07-split-testtrain.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# `07-split-testtrain`\n",
    "\n",
    "Note that since we have specified the random seed this is reproducible"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/fowler/packages/v2/predict-pyrazinamide-resistance/.venv/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n",
      "/Users/fowler/packages/v2/predict-pyrazinamide-resistance/.venv/lib/python3.11/site-packages/MDAnalysis/topology/TPRParser.py:161: DeprecationWarning: 'xdrlib' is deprecated and slated for removal in Python 3.13\n",
      "  import xdrlib\n"
     ]
    }
   ],
   "source": [
    "import pandas, numpy, joblib\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import RobustScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "\n",
    "import sbmlcore"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "random_state=42\n",
    "mutations ={}\n",
    "mapping_dict = {'S':0, 'R':1}"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's read in the testtrain dataset so we can split it into training and test datasets but now we can drop the six features shown by the univariate logistical regression to be not much better than random"
   ]
  },
  {
   "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>d_hydropathy_KD</th>\n",
       "      <th>d_rogov</th>\n",
       "      <th>psi</th>\n",
       "      <th>residue_sasa</th>\n",
       "      <th>n_hbond_donors</th>\n",
       "      <th>temp_factor</th>\n",
       "      <th>depth</th>\n",
       "      <th>deep_ddG</th>\n",
       "      <th>rasp_score_ml</th>\n",
       "      <th>snap2_score</th>\n",
       "      <th>dist_FE2</th>\n",
       "      <th>dist_PZA</th>\n",
       "      <th>mcsm_stability_ddG</th>\n",
       "      <th>mapp_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.9</td>\n",
       "      <td>-0.389</td>\n",
       "      <td>-35.2</td>\n",
       "      <td>86.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>59.720001</td>\n",
       "      <td>2.103049</td>\n",
       "      <td>-0.770</td>\n",
       "      <td>-0.107405</td>\n",
       "      <td>20</td>\n",
       "      <td>29.948318</td>\n",
       "      <td>27.641663</td>\n",
       "      <td>-0.771</td>\n",
       "      <td>13.49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-2.6</td>\n",
       "      <td>-0.548</td>\n",
       "      <td>-35.2</td>\n",
       "      <td>86.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>59.720001</td>\n",
       "      <td>2.103049</td>\n",
       "      <td>-1.096</td>\n",
       "      <td>0.655765</td>\n",
       "      <td>52</td>\n",
       "      <td>29.948318</td>\n",
       "      <td>27.641663</td>\n",
       "      <td>-0.367</td>\n",
       "      <td>25.97</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.3</td>\n",
       "      <td>-0.514</td>\n",
       "      <td>-35.2</td>\n",
       "      <td>86.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>59.720001</td>\n",
       "      <td>2.103049</td>\n",
       "      <td>-0.908</td>\n",
       "      <td>0.364061</td>\n",
       "      <td>4</td>\n",
       "      <td>29.948318</td>\n",
       "      <td>27.641663</td>\n",
       "      <td>-0.912</td>\n",
       "      <td>30.01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   d_hydropathy_KD  d_rogov   psi  residue_sasa  n_hbond_donors  temp_factor  \\\n",
       "0              1.9   -0.389 -35.2          86.2             0.0    59.720001   \n",
       "1             -2.6   -0.548 -35.2          86.2             0.0    59.720001   \n",
       "2              2.3   -0.514 -35.2          86.2             0.0    59.720001   \n",
       "\n",
       "      depth  deep_ddG  rasp_score_ml  snap2_score   dist_FE2   dist_PZA  \\\n",
       "0  2.103049    -0.770      -0.107405           20  29.948318  27.641663   \n",
       "1  2.103049    -1.096       0.655765           52  29.948318  27.641663   \n",
       "2  2.103049    -0.908       0.364061            4  29.948318  27.641663   \n",
       "\n",
       "   mcsm_stability_ddG  mapp_score  \n",
       "0              -0.771       13.49  \n",
       "1              -0.367       25.97  \n",
       "2              -0.912       30.01  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pandas.read_csv('data/ds-traintest-phen-features.csv')\n",
    "\n",
    "df.drop(columns=['secondary_structure_codes',\n",
    " 'd_MW',\n",
    " 'phi',\n",
    " 'd_volume',\n",
    " 'd_Pi',\n",
    " 'n_hbond_acceptors'],inplace=True)\n",
    "\n",
    "mutations['tt'] = df['mutation']\n",
    "features = df.iloc[:,3:]\n",
    "labels = df['phenotype'].map({'S':0, 'R':1})\n",
    "features[:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "mutations_train, mutations_test, X_train, X_test, Y_train, Y_test = train_test_split(mutations['tt'], features, labels, test_size=0.3,\n",
    "                                                    random_state=random_state)\n",
    "\n",
    "class DataFrameSelector(BaseEstimator, TransformerMixin):\n",
    "    def __init__(self, attribute_names):\n",
    "        self.attribute_names = attribute_names\n",
