remove duplicates but I can't find duplicates"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Merge data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/anne/anaconda/envs/psc/lib/python3.7/site-packages/ipykernel_launcher.py:7: FutureWarning: convert_objects is deprecated.  To re-infer data dtypes for object columns, use DataFrame.infer_objects()\n",
      "For all other conversions use the data-type specific converters pd.to_datetime, pd.to_timedelta and pd.to_numeric.\n",
      "  import sys\n"
     ]
    }
   ],
   "source": [
    "df_genes = pd.DataFrame(data=genes_dict_for_df)\n",
    "df_survival = pd.DataFrame(data=survival_dict_for_df)\n",
    "#df_phenotype = pd.DataFrame(data=phenotype_dict_for_df)\n",
    "\n",
    "df = pd.merge(df_genes, df_survival) # merge the survival and genes data according to sampleID\n",
    "#df = pd.merge(df, df_phenotype) #merge also the phenotype data \n",
    "df = df.convert_objects(convert_numeric=True)\n",
    "df = df[~pd.isna(df[\"survival\"])][~pd.isna(df[\"event\"])]\n",
    "#df = df[df[\"sample_type\"] == \"Primary Tumor\"] #keep only primary tumor to remove duplicate WARNING : check if name doesn't change if you use another dataset\n",
    "#df = df.drop(['sampleID','sample_type'], axis=1) #remove identifier for analysis\n",
    "\n",
    "df = df.drop('sampleID', axis=1) #remove identifier for analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>expressionNCKAP1</th>\n",
       "      <th>expressionCYFIP2</th>\n",
       "      <th>expressionNCKAP1L</th>\n",
       "      <th>expressionWASF2</th>\n",
       "      <th>expressionABI3</th>\n",
       "      <th>expressionWASF3</th>\n",
       "      <th>expressionABI1</th>\n",
       "      <th>expressionABI2</th>\n",
       "      <th>expressionCYFIP1</th>\n",
       "      <th>expressionWASF1</th>\n",
       "      <th>expressionBRK1</th>\n",
       "      <th>survival</th>\n",
       "      <th>event</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>17.61</td>\n",
       "      <td>16.41</td>\n",
       "      <td>14.69</td>\n",
       "      <td>20.28</td>\n",
       "      <td>16.10</td>\n",
       "      <td>15.46</td>\n",
       "      <td>17.68</td>\n",
       "      <td>15.84</td>\n",
       "      <td>18.34</td>\n",
       "      <td>14.41</td>\n",
       "      <td>20.59</td>\n",
       "      <td>4047</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>17.15</td>\n",
       "      <td>15.33</td>\n",
       "      <td>16.06</td>\n",
       "      <td>18.06</td>\n",
       "      <td>17.74</td>\n",
       "      <td>15.77</td>\n",
       "      <td>18.12</td>\n",
       "      <td>15.85</td>\n",
       "      <td>17.89</td>\n",
       "      <td>15.20</td>\n",
       "      <td>21.15</td>\n",
       "      <td>4005</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>16.70</td>\n",
       "      <td>16.09</td>\n",
       "      <td>16.10</td>\n",
       "      <td>19.24</td>\n",
       "      <td>16.71</td>\n",
       "      <td>14.70</td>\n",
       "      <td>18.47</td>\n",
       "      <td>15.53</td>\n",
       "      <td>17.82</td>\n",
       "      <td>15.06</td>\n",
       "      <td>21.31</td>\n",
       "      <td>1474</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>17.27</td>\n",
       "      <td>14.59</td>\n",
       "      <td>15.38</td>\n",
       "      <td>19.12</td>\n",
       "      <td>16.70</td>\n",
       "      <td>14.51</td>\n",
       "      <td>18.12</td>\n",
       "      <td>16.11</td>\n",
       "      <td>18.43</td>\n",
       "      <td>16.05</td>\n",
       "      <td>21.59</td>\n",
       "      <td>1448</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>17.11</td>\n",
       "      <td>16.21</td>\n",
       "      <td>15.67</td>\n",
       "      <td>19.44</td>\n",
       "      <td>16.39</td>\n",
       "      <td>15.60</td>\n",
       "      <td>18.34</td>\n",
       "      <td>16.26</td>\n",
       "      <td>17.93</td>\n",
       "      <td>16.86</td>\n",
       "      <td>21.63</td>\n",
       "      <td>348</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   expressionNCKAP1  expressionCYFIP2  expressionNCKAP1L  expressionWASF2  \\\n",
       "0             17.61             16.41              14.69            20.28   \n",
       "1             17.15             15.33              16.06            18.06   \n",
       "2             16.70             16.09              16.10            19.24   \n",
       "3             17.27             14.59              15.38            19.12   \n",
       "4             17.11             16.21              15.67            19.44   \n",
       "\n",
       "   expressionABI3  expressionWASF3  expressionABI1  expressionABI2  \\\n",
       "0           16.10            15.46           17.68           15.84   \n",
       "1           17.74            15.77           18.12           15.85   \n",
       "2           16.71            14.70           18.47           15.53   \n",
       "3           16.70            14.51           18.12           16.11   \n",
       "4           16.39            15.60           18.34           16.26   \n",
       "\n",
       "   expressionCYFIP1  expressionWASF1  expressionBRK1  survival  event  \n",
       "0             18.34            14.41           20.59      4047      0  \n",
       "1             17.89            15.20           21.15      4005      0  \n",
