pipeline/misc/tokenization/protein-tokenization-analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Protein Tokenization Analysis\n",
    "\n",
    "This notebook will compare tokenization strategies used to classify T cell phenotype strings according to their known (but withheld) subtype."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import warnings\n",
    "import tqdm\n",
    "import plotnine as pn\n",
    "from gensim.models import KeyedVectors\n",
    "from tcre.env import *\n",
    "from tcre import integration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 2780 entries, 0 to 2779\n",
      "Data columns (total 8 columns):\n",
      "id                2780 non-null object\n",
      "start_wrd         2780 non-null int64\n",
      "end_wrd           2780 non-null int64\n",
      "text              2780 non-null object\n",
      "cell_type         2780 non-null object\n",
      "cell_type_text    2780 non-null object\n",
      "ptkn              2780 non-null object\n",
      "clean_text        2780 non-null object\n",
      "dtypes: int64(2), object(6)\n",
      "memory usage: 173.8+ KB\n"
     ]
    }
   ],
   "source": [
    "path = osp.join(RESULTS_DATA_DIR, 'protein-tokenization', 'dataset.pkl')\n",
    "df = pd.read_pickle(path)\n",
    "df = df.reset_index(drop=True)\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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>id</th>\n",
       "      <th>start_wrd</th>\n",
       "      <th>end_wrd</th>\n",
       "      <th>text</th>\n",
       "      <th>cell_type</th>\n",
       "      <th>cell_type_text</th>\n",
       "      <th>ptkn</th>\n",
       "      <th>clean_text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>668</th>\n",
       "      <td>PMC3904036</td>\n",
       "      <td>2395</td>\n",
       "      <td>2400</td>\n",
       "      <td>Th1 (CD4+IL-17-IFN-γ+) cells</td>\n",
       "      <td>Th1</td>\n",
       "      <td>Th1</td>\n",
       "      <td>[(, CD4⁺(+), IL-17⁻(-), IFN-γ⁺(+), ), cells]</td>\n",
       "      <td>(CD4+IL-17-IFN-γ+) cells</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             id  start_wrd  end_wrd                          text cell_type  \\\n",
       "668  PMC3904036       2395     2400  Th1 (CD4+IL-17-IFN-γ+) cells       Th1   \n",
       "\n",
       "    cell_type_text                                          ptkn  \\\n",
       "668            Th1  [(, CD4⁺(+), IL-17⁻(-), IFN-γ⁺(+), ), cells]   \n",
       "\n",
       "                   clean_text  \n",
       "668  (CD4+IL-17-IFN-γ+) cells  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['text'].str.contains('Th1 \\(CD4\\+IL-17-IFN-γ\\+\\) cells')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>id</th>\n",
       "      <th>start_wrd</th>\n",
       "      <th>end_wrd</th>\n",
       "      <th>text</th>\n",
       "      <th>cell_type</th>\n",
       "      <th>cell_type_text</th>\n",
       "      <th>ptkn</th>\n",
       "      <th>clean_text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>PMC1555646</td>\n",
       "      <td>196</td>\n",
       "      <td>200</td>\n",
       "      <td>human CD4+CD25+Foxp3+ Tregs</td>\n",
       "      <td>Treg</td>\n",
       "      <td>Tregs</td>\n",
       "      <td>[human, CD4⁺(+), CD25⁺(+), Foxp3⁺(+)]</td>\n",
       "      <td>human CD4+CD25+Foxp3+</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PMC1867380</td>\n",
       "      <td>5134</td>\n",
       "      <td>5141</td>\n",
       "      <td>CD4+CD25+FoxP3+ IL-10–producing natural T regu...</td>\n",
       "      <td>nTreg</td>\n",
       "      <td>natural T regulatory</td>\n",
       "      <td>[CD4⁺(+), CD25⁺(+), FoxP3⁺(+), IL-10⁻(–), prod...</td>\n",
       "      <td>CD4+CD25+FoxP3+ IL-10–producing  cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>PMC1939810</td>\n",
       "      <td>6249</td>\n",
       "      <td>6254</td>\n",
       "      <td>CD4+CD25highCD62L−Foxp3+ regulatory T cells</td>\n",
       "      <td>Treg</td>\n",
       "      <td>regulatory T</td>\n",
       "      <td>[CD4⁺(+), CD25⁺(high), CD62L⁻(−), Foxp3⁺(+), c...</td>\n",
       "      <td>CD4+CD25highCD62L−Foxp3+  cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>PMC1940266</td>\n",
       "      <td>816</td>\n",
       "      <td>821</td>\n",
       "      <td>CD4+CD25+Foxp3+ regulatory T cells</td>\n",
       "      <td>Treg</td>\n",
       "      <td>regulatory T</td>\n",
       "      <td>[CD4⁺(+), CD25⁺(+), Foxp3⁺(+), cells]</td>\n",
       "      <td>CD4+CD25+Foxp3+  cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>PMC1945037</td>\n",
       "      <td>247</td>\n",
       "      <td>254</td>\n",
       "      <td>CD4+ CD25high FoxP3high regulatory T cells</td>\n",
       "      <td>Treg</td>\n",
       "      <td>regulatory T</td>\n",
       "      <td>[CD4⁺(+), CD25⁺(high), FoxP3⁺(high), cells]</td>\n",
       "      <td>CD4+ CD25high FoxP3high  cells</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           id  start_wrd  end_wrd  \\\n",
       "0  PMC1555646        196      200   \n",
       "1  PMC1867380       5134     5141   \n",
       "2  PMC1939810       6249     6254   \n",
       "3  PMC1940266        816      821   \n",
       "4  PMC1945037        247      254   \n",
       "\n",
       "                                                text cell_type  \\\n",
       "0                        human CD4+CD25+Foxp3+ Tregs      Treg   \n",
       "1  CD4+CD25+FoxP3+ IL-10–producing natural T regu...     nTreg   \n",
       "2        CD4+CD25highCD62L−Foxp3+ regulatory T cells      Treg   \n",
       "3                 CD4+CD25+Foxp3+ regulatory T cells      Treg   \n",
       "4         CD4+ CD25high FoxP3high regulatory T cells      Treg   \n",
       "\n",
       "         cell_type_text                                               ptkn  \\\n",
       "0                 Tregs              [human, CD4⁺(+), CD25⁺(+), Foxp3⁺(+)]   \n",
       "1  natural T regulatory  [CD4⁺(+), CD25⁺(+), FoxP3⁺(+), IL-10⁻(–), prod...   \n",
       "2          regulatory T  [CD4⁺(+), CD25⁺(high), CD62L⁻(−), Foxp3⁺(+), c...   \n",
       "3          regulatory T              [CD4⁺(+), CD25⁺(+), Foxp3⁺(+), cells]   \n",
       "4          regulatory T        [CD4⁺(+), CD25⁺(high), FoxP3⁺(high), cells]   \n",
       "\n",
       "                               clean_text  \n",
       "0                   human CD4+CD25+Foxp3+  \n",
       "1  CD4+CD25+FoxP3+ IL-10–producing  cells  \n",
       "2         CD4+CD25highCD62L−Foxp3+  cells  \n",
       "3                  CD4+CD25+Foxp3+  cells  \n",
       "4          CD4+ CD25high FoxP3high  cells  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "with warnings.catch_warnings():\n",
    "    warnings.filterwarnings(\"ignore\", message='.*')\n",
    "    w2v = KeyedVectors.load_word2vec_format(W2V_MODEL_01, binary=True, limit=500000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(500000, 200)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "w2v.vectors.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "nlp = integration.get_scispacy_pipeline()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_features(df, w2v, nlp, use_ptkn=False):\n",
    "    \n",
    "    def get_tokens(r):\n",
    "        if use_ptkn:\n",
    "            tkns = [((t.token_text or '') + (t.sign_text or '')) for t in r['ptkn']]\n",
    "        else:\n",
    "            tkns = list(nlp(r['clean_text']))\n",
    "        return tkns\n",
    "            \n",
    "        \n",
    "    oov = np.zeros(w2v.vectors.shape[1], dtype=w2v.vectors.dtype)\n",
    "    def get_vector(w):\n",
    "        return w2v.get_vector(w) if w in w2v else oov\n",
    "        \n",
    "    V = pd.Series([\n",
    "        np.stack([get_vector(t) for t in get_tokens(r)])\n",
    "        for i, r in df.iterrows()\n",
    "    ], index=df.index)\n",
    "    X = pd.DataFrame([\n",
    "        V[i].sum(axis=0) for i in range(len(V))], \n",
    "        index=df.index)\n",
    "    y = df['cell_type']\n",