    "    def fit(self, X, y=None):\n",
    "        return self\n",
    "    def transform(self, X):\n",
    "        return X[self.attribute_names].values\n",
    "\n",
    "numerical_attribs = list(features)\n",
    "\n",
    "headings = '\\n'.join(i for i in numerical_attribs)\n",
    "\n",
    "OUTPUT = open('data/ds-traintest-numericalattribs.txt', 'w')\n",
    "OUTPUT.write(headings)\n",
    "OUTPUT.close()\n",
    "\n",
    "pipe = joblib.load('data/pipeline.pkl')\n",
    "\n",
    "pipe = Pipeline([\n",
    "    ('selector', DataFrameSelector(numerical_attribs)),\n",
    "    ('scaling', RobustScaler())\n",
    "    ])\n",
    "\n",
    "X_train = pipe.fit_transform(X_train)    \n",
    "Y_train = Y_train.squeeze().to_numpy()\n",
    "Z_train = mutations_train.to_numpy()\n",
    "\n",
    "# let's save the pipeline to preserve the scaling etc\n",
    "joblib.dump(pipe, 'data/pipeline.pkl')\n",
    "\n",
    "X_test = pipe.transform(X_test)\n",
    "Y_test = Y_test.squeeze().to_numpy()   \n",
    "Z_test = mutations_test.to_numpy()\n",
    "\n",
    "with open('data/ds-train.npy', 'wb') as f:\n",
    "    numpy.save(f, Y_train)\n",
    "    numpy.save(f, X_train)\n",
    "    numpy.save(f, Z_train)\n",
    "\n",
    "with open('data/ds-test.npy', 'wb') as f:\n",
    "    numpy.save(f, Y_test)\n",
    "    numpy.save(f, X_test)    \n",
    "    numpy.save(f, Z_test)    \n",
    "\n",
    "df.loc[df.mutation.isin(mutations_test), 'set'] = 'test'\n",
    "df.loc[df.mutation.isin(mutations_train), 'set'] = 'train'\n",
    "df.to_csv('data/ds-traintest-phen-features-set.csv',index=False)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's check the % resistance is about the same between the train and test datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>phenotype</th>\n",
       "      <th>R</th>\n",
       "      <th>S</th>\n",
       "      <th>All</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>set</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>test</th>\n",
       "      <td>103</td>\n",
       "      <td>97</td>\n",
       "      <td>200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>train</th>\n",
       "      <td>246</td>\n",
       "      <td>218</td>\n",
       "      <td>464</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>All</th>\n",
       "      <td>349</td>\n",
       "      <td>315</td>\n",
       "      <td>664</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "phenotype    R    S  All\n",
       "set                     \n",
       "test       103   97  200\n",
       "train      246  218  464\n",
       "All        349  315  664"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pandas.crosstab(df.set,df.phenotype,margins=True)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's work out what the predictions using Suspect-PZA are for the training and test datasets for analysis later"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>prediction</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mutation</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>C72G</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D129G</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T167N</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>L85P</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>V139L</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I31T</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D8Y</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>S84I</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>R154G</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D110V</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1010 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          prediction\n",
       "mutation            \n",
       "C72G               1\n",
       "D129G              1\n",
       "T167N              0\n",
       "L85P               1\n",
       "V139L              1\n",
       "...              ...\n",
       "I31T               1\n",
       "D8Y                1\n",
       "S84I               0\n",
       "R154G              0\n",
       "D110V              0\n",
       "\n",
       "[1010 rows x 1 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "suspectpza = pandas.read_csv('data/reference/ds-suspectpza-results.csv')\n",
    "def create_mutation(row):\n",
    "    return sbmlcore.amino_acid_3to1letter[row['Wild Type']] + str(row['Position']) + sbmlcore.amino_acid_3to1letter[row['Mutant']]\n",
    "\n",
    "suspectpza['mutation'] = suspectpza.apply(create_mutation,axis=1)    \n",
    "suspectpza.set_index('mutation', inplace=True)\n",
    "suspectpza['prediction'] = suspectpza[' Prediction'].str[0].map(mapping_dict)\n",
    "suspectpza.drop(columns=['Chain','Wild Type','Position','Mutant','Distance to ligand',' Prediction'],inplace=True)\n",
    "suspectpza"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using the (fitted) pipeline we can repeat the process for the two types of `validation` dataset, depending whether just clinical samples or aggregated mutations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in ['samples','mutations']:\n",