       "2             17.82            15.06           21.31      1474      0  \n",
       "3             18.43            16.05           21.59      1448      0  \n",
       "4             17.93            16.86           21.63       348      0  "
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "   ## Export the data so we can use them with R"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.to_csv(\"data_base_Arpin\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analyse the data (using Lifelines)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a1a9390f0>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "kmf = KaplanMeierFitter()\n",
    "kmf.fit(df[\"survival\"], df[\"event\"]) #TODO : tell lifelines if the patient is censored or dead \n",
    "kmf.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<lifelines.CoxPHFitter: fitted with 1203 observations, 1003 censored>\n",
      "      duration col = 'survival'\n",
      "         event col = 'event'\n",
      "number of subjects = 1203\n",
      "  number of events = 200\n",
      "    log-likelihood = -1144.45\n",
      "  time fit was run = 2019-01-25 11:36:55 UTC\n",
      "\n",
      "---\n",
      "                   coef  exp(coef)  se(coef)     z    p  log(p)  lower 0.95  upper 0.95   \n",
      "expressionNCKAP1   0.44       1.55      0.20  2.22 0.03   -3.63        0.05        0.83  .\n",
      "expressionCYFIP2  -0.08       0.92      0.07 -1.26 0.21   -1.58       -0.21        0.05   \n",
      "expressionNCKAP1L -0.09       0.91      0.11 -0.85 0.40   -0.93       -0.30        0.12   \n",
      "expressionWASF2    0.08       1.08      0.14  0.55 0.58   -0.54       -0.20        0.35   \n",
      "expressionABI3     0.09       1.09      0.15  0.57 0.57   -0.57       -0.21        0.39   \n",
      "expressionWASF3    0.05       1.05      0.06  0.81 0.42   -0.87       -0.07        0.16   \n",
      "expressionABI1    -0.10       0.91      0.18 -0.56 0.58   -0.55       -0.44        0.25   \n",
      "expressionABI2    -0.22       0.80      0.16 -1.39 0.16   -1.81       -0.54        0.09   \n",
      "expressionCYFIP1   0.26       1.30      0.20  1.32 0.19   -1.68       -0.13        0.64   \n",
      "expressionWASF1    0.08       1.08      0.07  1.14 0.26   -1.36       -0.06        0.21   \n",
      "expressionBRK1     0.13       1.13      0.16  0.79 0.43   -0.84       -0.19        0.44   \n",
      "---\n",
      "Signif. codes: 0 '***' 0.0001 '**' 0.001 '*' 0.01 '.' 0.05 ' ' 1\n",
      "\n",
      "Concordance = 0.61\n",
      "Likelihood ratio test = 21.12 on 11 df, log(p)=-3.44\n"
     ]
    }
   ],
   "source": [
    "cph = CoxPHFitter()\n",
    "cph.fit(df, duration_col='survival', event_col='event')\n",
    "\n",
    "\n",
    "cph.print_summary() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a1bbf5048>"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cph.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a1bd1e400>"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cph.plot_covariate_groups('expressionNCKAP1', [0, 5, 10, 15])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The concordance and log-likelihood indicate that the model do not fit very well : let's check the assumptions for cox regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "ename": "IndexError",
     "evalue": "boolean index did not match indexed array along dimension 0; dimension is 1173 but corresponding boolean dimension is 1203",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mIndexError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-36-cfb1680b09c7>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mcph\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcheck_assumptions\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdf\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m~/anaconda/envs/psc/lib/python3.7/site-packages/lifelines/fitters/coxph_fitter.py\u001b[0m in \u001b[0;36mcheck_assumptions\u001b[0;34m(self, training_df, advice, show_plots, p_value_threshold, plot_n_bootstraps)\u001b[0m\n\u001b[1;32m   1534\u001b[0m         \u001b[0mhttp\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m//\u001b[0m\u001b[0meprints\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlse\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mac\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0muk\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0;36m84988\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;36m6\u001b[0m\u001b[0m_ParkHendry2015\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mReassessingSchoenfeldTests_Final\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpdf\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1535\u001b[0m         \"\"\"\n\u001b[0;32m-> 1536\u001b[0;31m         \u001b[0mresiduals\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcompute_residuals\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtraining_df\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkind\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m\"scaled_schoenfeld\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1537\u001b[0m         test_results = proportional_hazard_test(\n\u001b[1;32m   1538\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtraining_df\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtime_transform\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m\"all\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mprecomputed_residuals\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mresiduals\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda/envs/psc/lib/python3.7/site-packages/lifelines/fitters/coxph_fitter.py\u001b[0m