prf_demo.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "enormous-default",
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "import numpy as np\n",
    "from rdkit import Chem\n",
    "from rdkit.Chem import DataStructs\n",
    "from rdkit.Chem import rdFingerprintGenerator\n",
    "from rdkit.Chem import rdBase\n",
    "from rdkit import RDConfig\n",
    "from PRF import prf\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import metrics\n",
    "from scipy.stats import norm\n",
    "\n",
    "# if SOL > -1, solubility class is defined as high so I defined threshold is -1.\n",
    "def calc_delta_y(value, scale=0.00001):\n",
    "    v = norm.cdf(value, scale=scale)\n",
    "    return [v, 1-v]\n",
    "\n",
    "def mol2fp(mol, fpgen):\n",
    "    fp = fpgen.GetFingerprint(mol)\n",
    "    arr = np.zeros((1,))\n",
    "    DataStructs.ConvertToNumpyArray(fp, arr)\n",
    "    return arr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "hundred-smile",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = os.path.join(RDConfig.RDDocsDir,'Book/data/solubility.train.sdf')\n",
    "test = os.path.join(RDConfig.RDDocsDir,'Book/data/solubility.test.sdf')\n",
    "\n",
    "data_train = [m for m in Chem.SDMolSupplier(train)]\n",
    "data_test = [m for m in Chem.SDMolSupplier(test)]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "rocky-bruce",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1e-05\n",
      "0.8599221789883269\n",
      "0.8910505836575876\n"
     ]
    }
   ],
   "source": [
    "scale=0.00001\n",
    "fpgen = rdFingerprintGenerator.GetMorganGenerator(radius=2, fpSize=1024)\n",
    "X_train = np.asarray([mol2fp(m, fpgen) for m in data_train])\n",
    "X_test =  np.asarray([mol2fp(m, fpgen) for m in data_test])\n",
    "sol_train = [float(m.GetProp('SOL')) for m in data_train]\n",
    "delta_sol_train = [s-(-1) for s in sol_train]\n",
    "delta_y_train = np.asarray([calc_delta_y(v, scale=scale) for v in delta_sol_train])\n",
    "y_train = [np.argmax(y) for y in delta_y_train]\n",
    "\n",
    "sol_test = [float(m.GetProp('SOL')) for m in data_test]\n",
    "delta_sol_test = [s-(-1) for s in sol_test]\n",
    "delta_y_test = np.asanyarray([calc_delta_y(v, scale=scale) for v in delta_sol_test])\n",
    "y_test = [np.argmax(y) for y in delta_y_test]\n",
    "\n",
    "\n",
    "prf_cls = prf()\n",
    "rf_cls = RandomForestClassifier(random_state=1)\n",
    "prf_cls.fit(X_train, py=delta_y_train)\n",
    "rf_cls.fit(X_train, y_train)\n",
    "print(scale)\n",
    "print(prf_cls.score(X_test, y_test))\n",
    "print(rf_cls.score(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "acquired-arena",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.3\n",
      "0.8793774319066148\n",
      "0.8910505836575876\n"
     ]
    }
   ],
   "source": [
    "scale=0.3\n",
    "fpgen = rdFingerprintGenerator.GetMorganGenerator(radius=2, fpSize=1024)\n",
    "X_train = np.asarray([mol2fp(m, fpgen) for m in data_train])\n",
    "X_test =  np.asarray([mol2fp(m, fpgen) for m in data_test])\n",
    "sol_train = [float(m.GetProp('SOL')) for m in data_train]\n",
    "delta_sol_train = [s-(-1) for s in sol_train]\n",
    "delta_y_train = np.asarray([calc_delta_y(v, scale=scale) for v in delta_sol_train])\n",
    "y_train = [np.argmax(y) for y in delta_y_train]\n",
    "\n",
    "sol_test = [float(m.GetProp('SOL')) for m in data_test]\n",
    "delta_sol_test = [s-(-1) for s in sol_test]\n",
    "delta_y_test = np.asanyarray([calc_delta_y(v, scale=scale) for v in delta_sol_test])\n",
    "y_test = [np.argmax(y) for y in delta_y_test]\n",
    "\n",
    "\n",
    "prf_cls = prf()\n",
    "rf_cls = RandomForestClassifier(random_state=1)\n",
    "prf_cls.fit(X_train, py=delta_y_train)\n",
    "rf_cls.fit(X_train, y_train)\n",
    "print(scale)\n",
    "print(prf_cls.score(X_test, y_test))\n",