    "    df = pandas.read_csv('data/ds-validation-' + i + '-phen-features.csv')\n",
    "\n",
    "    df.drop(columns=['secondary_structure_codes',\n",
    "        'd_MW',\n",
    "        'phi',\n",
    "        'd_volume',\n",
    "        'd_Pi',\n",
    "        'n_hbond_acceptors'],inplace=True)\n",
    "\n",
    "    mutations['v'] = df['mutation']\n",
    "    features = df.iloc[:,3:]\n",
    "    labels = df['phenotype'].map({'S':0, 'R':1, 'U':2})\n",
    "\n",
    "    X_validate = pipe.transform(features)\n",
    "    Y_validate = labels.squeeze().to_numpy()\n",
    "    Z_validate = mutations['v'].to_numpy()\n",
    "\n",
    "    with open('data/ds-validation-' + i + '.npy', 'wb') as f:\n",
    "        numpy.save(f, Y_validate)\n",
    "        numpy.save(f, X_validate)\n",
    "        numpy.save(f, Z_validate)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "..and again label the mutations present in the `validation` dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>prediction</th>\n",
       "      <th>validation-mutations</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mutation</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A102E</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102G</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102P</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102R</th>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102S</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99C</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99D</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99F</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99H</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99N</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1010 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          prediction  validation-mutations\n",
       "mutation                                  \n",
       "A102E              1                   NaN\n",
       "A102G              0                   NaN\n",
       "A102P              0                   NaN\n",
       "A102R              1                   0.0\n",
       "A102S              1                   NaN\n",
       "...              ...                   ...\n",
       "Y99C               0                   NaN\n",
       "Y99D               0                   NaN\n",
       "Y99F               0                   NaN\n",
       "Y99H               0                   NaN\n",
       "Y99N               0                   NaN\n",
       "\n",
       "[1010 rows x 2 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.set_index('mutation', inplace=True)\n",
    "suspectpza = suspectpza.join(df[['phenotype']], how='outer')\n",
    "suspectpza['validation-mutations'] = suspectpza.phenotype.map(mapping_dict)\n",
    "suspectpza.drop(columns=['phenotype'], inplace=True)\n",
    "suspectpza"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pandas.read_csv('data/ds-mic-phen-features.csv')\n",
    "\n",
    "df.drop(columns=['secondary_structure_codes',\n",
    "        'd_MW',\n",
    "        'phi',\n",
    "        'd_volume',\n",
    "        'd_Pi',\n",
    "        'n_hbond_acceptors'],inplace=True)\n",
    "\n",
    "\n",
    "mutations['mic'] = df['mutation']\n",
    "features = df.iloc[:,3:]\n",
    "labels = df['phenotype'].map({'S':0, 'R':1})\n",
    "\n",
    "X_mic = pipe.transform(features)\n",
    "Y_mic = labels.squeeze().to_numpy()\n",
    "Z_mic = mutations['mic'].to_numpy()\n",
    "\n",
    "with open('data/ds-mic.npy', 'wb') as f:\n",
    "    numpy.save(f, Y_mic)\n",
    "    numpy.save(f, X_mic)\n",
    "    numpy.save(f, Z_mic)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>prediction</th>\n",
       "      <th>validation-mutations</th>\n",
       "      <th>mic</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mutation</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A102E</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102G</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102P</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          prediction  validation-mutations  mic\n",
       "mutation                                       \n",
       "A102E              1                   NaN  NaN\n",
       "A102G              0                   NaN  NaN\n",
       "A102P              0                   NaN  NaN"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.set_index('mutation', inplace=True)\n",
    "suspectpza = suspectpza.join(df[['phenotype']], how='outer')\n",
    "suspectpza['mic'] = suspectpza.phenotype.map(mapping_dict)\n",
    "suspectpza.drop(columns=['phenotype'], inplace=True)\n",
    "suspectpza[:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert len(Y_mic) == suspectpza.mic.notna().sum(), suspectpza.mic.notna().sum()\n",
    "assert len(Y_validate) == suspectpza['validation-mutations'].notna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>prediction</th>\n",
       "      <th>validation-mutations</th>\n",