    "    return V, X, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "#tokenization = 'scispacy'\n",
    "tokenization = 'ptkn'\n",
    "V, X, y = get_features(df, w2v, nlp, use_ptkn=tokenization == 'ptkn')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((2780, 200), (2780,))"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "iTreg      10\n",
       "Tscm       13\n",
       "TEMRA      28\n",
       "MAIT       29\n",
       "Treg1      37\n",
       "γδT        39\n",
       "nTreg      40\n",
       "Th2        46\n",
       "TEM        71\n",
       "TCM        73\n",
       "NKT        75\n",
       "Th1        78\n",
       "Th17      144\n",
       "Tfh       352\n",
       "Treg     1745\n",
       "Name: cell_type, dtype: int64"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.value_counts().sort_values()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "from xgboost import XGBClassifier\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.linear_model import RidgeClassifierCV\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.dummy import DummyClassifier\n",
    "from sklearn.metrics import accuracy_score, f1_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "ests = {\n",
    "    'xgb': XGBClassifier(random_state=TCRE_SEED),\n",
    "    'gbr': GradientBoostingClassifier(random_state=TCRE_SEED),\n",
    "    'ridge': RidgeClassifierCV(normalize=True),\n",
    "    'mlp': Pipeline([\n",
    "        ('scale', StandardScaler()), \n",
    "        ('est', MLPClassifier(hidden_layer_sizes=(15,), random_state=TCRE_SEED, max_iter=2000))\n",
    "    ]),\n",
    "    'svc': Pipeline([\n",
    "        ('scale', StandardScaler()), \n",
    "        ('est', GridSearchCV(SVC(random_state=TCRE_SEED, gamma='auto'), {'C': [.01, .1, .5, 1., 5.]}))\n",
    "    ]),\n",
    "    'rnd-strat': DummyClassifier(strategy='stratified', random_state=TCRE_SEED),\n",
    "    'rnd-mfreq': DummyClassifier(strategy='most_frequent', random_state=TCRE_SEED),\n",
    "    'rnd-runif': DummyClassifier(strategy='uniform', random_state=TCRE_SEED),\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/10 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting xgb ...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "100%|██████████| 10/10 [04:29<00:00, 27.01s/it]\n",
      "  0%|          | 0/10 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done\n",
      "Fitting gbr ...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 10%|█         | 1/10 [00:32<04:52, 32.46s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 20%|██        | 2/10 [01:05<04:21, 32.67s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 30%|███       | 3/10 [01:38<03:48, 32.64s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 50%|█████     | 5/10 [02:45<02:45, 33.09s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 60%|██████    | 6/10 [03:18<02:12, 33.09s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 80%|████████  | 8/10 [04:24<01:06, 33.24s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 90%|█████████ | 9/10 [04:58<00:33, 33.18s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "100%|██████████| 10/10 [05:31<00:00, 33.18s/it]\n",
      "  0%|          | 0/10 [00:00<?, ?it/s]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 20%|██        | 2/10 [00:00<00:00, 15.35it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done\n",
      "Fitting ridge ...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 40%|████      | 4/10 [00:00<00:00, 15.14it/s]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 60%|██████    | 6/10 [00:00<00:00, 15.70it/s]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 80%|████████  | 8/10 [00:00<00:00, 16.01it/s]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "100%|██████████| 10/10 [00:00<00:00, 15.99it/s]\n",
      "  0%|          | 0/10 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done\n",
      "Fitting mlp ...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 10%|█         | 1/10 [00:03<00:35,  3.92s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 20%|██        | 2/10 [00:07<00:29,  3.73s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 30%|███       | 3/10 [00:11<00:27,  3.88s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 50%|█████     | 5/10 [00:19<00:18,  3.78s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 60%|██████    | 6/10 [00:23<00:15,  3.94s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 70%|███████   | 7/10 [00:27<00:11,  3.88s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 80%|████████  | 8/10 [00:30<00:07,  3.62s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 90%|█████████ | 9/10 [00:34<00:03,  3.72s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "100%|██████████| 10/10 [00:37<00:00,  3.68s/it]\n",
      "  0%|          | 0/10 [00:00<?, ?it/s]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done\n",
      "Fitting svc ...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 10%|█         | 1/10 [00:15<02:17, 15.28s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 20%|██        | 2/10 [00:30<02:02, 15.30s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 30%|███       | 3/10 [00:45<01:46, 15.27s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 40%|████      | 4/10 [01:01<01:31, 15.28s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 50%|█████     | 5/10 [01:16<01:16, 15.40s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 60%|██████    | 6/10 [01:32<01:01, 15.40s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 70%|███████   | 7/10 [01:47<00:46, 15.45s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 80%|████████  | 8/10 [02:03<00:31, 15.55s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      " 90%|█████████ | 9/10 [02:19<00:15, 15.57s/it]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/model_selection/_split.py:1978: FutureWarning: The default value of cv will change from 3 to 5 in version 0.22. Specify it explicitly to silence this warning.\n",
      "  warnings.warn(CV_WARNING, FutureWarning)\n",
      "/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "100%|██████████| 10/10 [02:34<00:00, 15.57s/it]\n",
      "  0%|          | 0/10 [00:00<?, ?it/s]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "100%|██████████| 10/10 [00:00<00:00, 102.62it/s]\n",
      "  0%|          | 0/10 [00:00<?, ?it/s]/opt/conda/envs/nlp/lib/python3.6/site-packages/sklearn/metrics/classification.py:1437: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done\n",
      "Fitting rnd-strat ...\n",
      "Done\n",
      "Fitting rnd-mfreq ...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10/10 [00:00<00:00, 93.71it/s]\n",
      "100%|██████████| 10/10 [00:00<00:00, 98.53it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done\n",
      "Fitting rnd-runif ...\n",
      "Done\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\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>accuracy</th>\n",
       "      <th>est</th>\n",
       "      <th>est_name</th>\n",
       "      <th>fold</th>\n",
       "      <th>macro-f1</th>\n",
       "      <th>micro-f1</th>\n",
       "      <th>y_pred</th>\n",
       "      <th>y_true</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.855634</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>0</td>\n",
       "      <td>0.584804</td>\n",
       "      <td>0.855634</td>\n",
       "      <td>[Treg, Treg, Treg, Th17, Treg, Treg, Treg, Tre...</td>\n",
       "      <td>5        Treg\n",
       "6        Treg\n",
       "21       Treg\n",
       "28  ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.833922</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>1</td>\n",
       "      <td>0.608914</td>\n",
       "      <td>0.833922</td>\n",
       "      <td>[Th2, Treg, Treg, Treg, Treg, Treg, Treg, Treg...</td>\n",
       "      <td>10        Th2\n",
       "24       Treg\n",
       "30       Treg\n",
       "41  ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.854610</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>2</td>\n",
       "      <td>0.490487</td>\n",