    "print(rf_cls.score(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "documented-buffalo",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.6\n",
      "0.8988326848249028\n",
      "0.8910505836575876\n"
     ]
    }
   ],
   "source": [
    "scale=0.6\n",
    "fpgen = rdFingerprintGenerator.GetMorganGenerator(radius=2, fpSize=1024)\n",
    "X_train = np.asarray([mol2fp(m, fpgen) for m in data_train])\n",
    "X_test =  np.asarray([mol2fp(m, fpgen) for m in data_test])\n",
    "sol_train = [float(m.GetProp('SOL')) for m in data_train]\n",
    "delta_sol_train = [s-(-1) for s in sol_train]\n",
    "delta_y_train = np.asarray([calc_delta_y(v, scale=scale) for v in delta_sol_train])\n",
    "y_train = [np.argmax(y) for y in delta_y_train]\n",
    "\n",
    "sol_test = [float(m.GetProp('SOL')) for m in data_test]\n",
    "delta_sol_test = [s-(-1) for s in sol_test]\n",
    "delta_y_test = np.asanyarray([calc_delta_y(v, scale=scale) for v in delta_sol_test])\n",
    "y_test = [np.argmax(y) for y in delta_y_test]\n",
    "\n",
    "\n",
    "prf_cls = prf()\n",
    "rf_cls = RandomForestClassifier(random_state=1)\n",
    "prf_cls.fit(X_train, py=delta_y_train)\n",
    "rf_cls.fit(X_train, y_train)\n",
    "print(scale)\n",
    "print(prf_cls.score(X_test, y_test))\n",
    "print(rf_cls.score(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "static-freedom",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8\n",
      "0.8910505836575876\n",
      "0.8910505836575876\n"
     ]
    }
   ],
   "source": [
    "scale=0.8\n",
    "fpgen = rdFingerprintGenerator.GetMorganGenerator(radius=2, fpSize=1024)\n",
    "X_train = np.asarray([mol2fp(m, fpgen) for m in data_train])\n",
    "X_test =  np.asarray([mol2fp(m, fpgen) for m in data_test])\n",
    "sol_train = [float(m.GetProp('SOL')) for m in data_train]\n",
    "delta_sol_train = [s-(-1) for s in sol_train]\n",
    "delta_y_train = np.asarray([calc_delta_y(v, scale=scale) for v in delta_sol_train])\n",
    "y_train = [np.argmax(y) for y in delta_y_train]\n",
    "\n",
    "sol_test = [float(m.GetProp('SOL')) for m in data_test]\n",
    "delta_sol_test = [s-(-1) for s in sol_test]\n",
    "delta_y_test = np.asanyarray([calc_delta_y(v, scale=scale) for v in delta_sol_test])\n",
    "y_test = [np.argmax(y) for y in delta_y_test]\n",
    "\n",
    "\n",
    "prf_cls = prf()\n",
    "rf_cls = RandomForestClassifier(random_state=1)\n",
    "prf_cls.fit(X_train, py=delta_y_train)\n",
    "rf_cls.fit(X_train, y_train)\n",
    "print(scale)\n",
    "print(prf_cls.score(X_test, y_test))\n",
    "print(rf_cls.score(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "simple-friendship",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.0\n",
      "0.9027237354085603\n",
      "0.8910505836575876\n"
     ]
    }
   ],
   "source": [
    "scale=1.0\n",
    "fpgen = rdFingerprintGenerator.GetMorganGenerator(radius=2, fpSize=1024)\n",
    "X_train = np.asarray([mol2fp(m, fpgen) for m in data_train])\n",
    "X_test =  np.asarray([mol2fp(m, fpgen) for m in data_test])\n",
    "sol_train = [float(m.GetProp('SOL')) for m in data_train]\n",
    "delta_sol_train = [s-(-1) for s in sol_train]\n",
    "delta_y_train = np.asarray([calc_delta_y(v, scale=scale) for v in delta_sol_train])\n",
    "y_train = [np.argmax(y) for y in delta_y_train]\n",
    "\n",
    "sol_test = [float(m.GetProp('SOL')) for m in data_test]\n",
    "delta_sol_test = [s-(-1) for s in sol_test]\n",
    "delta_y_test = np.asanyarray([calc_delta_y(v, scale=scale) for v in delta_sol_test])\n",
    "y_test = [np.argmax(y) for y in delta_y_test]\n",
    "\n",
    "\n",
    "prf_cls = prf()\n",
    "rf_cls = RandomForestClassifier(random_state=1)\n",
    "prf_cls.fit(X_train, py=delta_y_train)\n",
    "rf_cls.fit(X_train, y_train)\n",
    "print(scale)\n",
    "print(prf_cls.score(X_test, y_test))\n",
    "print(rf_cls.score(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ideal-algorithm",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}