experimental_results.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ae9d8848-cb85-4306-a0b9-b5d4a00830a9",
   "metadata": {
    "tags": []
   },
   "source": [
    "# HiSIF-DTA\n",
    "\n",
    "\n",
    "A notebook for \"**HiSIF-DTA: A Hierarchical Semantic Information Fusion Framework for Drug-Target Affinity Prediction**\". In this study, we propose a protein semantic fusion-based DTA (Drug-Target Affinity) prediction method and design two different backbones for the semantic fusion process, namely **_TDNet (Top-Down)_** and **_BUNet (Bottom-Up)_**. To validate the effectiveness of our proposed method, we selected three benchmark datasets, namely **Davis**, **KIBA**, and **Human**, along with several relevant baseline models for evaluation. It is worth noting that, to ensure fair comparison, we retrained the baseline models, and their respective training code can be found in the **'baselines'** directory of our repository. The following presents the prediction results and performance comparison between our proposed model and other baseline models.\n",
    "\n",
    "==============================================================================================================\n",
    "\n",
    "## 1. Baseline models\n",
    "\n",
    "Detailed information about each baseline model can be found at the following link: [Click](https://github.com/bixiangpeng/HiSIF-DTA/blob/main/baselines/README.md).\n",
    "\n",
    "## 2. Parameter tuning\n",
    "\n",
    "### 2.1 _Top-Down_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "4b7452a4-1b63-449b-95a0-faaedf3c6d93",
   "metadata": {},
   "outputs": [],
   "source": [
    "#!pip install lifelines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1147aeca-93d0-4ad1-a215-186e5fca486e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from urllib.parse import quote\n",
    "from lifelines.utils import concordance_index\n",
    "from sklearn.metrics import roc_auc_score, precision_score, recall_score\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "def mse(y,f):\n",
    "    mse = ((y - f)**2).mean(axis=0)\n",
    "    return mse\n",
    "\n",
    "backbones = ['TDNet','BUNet']\n",
    "para_space = [0, 0.1, 0.2, 0.3, 0.4]\n",
    "TD_MSE_list = []\n",
    "TD_CI_list = []\n",
    "for para in para_space:\n",
    "    results = []\n",
    "    for fold in range(1,6):\n",
    "        url = f'https://zenodo.org/record/8402218/files/predicted_value_of_{backbones[0]}_dropedge_{para}_on_davis_test{fold} .csv'\n",
    "        url = quote(url, safe=':/')\n",
    "        df = pd.read_csv(url)\n",
    "        G = np.array(list(df['label']))\n",
    "        P = np.array(list(df['predicted value']))\n",
    "        results.append([mse(G,P),concordance_index(G,P)])\n",
    "    valid_results = np.array(results)\n",
    "    valid_results = [np.mean(valid_results, axis=0), np.std(valid_results, axis=0)]\n",
    "    TD_MSE_list.append(eval('{:.3f}'.format(valid_results[0][0])))\n",
    "    TD_CI_list.append(eval('{:.3f}'.format(valid_results[0][1])))\n",
    "\n",
    "plt.figure(1,(8,4))\n",
    "ax1 = plt.subplot(1,2,1)\n",
    "ax2 = plt.subplot(1,2,2)\n",
    "\n",
    "ax1.plot(para_space, TD_MSE_list, c = 'lightcoral', marker = '^', lw = 3)\n",
    "ax1.bar(para_space, TD_MSE_list,width=0.05,color='royalblue')\n",
    "ax1_x_ticks = np.arange(0,0.5,0.1)\n",
    "ax1_y_ticks = np.arange(0.195,0.205,0.003)\n",
    "ax1.set_xlim((-0.05, 0.45))\n",
    "ax1.set_ylim((0.195, 0.205))\n",
    "ax1.set_xticks(ax1_x_ticks)\n",
    "ax1.set_yticks(ax1_y_ticks)\n",
    "ax1.set_title('MSE')\n",
    "\n",
    "ax2.plot(para_space, TD_CI_list, c = 'lightcoral', marker = '^', lw = 3)\n",