       "      <td>0.854610</td>\n",
       "      <td>[nTreg, Treg, Treg, Treg, TEM, Tfh, Treg, Treg...</td>\n",
       "      <td>17       Treg\n",
       "31       Treg\n",
       "32        NKT\n",
       "35  ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.803571</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>3</td>\n",
       "      <td>0.436746</td>\n",
       "      <td>0.803571</td>\n",
       "      <td>[Treg, Treg, Th2, Treg, Treg, Treg, MAIT, γδT,...</td>\n",
       "      <td>0        Treg\n",
       "2        Treg\n",
       "33       Th17\n",
       "38  ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.845878</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>4</td>\n",
       "      <td>0.622824</td>\n",
       "      <td>0.845878</td>\n",
       "      <td>[Treg, Treg, Tfh, Th17, Treg, Treg, Treg, Treg...</td>\n",
       "      <td>13      nTreg\n",
       "44       Treg\n",
       "47        Th1\n",
       "53  ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   accuracy                                                est est_name  fold  \\\n",
       "0  0.855634  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     0   \n",
       "1  0.833922  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     1   \n",
       "2  0.854610  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     2   \n",
       "3  0.803571  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     3   \n",
       "4  0.845878  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     4   \n",
       "\n",
       "   macro-f1  micro-f1                                             y_pred  \\\n",
       "0  0.584804  0.855634  [Treg, Treg, Treg, Th17, Treg, Treg, Treg, Tre...   \n",
       "1  0.608914  0.833922  [Th2, Treg, Treg, Treg, Treg, Treg, Treg, Treg...   \n",
       "2  0.490487  0.854610  [nTreg, Treg, Treg, Treg, TEM, Tfh, Treg, Treg...   \n",
       "3  0.436746  0.803571  [Treg, Treg, Th2, Treg, Treg, Treg, MAIT, γδT,...   \n",
       "4  0.622824  0.845878  [Treg, Treg, Tfh, Th17, Treg, Treg, Treg, Treg...   \n",
       "\n",
       "                                              y_true  \n",
       "0  5        Treg\n",
       "6        Treg\n",
       "21       Treg\n",
       "28  ...  \n",
       "1  10        Th2\n",
       "24       Treg\n",
       "30       Treg\n",
       "41  ...  \n",
       "2  17       Treg\n",
       "31       Treg\n",
       "32        NKT\n",
       "35  ...  \n",
       "3  0        Treg\n",
       "2        Treg\n",
       "33       Th17\n",
       "38  ...  \n",
       "4  13      nTreg\n",
       "44       Treg\n",
       "47        Th1\n",
       "53  ...  "
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfr = []\n",
    "\n",
    "def get_scores(y_true, y_pred):\n",
    "    return {\n",
    "        'micro-f1': f1_score(y_true, y_pred, average='micro'),\n",
    "        'macro-f1': f1_score(y_true, y_pred, average='macro'),\n",
    "        'accuracy': accuracy_score(y_true, y_pred)\n",
    "    }\n",
    "    \n",
    "for k in ests.keys():\n",
    "    print(f'Fitting {k} ...')\n",
    "    cv = StratifiedKFold(n_splits=10, shuffle=True, random_state=TCRE_SEED)\n",
    "    for i, (train, test) in tqdm.tqdm(list(enumerate(cv.split(X, y)))):\n",
    "        X_train, y_train = X.iloc[train], y.iloc[train]\n",
    "        X_test, y_test = X.iloc[test], y.iloc[test]\n",
    "        est = ests[k]\n",
    "        est.fit(X_train, y_train)\n",
    "        y_pred = est.predict(X_test)\n",
    "        dfr.append({**dict(est=est, est_name=k, fold=i, y_true=y_test, y_pred=y_pred), **get_scores(y_test, y_pred)})\n",
    "    print('Done')\n",
    "\n",
    "dfr = pd.DataFrame(dfr)\n",
    "dfr.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Export"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/lab/data/results/protein-tokenization/modeling-results-ptkn.pkl'"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "path = osp.join(RESULTS_DATA_DIR, 'protein-tokenization', 'modeling-results-{}.pkl'.format(tokenization))\n",
    "dfr.to_pickle(path)\n",
    "path"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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>accuracy</th>\n",
       "      <th>est</th>\n",
       "      <th>est_name</th>\n",
       "      <th>fold</th>\n",
       "      <th>macro-f1</th>\n",
       "      <th>micro-f1</th>\n",
       "      <th>y_pred</th>\n",
       "      <th>y_true</th>\n",
       "      <th>tokenization</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.855634</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>0</td>\n",
       "      <td>0.584804</td>\n",
       "      <td>0.855634</td>\n",
       "      <td>[Treg, Treg, Treg, Th17, Treg, Treg, Treg, Tre...</td>\n",
       "      <td>5        Treg\n",
       "6        Treg\n",
       "21       Treg\n",
       "28  ...</td>\n",
       "      <td>ptkn</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.833922</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>1</td>\n",
       "      <td>0.608914</td>\n",
       "      <td>0.833922</td>\n",
       "      <td>[Th2, Treg, Treg, Treg, Treg, Treg, Treg, Treg...</td>\n",
       "      <td>10        Th2\n",
       "24       Treg\n",
       "30       Treg\n",
       "41  ...</td>\n",
       "      <td>ptkn</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.854610</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>2</td>\n",
       "      <td>0.490487</td>\n",
       "      <td>0.854610</td>\n",
       "      <td>[nTreg, Treg, Treg, Treg, TEM, Tfh, Treg, Treg...</td>\n",
       "      <td>17       Treg\n",
       "31       Treg\n",
       "32        NKT\n",
       "35  ...</td>\n",
       "      <td>ptkn</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.803571</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>3</td>\n",
       "      <td>0.436746</td>\n",
       "      <td>0.803571</td>\n",
       "      <td>[Treg, Treg, Th2, Treg, Treg, Treg, MAIT, γδT,...</td>\n",
       "      <td>0        Treg\n",
       "2        Treg\n",
       "33       Th17\n",
       "38  ...</td>\n",
       "      <td>ptkn</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.845878</td>\n",
       "      <td>XGBClassifier(base_score=0.5, booster='gbtree'...</td>\n",
       "      <td>xgb</td>\n",
       "      <td>4</td>\n",
       "      <td>0.622824</td>\n",
       "      <td>0.845878</td>\n",
       "      <td>[Treg, Treg, Tfh, Th17, Treg, Treg, Treg, Treg...</td>\n",
       "      <td>13      nTreg\n",
       "44       Treg\n",
       "47        Th1\n",
       "53  ...</td>\n",
       "      <td>ptkn</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   accuracy                                                est est_name  fold  \\\n",
       "0  0.855634  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     0   \n",
       "1  0.833922  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     1   \n",
       "2  0.854610  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     2   \n",
       "3  0.803571  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     3   \n",
       "4  0.845878  XGBClassifier(base_score=0.5, booster='gbtree'...      xgb     4   \n",
       "\n",
       "   macro-f1  micro-f1                                             y_pred  \\\n",
       "0  0.584804  0.855634  [Treg, Treg, Treg, Th17, Treg, Treg, Treg, Tre...   \n",
       "1  0.608914  0.833922  [Th2, Treg, Treg, Treg, Treg, Treg, Treg, Treg...   \n",
       "2  0.490487  0.854610  [nTreg, Treg, Treg, Treg, TEM, Tfh, Treg, Treg...   \n",
       "3  0.436746  0.803571  [Treg, Treg, Th2, Treg, Treg, Treg, MAIT, γδT,...   \n",
       "4  0.622824  0.845878  [Treg, Treg, Tfh, Th17, Treg, Treg, Treg, Treg...   \n",
       "\n",
       "                                              y_true tokenization  \n",
       "0  5        Treg\n",
       "6        Treg\n",
       "21       Treg\n",
       "28  ...         ptkn  \n",
       "1  10        Th2\n",
       "24       Treg\n",
       "30       Treg\n",
       "41  ...         ptkn  \n",