       "      <th>mic</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mutation</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A102E</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102G</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102P</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102R</th>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102S</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99C</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99D</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99F</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99H</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Y99N</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1010 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          prediction  validation-mutations  mic\n",
       "mutation                                       \n",
       "A102E              1                   NaN  NaN\n",
       "A102G              0                   NaN  NaN\n",
       "A102P              0                   NaN  NaN\n",
       "A102R              1                   0.0  NaN\n",
       "A102S              1                   NaN  NaN\n",
       "...              ...                   ...  ...\n",
       "Y99C               0                   NaN  NaN\n",
       "Y99D               0                   NaN  NaN\n",
       "Y99F               0                   NaN  NaN\n",
       "Y99H               0                   NaN  NaN\n",
       "Y99N               0                   NaN  NaN\n",
       "\n",
       "[1010 rows x 3 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "suspectpza"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in ['validation-mutations', 'mic']:\n",
    "    Y=suspectpza[suspectpza[i].notna()][i].squeeze().to_numpy()\n",
    "    Z=suspectpza[suspectpza[i].notna()].prediction.squeeze().to_numpy()\n",
    "    ZZ=suspectpza[suspectpza[i].notna()].index.to_numpy()\n",
    "    with open('data/suspectpza-'+i+'.npy', 'wb') as f:\n",
    "        numpy.save(f, Y)\n",
    "        numpy.save(f, Z)\n",
    "        numpy.save(f, ZZ)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>phenotype</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mutation</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A102P</th>\n",
       "      <td>R</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102P</th>\n",
       "      <td>R</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102P</th>\n",
       "      <td>R</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         phenotype\n",
       "mutation          \n",
       "A102P            R\n",
       "A102P            R\n",
       "A102P            R"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pandas.read_csv('data/ds-validation-samples-phen.csv')\n",
    "df.rename(columns={'MUTATION': 'mutation', 'PHENOTYPE': 'phenotype'}, inplace=True)\n",
    "df.set_index('mutation',inplace=True)\n",
    "df[:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "prediction\n",
       "1    3513\n",
       "0    1157\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "suspectpza = suspectpza.join(df[['phenotype']], how='outer')\n",
    "suspectpza.prediction.value_counts(dropna=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>prediction</th>\n",
       "      <th>validation-mutations</th>\n",
       "      <th>mic</th>\n",
       "      <th>phenotype</th>\n",
       "      <th>validation-samples</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mutation</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A102E</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102G</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A102P</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>R</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          prediction  validation-mutations  mic phenotype  validation-samples\n",
       "mutation                                                                     \n",
       "A102E              1                   NaN  NaN       NaN                 NaN\n",
       "A102G              0                   NaN  NaN       NaN                 NaN\n",
       "A102P              0                   NaN  NaN         R                 1.0"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "suspectpza['validation-samples'] = suspectpza.phenotype.map(mapping_dict)\n",
    "suspectpza[:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in ['validation-samples']:\n",
    "    Y=suspectpza[suspectpza[i].notna()][i].squeeze().to_numpy()\n",
    "    Z=suspectpza[suspectpza[i].notna()].prediction.squeeze().to_numpy()\n",
    "    ZZ=suspectpza[suspectpza[i].notna()].index.to_numpy()\n",
    "    with open('data/suspectpza-'+i+'.npy', 'wb') as f:\n",
    "        numpy.save(f, Y)\n",
    "        numpy.save(f, Z)\n",
    "        numpy.save(f, ZZ)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "suspectpza = suspectpza[~(suspectpza['validation-mutations'].isna() & suspectpza.mic.isna())]\n",
    "suspectpza.to_csv('data/reference/ds-suspectpza-results-sets.csv')"
   ]
  }
 ],
 "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": "3.11.6"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}