    "ax2.bar(para_space, TD_CI_list,width=0.05,color='royalblue')\n",
    "ax2_x_ticks = np.arange(0,0.5,0.1)\n",
    "ax2_y_ticks = np.arange(0.891,0.899,0.002)\n",
    "ax2.set_xlim((-0.05, 0.45))\n",
    "ax2.set_ylim((0.891, 0.900))\n",
    "ax2.set_xticks(ax2_x_ticks)\n",
    "ax2.set_yticks(ax2_y_ticks)\n",
    "ax2.set_title('CI')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cfead643",
   "metadata": {},
   "source": [
    "### 2.2 _Bottom-Up_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "bd39dcc7",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "BU_MSE_list = []\n",
    "BU_CI_list = []\n",
    "for para in para_space:\n",
    "    results = []\n",
    "    for fold in range(1,6):\n",
    "        url = f'https://zenodo.org/record/8402218/files/predicted_value_of_{backbones[1]}_dropedge_{para}_on_davis_test{fold} .csv'\n",
    "        url = quote(url, safe=':/')\n",
    "        df = pd.read_csv(url)\n",
    "        G = np.array(list(df['label']))\n",
    "        P = np.array(list(df['predicted value']))\n",
    "        results.append([mse(G,P),concordance_index(G,P)])\n",
    "    valid_results = np.array(results)\n",
    "    valid_results = [np.mean(valid_results, axis=0), np.std(valid_results, axis=0)]\n",
    "    BU_MSE_list.append(eval('{:.3f}'.format(valid_results[0][0])))\n",
    "    BU_CI_list.append(eval('{:.3f}'.format(valid_results[0][1])))\n",
    "\n",
    "plt.figure(2,(8,4))\n",
    "ax1 = plt.subplot(1,2,1)\n",
    "ax2 = plt.subplot(1,2,2)\n",
    "\n",
    "ax1.plot(para_space, BU_MSE_list, c = 'lightcoral', marker = '^', lw = 3)\n",
    "ax1.bar(para_space, BU_MSE_list,width=0.05,color='royalblue')\n",
    "ax1_x_ticks = np.arange(0,0.5,0.1)\n",
    "ax1_y_ticks = np.arange(0.2,0.213,0.003)\n",
    "ax1.set_xlim((-0.05, 0.45))\n",
    "ax1.set_ylim((0.2, 0.213))\n",
    "ax1.set_xticks(ax1_x_ticks)\n",
    "ax1.set_yticks(ax1_y_ticks)\n",
    "ax1.set_title('MSE')\n",
    "\n",
    "ax2.plot(para_space, BU_CI_list, c = 'lightcoral', marker = '^', lw = 3)\n",
    "ax2.bar(para_space, BU_CI_list,width=0.05,color='royalblue')\n",
    "ax2_x_ticks = np.arange(0,0.5,0.1)\n",
    "ax2_y_ticks = np.arange(0.89,0.899,0.002)\n",
    "ax2.set_xlim((-0.05, 0.45))\n",
    "ax2.set_ylim((0.89, 0.899))\n",
    "ax2.set_xticks(ax2_x_ticks)\n",
    "ax2.set_yticks(ax2_y_ticks)\n",
    "ax2.set_title('CI')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0762ac65-5eaa-46e5-994a-696f370c0325",
   "metadata": {
    "tags": []
   },
   "source": [
    "\n",
    "## 3. Comparative experiments\n",
    "\n",
    "### 3.1 Davis "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "dab40e8d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                MSE     CI\n",
      "DeepDTA       0.253  0.879\n",
      "AttentionDTA  0.244  0.885\n",
      "GraphDTA      0.230  0.885\n",
      "MGraphDTA     0.207  0.895\n",
      "DGraphDTA     0.202  0.905\n",
      "TDNet         0.193  0.907\n",
      "BUNet         0.191  0.906\n"
     ]
    }
   ],
   "source": [
    "# Davis\n",
    "DTA_models = ['DeepDTA','AttentionDTA','GraphDTA','MGraphDTA','DGraphDTA','TDNet','BUNet']\n",
    "davis_MSE_list = []\n",
    "davis_CI_list = []\n",
    "for model in DTA_models:\n",
    "    url = f'https://zenodo.org/record/8385073/files/predicted_value_of_{model}_on_davis .csv'\n",
    "    url = quote(url, safe=':/')\n",
    "    df = pd.read_csv(url)\n",
    "    G = np.array(list(df['label']))\n",
    "    P = np.array(list(df['predicted value']))\n",
    "    davis_MSE_list.append(\"{:.3f}\".format(mse(G,P)))\n",
    "    davis_CI_list.append(\"{:.3f}\".format(concordance_index(G,P)))\n",