       "2  17       Treg\n",
       "31       Treg\n",
       "32        NKT\n",
       "35  ...         ptkn  \n",
       "3  0        Treg\n",
       "2        Treg\n",
       "33       Th17\n",
       "38  ...         ptkn  \n",
       "4  13      nTreg\n",
       "44       Treg\n",
       "47        Th1\n",
       "53  ...         ptkn  "
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfr = pd.concat([\n",
    "    pd.read_pickle(osp.join(RESULTS_DATA_DIR, 'protein-tokenization', 'modeling-results-{}.pkl'.format(t)))\n",
    "    .assign(tokenization=t)\n",
    "    for t in ['ptkn', 'scispacy']\n",
    "]).reset_index(drop=True)\n",
    "dfr.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>tokenization</th>\n",
       "      <th>est_name</th>\n",
       "      <th>fold</th>\n",
       "      <th>metric</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ptkn</td>\n",
       "      <td>xgb</td>\n",
       "      <td>0</td>\n",
       "      <td>accuracy</td>\n",
       "      <td>0.855634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ptkn</td>\n",
       "      <td>xgb</td>\n",
       "      <td>0</td>\n",
       "      <td>macro-f1</td>\n",
       "      <td>0.584804</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ptkn</td>\n",
       "      <td>xgb</td>\n",
       "      <td>0</td>\n",
       "      <td>micro-f1</td>\n",
       "      <td>0.855634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ptkn</td>\n",
       "      <td>xgb</td>\n",
       "      <td>1</td>\n",
       "      <td>accuracy</td>\n",
       "      <td>0.833922</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ptkn</td>\n",
       "      <td>xgb</td>\n",
       "      <td>1</td>\n",
       "      <td>macro-f1</td>\n",
       "      <td>0.608914</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  tokenization est_name  fold    metric     value\n",
       "0         ptkn      xgb     0  accuracy  0.855634\n",
       "1         ptkn      xgb     0  macro-f1  0.584804\n",
       "2         ptkn      xgb     0  micro-f1  0.855634\n",
       "3         ptkn      xgb     1  accuracy  0.833922\n",
       "4         ptkn      xgb     1  macro-f1  0.608914"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfs = (\n",
    "    dfr\n",
    "    .drop(['est', 'y_pred', 'y_true'], axis=1)\n",
    "    .set_index(['tokenization', 'est_name', 'fold'])\n",
    "    .rename_axis('metric', axis='columns').stack().rename('value')\n",
    "    .reset_index()\n",
    ")\n",
    "dfs.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1600x500 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<ggplot: (-9223363274526249819)>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(\n",
    "    pn.ggplot(\n",
    "        dfs\n",
    "            .assign(est_name=lambda df: pd.Categorical(\n",
    "                df['est_name'], ordered=True, \n",
    "                categories=df.groupby(['est_name', 'metric'])['value'].mean().unstack()['micro-f1'].sort_values().index\n",
    "            ))\n",
    "        , \n",
    "        pn.aes(x='est_name', y='value')\n",
    "    ) + \n",
    "    pn.geom_jitter(random_state=1, width=.15) + \n",
    "    pn.facet_grid('metric ~ tokenization', scales='free') + \n",
    "    pn.theme(figure_size=(16, 5), axis_text_x=pn.element_text(rotation=45, hjust=1))\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1200x500 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<ggplot: (-9223363274484486333)>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def plot_metrics(metrics):\n",
    "    return (\n",
    "        pn.ggplot(\n",
    "            dfs\n",
    "                .pipe(lambda df: df[df['metric'].isin(metrics)])\n",
    "                #.pipe(lambda df: df[df['tokenization'].isin([tokenizations])])\n",
    "                .groupby(['tokenization', 'est_name', 'metric'])['value'].mean().rename('value').reset_index()\n",
    "                .assign(value=lambda df: 100*df['value'])\n",
    "                .assign(text=lambda df: df['value'].map('{:.1f}%'.format))\n",
    "                .assign(est_name=lambda df: pd.Categorical(\n",
    "                    df['est_name'], ordered=True, \n",
    "                    categories=df.groupby(['est_name', 'metric'])['value'].mean().unstack()[metrics[0]].sort_values().index\n",
    "                ))\n",
    "            , \n",
    "            pn.aes(x='est_name', y='value', label='text')\n",
    "        ) + \n",
    "        pn.geom_bar(stat='identity', position=pn.position_dodge(width=1)) +\n",
    "        pn.geom_text(nudge_y=2, size=8) + \n",
    "        pn.facet_grid('metric ~ tokenization', scales='free') + \n",
    "        pn.labs(x='Estimator', y='Score (%)', title='Classification Scores by Tokenization Method') + \n",
    "        pn.theme_bw() + \n",
    "        pn.theme(figure_size=(12, 5), axis_text_x=pn.element_text(rotation=45, hjust=1))\n",
    "    )\n",
    "plot_metrics(['accuracy', 'macro-f1'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "best_est = 'svc'\n",
    "best_tkn = 'ptkn'\n",
    "dft = dfr.pipe(lambda df: df[df['est_name'] == best_est]).pipe(lambda df: df[df['tokenization'] == best_tkn])\n",
    "est = dft['est'].iloc[0]\n",
    "y_pred = np.concatenate([v for v in dft['y_pred']])\n",
    "y_true = np.concatenate([v for v in dft['y_true']])\n",
    "assert y_pred.shape == y_true.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "        MAIT       0.69      0.62      0.65        29\n",
      "         NKT       0.84      0.63      0.72        75\n",
      "         TCM       0.71      0.66      0.68        73\n",
      "         TEM       0.50      0.48      0.49        71\n",
      "       TEMRA       0.74      0.50      0.60        28\n",
      "         Tfh       0.79      0.89      0.84       352\n",
      "         Th1       0.65      0.42      0.51        78\n",
      "        Th17       0.82      0.71      0.76       144\n",
      "         Th2       0.88      0.61      0.72        46\n",
      "        Treg       0.91      0.97      0.94      1745\n",
      "       Treg1       0.77      0.73      0.75        37\n",
      "        Tscm       0.75      0.46      0.57        13\n",
      "       iTreg       0.75      0.30      0.43        10\n",
      "       nTreg       0.40      0.10      0.16        40\n",
      "         γδT       0.74      0.44      0.55        39\n",
      "\n",
      "    accuracy                           0.86      2780\n",
      "   macro avg       0.73      0.57      0.62      2780\n",
      "weighted avg       0.85      0.86      0.85      2780\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_true, y_pred, labels=est.classes_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style  type=\"text/css\" >\n",
       "    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col0 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col1 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col2 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col3 {\n",
       "            background-color:  #ede8f3;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col5 {\n",
       "            background-color:  #fef6fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col6 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col7 {\n",
       "            background-color:  #fef6fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col9 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col14 {\n",
       "            background-color:  #f6eff7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col0 {\n",
       "            background-color:  #f7f0f7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col1 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col2 {\n",
       "            background-color:  #f9f2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col3 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col5 {\n",
       "            background-color:  #faf3f9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col6 {\n",
       "            background-color:  #fbf3f9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col7 {\n",