in \u001b[0;36mcompute_residuals\u001b[0;34m(self, training_dataframe, kind)\u001b[0m\n\u001b[1;32m   1003\u001b[0m         \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mT\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mE\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweights\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mshuffled_original_index\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_preprocess_dataframe\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtraining_dataframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1004\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1005\u001b[0;31m         \u001b[0mresids\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgetattr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"_compute_%s\"\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0mkind\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mT\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mE\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweights\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mshuffled_original_index\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1006\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresids\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1007\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda/envs/psc/lib/python3.7/site-packages/lifelines/fitters/coxph_fitter.py\u001b[0m in \u001b[0;36m_compute_scaled_schoenfeld\u001b[0;34m(self, X, T, E, weights, index)\u001b[0m\n\u001b[1;32m    799\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    800\u001b[0m         \u001b[0mn_deaths\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msum\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mevent_observed\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 801\u001b[0;31m         scaled_schoenfeld_resids = n_deaths * self._compute_schoenfeld(X, T, E, weights, index).dot(\n\u001b[0m\u001b[1;32m    802\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvariance_matrix_\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    803\u001b[0m         )\n",
      "\u001b[0;32m~/anaconda/envs/psc/lib/python3.7/site-packages/lifelines/fitters/coxph_fitter.py\u001b[0m in \u001b[0;36m_compute_schoenfeld\u001b[0;34m(self, X, T, E, weights, index)\u001b[0m\n\u001b[1;32m    825\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    826\u001b[0m         \u001b[0;31m# schoenfeld residuals are only defined for subjects with a non-zero event.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 827\u001b[0;31m         \u001b[0mdf\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDataFrame\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mschoenfeld_residuals\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mE\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcolumns\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhazards_\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcolumns\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindex\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mE\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    828\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mdf\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    829\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mIndexError\u001b[0m: boolean index did not match indexed array along dimension 0; dimension is 1173 but corresponding boolean dimension is 1203"
     ]
    }
   ],
   "source": [
    "cph.check_assumptions(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# doesn't work so go to R"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## TODO : do we have to remove tumors that are not Primary tumors from the list ? make the same transformation than Marc Lavielle, check assumption for Cox Regression, use other form of gene expression (RNAseq) to compare, use other datasets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Liens : \n",
    "\n",
    "objectifs en cours : https://docs.google.com/document/d/1A5wxgQ_ZPry4aRH1BsmdbmPJfJoIsy5sa-4P5RmWaTs/edit\n",
    "\n",
    "\n",
    "dossier partagé : https://drive.google.com/drive/u/0/folders/1HotcSiF2BzDuzXMhKNX1p-7lI1U_3UQ9\n",
    "\n",
    "rapport : https://www.overleaf.com/project/5c34f6b7f99dfe3f36bf3a55\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generate data for report (formatting and renaming)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2 = pd.read_csv(\"data_for_report_arp23\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>coef</th>\n",
       "      <th>exp(coef)</th>\n",
       "      <th>se(coef)</th>\n",
       "      <th>z</th>\n",
       "      <th>Pr(&gt;|z|)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>expressionACTR2</td>\n",
       "      <td>-0.175868</td>\n",
       "      <td>0.838728</td>\n",
       "      <td>0.247238</td>\n",
       "      <td>-0.711333</td>\n",
       "      <td>0.476878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>expressionACTR3</td>\n",
       "      <td>0.111441</td>\n",
       "      <td>1.117888</td>\n",
       "      <td>0.266382</td>\n",
       "      <td>0.418351</td>\n",