       "            background-color:  #f9f2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col8 {\n",
       "            background-color:  #faf2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col9 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col14 {\n",
       "            background-color:  #f6eff7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col1 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col2 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col3 {\n",
       "            background-color:  #c1cae2;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col5 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col6 {\n",
       "            background-color:  #fbf3f9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col7 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col9 {\n",
       "            background-color:  #fef6fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col11 {\n",
       "            background-color:  #e3e0ee;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col0 {\n",
       "            background-color:  #eee9f3;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col1 {\n",
       "            background-color:  #f9f2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col2 {\n",
       "            background-color:  #e3e0ee;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col3 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col4 {\n",
       "            background-color:  #d9d8ea;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col5 {\n",
       "            background-color:  #fdf5fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col6 {\n",
       "            background-color:  #fbf3f9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col7 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col9 {\n",
       "            background-color:  #fef6fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col13 {\n",
       "            background-color:  #73a9cf;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col14 {\n",
       "            background-color:  #e0deed;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col1 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col2 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col3 {\n",
       "            background-color:  #ede8f3;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col4 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col5 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col6 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col7 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col9 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col11 {\n",
       "            background-color:  #e3e0ee;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col13 {\n",
       "            background-color:  #d0d1e6;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col1 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col2 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col3 {\n",
       "            background-color:  #e7e3f0;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col5 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col6 {\n",
       "            background-color:  #d2d2e7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col7 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col8 {\n",
       "            background-color:  #faf2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col9 {\n",
       "            background-color:  #fdf5fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col1 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col2 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col3 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col5 {\n",
       "            background-color:  #f3edf5;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col6 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col7 {\n",
       "            background-color:  #fbf3f9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col8 {\n",
       "            background-color:  #faf2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col9 {\n",
       "            background-color:  #fef6fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col1 {\n",
       "            background-color:  #f9f2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col2 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col3 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col5 {\n",
       "            background-color:  #f7f0f7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col6 {\n",
       "            background-color:  #f1ebf5;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col7 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col8 {\n",
       "            background-color:  #faf2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col9 {\n",
       "            background-color:  #fef6fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col14 {\n",
       "            background-color:  #f6eff7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col1 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col2 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col3 {\n",
       "            background-color:  #f6eff7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col5 {\n",
       "            background-color:  #fef6fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col6 {\n",
       "            background-color:  #f6eff7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col7 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col8 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col9 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col0 {\n",
       "            background-color:  #f7f0f7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col1 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col2 {\n",
       "            background-color:  #e7e3f0;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col3 {\n",
       "            background-color:  #e0deed;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col4 {\n",
       "            background-color:  #f4eef6;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col5 {\n",
       "            background-color:  #f7f0f7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col6 {\n",
       "            background-color:  #fbf3f9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col7 {\n",
       "            background-color:  #f9f2f8;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col9 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col10 {\n",
       "            background-color:  #c1cae2;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col12 {\n",
       "            background-color:  #b4c4df;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col13 {\n",
       "            background-color:  #056faf;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col1 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col2 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col3 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col5 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col6 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col7 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col9 {\n",
       "            background-color:  #fef6fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col10 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col1 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col2 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col3 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col5 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col6 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col7 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col9 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col11 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col1 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col2 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col3 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col5 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col6 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col7 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col9 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col12 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col0 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col1 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col2 {\n",
       "            background-color:  #fcf4fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col3 {\n",
       "            background-color:  #fbf3f9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col4 {\n",
       "            background-color:  #f4eef6;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col5 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col6 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col7 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col9 {\n",
       "            background-color:  #fdf5fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col13 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col14 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col0 {\n",
       "            background-color:  #d7d6e9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col1 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col2 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col3 {\n",
       "            background-color:  #f6eff7;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col4 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col5 {\n",
       "            background-color:  #fef6fa;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col6 {\n",
       "            background-color:  #fbf3f9;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col7 {\n",
       "            background-color:  #f3edf5;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col8 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col9 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col10 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col11 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col12 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col13 {\n",
       "            background-color:  #fff7fb;\n",
       "            color:  #000000;\n",
       "        }    #T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col14 {\n",
       "            background-color:  #023858;\n",
       "            color:  #f1f1f1;\n",
       "        }</style><table id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002\" ><thead>    <tr>        <th class=\"blank level0\" ></th>        <th class=\"col_heading level0 col0\" >MAIT</th>        <th class=\"col_heading level0 col1\" >NKT</th>        <th class=\"col_heading level0 col2\" >TCM</th>        <th class=\"col_heading level0 col3\" >TEM</th>        <th class=\"col_heading level0 col4\" >TEMRA</th>        <th class=\"col_heading level0 col5\" >Tfh</th>        <th class=\"col_heading level0 col6\" >Th1</th>        <th class=\"col_heading level0 col7\" >Th17</th>        <th class=\"col_heading level0 col8\" >Th2</th>        <th class=\"col_heading level0 col9\" >Treg</th>        <th class=\"col_heading level0 col10\" >Treg1</th>        <th class=\"col_heading level0 col11\" >Tscm</th>        <th class=\"col_heading level0 col12\" >iTreg</th>        <th class=\"col_heading level0 col13\" >nTreg</th>        <th class=\"col_heading level0 col14\" >γδT</th>    </tr></thead><tbody>\n",
       "                <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row0\" class=\"row_heading level0 row0\" >MAIT</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col0\" class=\"data row0 col0\" >18</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col1\" class=\"data row0 col1\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col2\" class=\"data row0 col2\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col3\" class=\"data row0 col3\" >4</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col4\" class=\"data row0 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col5\" class=\"data row0 col5\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col6\" class=\"data row0 col6\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col7\" class=\"data row0 col7\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col8\" class=\"data row0 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col9\" class=\"data row0 col9\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col10\" class=\"data row0 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col11\" class=\"data row0 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col12\" class=\"data row0 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col13\" class=\"data row0 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row0_col14\" class=\"data row0 col14\" >1</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row1\" class=\"row_heading level0 row1\" >NKT</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col0\" class=\"data row1 col0\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col1\" class=\"data row1 col1\" >47</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col2\" class=\"data row1 col2\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col3\" class=\"data row1 col3\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col4\" class=\"data row1 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col5\" class=\"data row1 col5\" >10</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col6\" class=\"data row1 col6\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col7\" class=\"data row1 col7\" >4</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col8\" class=\"data row1 col8\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col9\" class=\"data row1 col9\" >8</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col10\" class=\"data row1 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col11\" class=\"data row1 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col12\" class=\"data row1 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col13\" class=\"data row1 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row1_col14\" class=\"data row1 col14\" >1</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row2\" class=\"row_heading level0 row2\" >TCM</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col0\" class=\"data row2 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col1\" class=\"data row2 col1\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col2\" class=\"data row2 col2\" >48</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col3\" class=\"data row2 col3\" >10</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col4\" class=\"data row2 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col5\" class=\"data row2 col5\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col6\" class=\"data row2 col6\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col7\" class=\"data row2 col7\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col8\" class=\"data row2 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col9\" class=\"data row2 col9\" >12</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col10\" class=\"data row2 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col11\" class=\"data row2 col11\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col12\" class=\"data row2 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col13\" class=\"data row2 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row2_col14\" class=\"data row2 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row3\" class=\"row_heading level0 row3\" >TEM</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col0\" class=\"data row3 col0\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col1\" class=\"data row3 col1\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col2\" class=\"data row3 col2\" >8</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col3\" class=\"data row3 col3\" >34</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col4\" class=\"data row3 