       "      <td>0.675691</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>expressionARPC1A</td>\n",
       "      <td>0.106827</td>\n",
       "      <td>1.112742</td>\n",
       "      <td>0.188127</td>\n",
       "      <td>0.567846</td>\n",
       "      <td>0.570140</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>expressionARPC2</td>\n",
       "      <td>0.166429</td>\n",
       "      <td>1.181080</td>\n",
       "      <td>0.237830</td>\n",
       "      <td>0.699781</td>\n",
       "      <td>0.484064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>expressionARPC3</td>\n",
       "      <td>-0.220496</td>\n",
       "      <td>0.802121</td>\n",
       "      <td>0.177576</td>\n",
       "      <td>-1.241698</td>\n",
       "      <td>0.214348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>expressionARPC4</td>\n",
       "      <td>-0.127033</td>\n",
       "      <td>0.880705</td>\n",
       "      <td>0.219331</td>\n",
       "      <td>-0.579181</td>\n",
       "      <td>0.562467</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>expressionARPC5</td>\n",
       "      <td>0.137872</td>\n",
       "      <td>1.147829</td>\n",
       "      <td>0.186991</td>\n",
       "      <td>0.737321</td>\n",
       "      <td>0.460927</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>expressionACTR3B</td>\n",
       "      <td>-0.200473</td>\n",
       "      <td>0.818344</td>\n",
       "      <td>0.100017</td>\n",
       "      <td>-2.004386</td>\n",
       "      <td>0.045029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>expressionARPC1B</td>\n",
       "      <td>-0.246536</td>\n",
       "      <td>0.781503</td>\n",
       "      <td>0.114362</td>\n",
       "      <td>-2.155755</td>\n",
       "      <td>0.031103</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>expressionARPC5L</td>\n",
       "      <td>0.060699</td>\n",
       "      <td>1.062580</td>\n",
       "      <td>0.171384</td>\n",
       "      <td>0.354172</td>\n",
       "      <td>0.723210</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Unnamed: 0      coef  exp(coef)  se(coef)         z  Pr(>|z|)\n",
       "0   expressionACTR2 -0.175868   0.838728  0.247238 -0.711333  0.476878\n",
       "1   expressionACTR3  0.111441   1.117888  0.266382  0.418351  0.675691\n",
       "2  expressionARPC1A  0.106827   1.112742  0.188127  0.567846  0.570140\n",
       "3   expressionARPC2  0.166429   1.181080  0.237830  0.699781  0.484064\n",
       "4   expressionARPC3 -0.220496   0.802121  0.177576 -1.241698  0.214348\n",
       "5   expressionARPC4 -0.127033   0.880705  0.219331 -0.579181  0.562467\n",
       "6   expressionARPC5  0.137872   1.147829  0.186991  0.737321  0.460927\n",
       "7  expressionACTR3B -0.200473   0.818344  0.100017 -2.004386  0.045029\n",
       "8  expressionARPC1B -0.246536   0.781503  0.114362 -2.155755  0.031103\n",
       "9  expressionARPC5L  0.060699   1.062580  0.171384  0.354172  0.723210"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2 = df2[[\"Unnamed: 0\", \"coef\", \"Pr(>|z|)\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2 = df2.rename(columns={\"Unnamed: 0\": \"gene\", \"coef\": \"coefficient\", \"Pr(>|z|)\" : \"p-value\"})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2[\"gene\"] = df2[\"gene\"].apply(lambda x: x[10:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using matplotlib backend: MacOSX\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>gene</th>\n",
       "      <th>coefficient</th>\n",
       "      <th>p-value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ACTR2</td>\n",
       "      <td>-0.175868</td>\n",
       "      <td>0.476878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ACTR3</td>\n",
       "      <td>0.111441</td>\n",
       "      <td>0.675691</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ARPC1A</td>\n",
       "      <td>0.106827</td>\n",
       "      <td>0.570140</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ARPC2</td>\n",
       "      <td>0.166429</td>\n",
       "      <td>0.484064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ARPC3</td>\n",
       "      <td>-0.220496</td>\n",
       "      <td>0.214348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>ARPC4</td>\n",
       "      <td>-0.127033</td>\n",
       "      <td>0.562467</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>ARPC5</td>\n",
       "      <td>0.137872</td>\n",
       "      <td>0.460927</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>ACTR3B</td>\n",
       "      <td>-0.200473</td>\n",
       "      <td>0.045029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>ARPC1B</td>\n",
       "      <td>-0.246536</td>\n",
       "      <td>0.031103</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>ARPC5L</td>\n",
       "      <td>0.060699</td>\n",
       "      <td>0.723210</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     gene  coefficient   p-value\n",
       "0   ACTR2    -0.175868  0.476878\n",
       "1   ACTR3     0.111441  0.675691\n",
       "2  ARPC1A     0.106827  0.570140\n",
       "3   ARPC2     0.166429  0.484064\n",
       "4   ARPC3    -0.220496  0.214348\n",
       "5   ARPC4    -0.127033  0.562467\n",
       "6   ARPC5     0.137872  0.460927\n",
       "7  ACTR3B    -0.200473  0.045029\n",
       "8  ARPC1B    -0.246536  0.031103\n",
       "9  ARPC5L     0.060699  0.723210"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}