col4\" >3</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col5\" class=\"data row3 col5\" >5</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col6\" class=\"data row3 col6\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col7\" class=\"data row3 col7\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col8\" class=\"data row3 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col9\" class=\"data row3 col9\" >11</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col10\" class=\"data row3 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col11\" class=\"data row3 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col12\" class=\"data row3 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col13\" class=\"data row3 col13\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row3_col14\" class=\"data row3 col14\" >3</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row4\" class=\"row_heading level0 row4\" >TEMRA</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col0\" class=\"data row4 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col1\" class=\"data row4 col1\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col2\" class=\"data row4 col2\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col3\" class=\"data row4 col3\" >4</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col4\" class=\"data row4 col4\" >14</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col5\" class=\"data row4 col5\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col6\" class=\"data row4 col6\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col7\" class=\"data row4 col7\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col8\" class=\"data row4 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col9\" class=\"data row4 col9\" >7</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col10\" class=\"data row4 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col11\" class=\"data row4 col11\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col12\" class=\"data row4 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col13\" class=\"data row4 col13\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row4_col14\" class=\"data row4 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row5\" class=\"row_heading level0 row5\" >Tfh</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col0\" class=\"data row5 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col1\" class=\"data row5 col1\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col2\" class=\"data row5 col2\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col3\" class=\"data row5 col3\" >5</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col4\" class=\"data row5 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col5\" class=\"data row5 col5\" >314</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col6\" class=\"data row5 col6\" >8</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col7\" class=\"data row5 col7\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col8\" class=\"data row5 col8\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col9\" class=\"data row5 col9\" >23</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col10\" class=\"data row5 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col11\" class=\"data row5 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col12\" class=\"data row5 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col13\" class=\"data row5 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row5_col14\" class=\"data row5 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row6\" class=\"row_heading level0 row6\" >Th1</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col0\" class=\"data row6 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col1\" class=\"data row6 col1\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col2\" class=\"data row6 col2\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col3\" class=\"data row6 col3\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col4\" class=\"data row6 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col5\" class=\"data row6 col5\" >25</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col6\" class=\"data row6 col6\" >33</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col7\" class=\"data row6 col7\" >3</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col8\" class=\"data row6 col8\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col9\" class=\"data row6 col9\" >15</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col10\" class=\"data row6 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col11\" class=\"data row6 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col12\" class=\"data row6 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col13\" class=\"data row6 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row6_col14\" class=\"data row6 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row7\" class=\"row_heading level0 row7\" >Th17</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col0\" class=\"data row7 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col1\" class=\"data row7 col1\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col2\" class=\"data row7 col2\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col3\" class=\"data row7 col3\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col4\" class=\"data row7 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col5\" class=\"data row7 col5\" >16</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col6\" class=\"data row7 col6\" >3</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col7\" class=\"data row7 col7\" >102</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col8\" class=\"data row7 col8\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col9\" class=\"data row7 col9\" >19</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col10\" class=\"data row7 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col11\" class=\"data row7 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col12\" class=\"data row7 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col13\" class=\"data row7 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row7_col14\" class=\"data row7 col14\" >1</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row8\" class=\"row_heading level0 row8\" >Th2</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col0\" class=\"data row8 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col1\" class=\"data row8 col1\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col2\" class=\"data row8 col2\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col3\" class=\"data row8 col3\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col4\" class=\"data row8 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col5\" class=\"data row8 col5\" >3</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col6\" class=\"data row8 col6\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col7\" class=\"data row8 col7\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col8\" class=\"data row8 col8\" >28</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col9\" class=\"data row8 col9\" >8</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col10\" class=\"data row8 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col11\" class=\"data row8 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col12\" class=\"data row8 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col13\" class=\"data row8 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row8_col14\" class=\"data row8 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row9\" class=\"row_heading level0 row9\" >Treg</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col0\" class=\"data row9 col0\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col1\" class=\"data row9 col1\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col2\" class=\"data row9 col2\" >7</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col3\" class=\"data row9 col3\" >6</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col4\" class=\"data row9 col4\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col5\" class=\"data row9 col5\" >16</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col6\" class=\"data row9 col6\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col7\" class=\"data row9 col7\" >4</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col8\" class=\"data row9 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col9\" class=\"data row9 col9\" >1696</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col10\" class=\"data row9 col10\" >8</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col11\" class=\"data row9 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col12\" class=\"data row9 col12\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col13\" class=\"data row9 col13\" >3</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row9_col14\" class=\"data row9 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row10\" class=\"row_heading level0 row10\" >Treg1</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col0\" class=\"data row10 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col1\" class=\"data row10 col1\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col2\" class=\"data row10 col2\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col3\" class=\"data row10 col3\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col4\" class=\"data row10 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col5\" class=\"data row10 col5\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col6\" class=\"data row10 col6\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col7\" class=\"data row10 col7\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col8\" class=\"data row10 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col9\" class=\"data row10 col9\" >9</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col10\" class=\"data row10 col10\" >27</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col11\" class=\"data row10 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col12\" class=\"data row10 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col13\" class=\"data row10 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row10_col14\" class=\"data row10 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row11\" class=\"row_heading level0 row11\" >Tscm</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col0\" class=\"data row11 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col1\" class=\"data row11 col1\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col2\" class=\"data row11 col2\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col3\" class=\"data row11 col3\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col4\" class=\"data row11 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col5\" class=\"data row11 col5\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col6\" class=\"data row11 col6\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col7\" class=\"data row11 col7\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col8\" class=\"data row11 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col9\" class=\"data row11 col9\" >6</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col10\" class=\"data row11 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col11\" class=\"data row11 col11\" >6</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col12\" class=\"data row11 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col13\" class=\"data row11 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row11_col14\" class=\"data row11 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row12\" class=\"row_heading level0 row12\" >iTreg</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col0\" class=\"data row12 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col1\" class=\"data row12 col1\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col2\" class=\"data row12 col2\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col3\" class=\"data row12 col3\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col4\" class=\"data row12 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col5\" class=\"data row12 col5\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col6\" class=\"data row12 col6\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col7\" class=\"data row12 col7\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col8\" class=\"data row12 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col9\" class=\"data row12 col9\" >7</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col10\" class=\"data row12 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col11\" class=\"data row12 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col12\" class=\"data row12 col12\" >3</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col13\" class=\"data row12 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row12_col14\" class=\"data row12 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row13\" class=\"row_heading level0 row13\" >nTreg</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col0\" class=\"data row13 col0\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col1\" class=\"data row13 col1\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col2\" class=\"data row13 col2\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col3\" class=\"data row13 col3\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col4\" class=\"data row13 col4\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col5\" class=\"data row13 col5\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col6\" class=\"data row13 col6\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col7\" class=\"data row13 col7\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col8\" class=\"data row13 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col9\" class=\"data row13 col9\" >33</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col10\" class=\"data row13 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col11\" class=\"data row13 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col12\" class=\"data row13 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col13\" class=\"data row13 col13\" >4</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row13_col14\" class=\"data row13 col14\" >0</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002level0_row14\" class=\"row_heading level0 row14\" >γδT</th>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col0\" class=\"data row14 col0\" >4</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col1\" class=\"data row14 col1\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col2\" class=\"data row14 col2\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col3\" class=\"data row14 col3\" >2</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col4\" class=\"data row14 col4\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col5\" class=\"data row14 col5\" >3</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col6\" class=\"data row14 col6\" >1</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col7\" class=\"data row14 col7\" >8</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col8\" class=\"data row14 col8\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col9\" class=\"data row14 col9\" >4</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col10\" class=\"data row14 col10\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col11\" class=\"data row14 col11\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col12\" class=\"data row14 col12\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col13\" class=\"data row14 col13\" >0</td>\n",
       "                        <td id=\"T_933a74ea_dc74_11e9_b8ea_0242ac110002row14_col14\" class=\"data row14 col14\" >17</td>\n",
       "            </tr>\n",
       "    </tbody></table>"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f821e8115c0>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(confusion_matrix(y_true, y_pred), index=est.classes_, columns=est.classes_).style.background_gradient()"
   ]
  }
 ],
 "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": 4
}