    "\n",
    "performance_on_davis = {\n",
    "    'MSE':davis_MSE_list,\n",
    "    'CI':davis_CI_list\n",
    "}\n",
    "df_performance_on_davis = pd.DataFrame(performance_on_davis)\n",
    "df_performance_on_davis.index = DTA_models\n",
    "print(df_performance_on_davis)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec819594",
   "metadata": {},
   "source": [
    "### 3.2 KIBA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "617a119a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                MSE     CI\n",
      "DeepDTA       0.187  0.854\n",
      "AttentionDTA  0.175  0.867\n",
      "GraphDTA      0.146  0.885\n",
      "MGraphDTA     0.128  0.902\n",
      "DGraphDTA     0.127  0.902\n",
      "TDNet         0.120  0.904\n",
      "BUNet         0.121  0.904\n"
     ]
    }
   ],
   "source": [
    "kiba_MSE_list = []\n",
    "kiba_CI_list = []\n",
    "for model in DTA_models:\n",
    "    url = f'https://zenodo.org/record/8385073/files/predicted_value_of_{model}_on_kiba .csv'\n",
    "    url = quote(url, safe=':/')\n",
    "    df = pd.read_csv(url)\n",
    "    G = np.array(list(df['label']))\n",
    "    P = np.array(list(df['predicted value']))\n",
    "    kiba_MSE_list.append(\"{:.3f}\".format(mse(G,P)))\n",
    "    kiba_CI_list.append(\"{:.3f}\".format(concordance_index(G,P)))\n",
    "\n",
    "performance_on_kiba = {\n",
    "    'MSE':kiba_MSE_list,\n",
    "    'CI':kiba_CI_list\n",
    "}\n",
    "df_performance_on_kiba = pd.DataFrame(performance_on_kiba)\n",
    "df_performance_on_kiba.index = DTA_models\n",
    "print(df_performance_on_kiba)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5542e8ac",
   "metadata": {},
   "source": [
    "### 3.3 Human"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "d9daa81f",
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                  AUC Precision Recall\n",
      "CPI-GNN         0.965     0.919  0.912\n",
      "DrugVQA         0.966     0.921  0.914\n",
      "TransformerCPI  0.974     0.914  0.923\n",
      "GraphDTA        0.975     0.930  0.917\n",
      "TDNet           0.988     0.945  0.952\n",
      "BUNet           0.986     0.947  0.947\n"
     ]
    }
   ],
   "source": [
    "CPI_models = ['CPI-GNN','DrugVQA','TransformerCPI','GraphDTA','TDNet','BUNet']\n",
    "AUC_list = []\n",
    "Precision_list = []\n",
    "Recall_list = []\n",
    "for model in CPI_models:\n",
    "    results = []\n",
    "    for fold in range(1,6):\n",
    "        url = f'https://zenodo.org/record/8385073/files/predicted_value_of_{model}_on_Human_test{fold} .csv'\n",
    "        url = quote(url, safe=':/')\n",
    "        df = pd.read_csv(url)\n",
    "        G = np.array(list(df['label']))\n",
    "        P = np.array(list(df['predicted value']))\n",
    "        P_label = np.array(list(df['predicted label']))\n",
    "        AUC = roc_auc_score(G, P)\n",
    "        precision = precision_score(G, P_label)\n",
    "        recall = recall_score(G, P_label)\n",
    "        results.append([AUC, precision, recall])\n",
    "    valid_results = np.array(results)\n",
    "    valid_results = [np.mean(valid_results, axis=0), np.std(valid_results, axis=0)]\n",
    "    AUC_list.append('{:.3f}'.format(valid_results[0][0]))\n",
    "    Precision_list.append('{:.3f}'.format(valid_results[0][1]))\n",
    "    Recall_list.append('{:.3f}'.format(valid_results[0][2]))\n",
    "performance_on_Human = {\n",
    "    'AUC': AUC_list,\n",
    "    'Precision': Precision_list,\n",
    "    'Recall':Recall_list\n",
    "}\n",
    "df_performance_on_Human = pd.DataFrame(performance_on_Human)\n",
    "df_performance_on_Human.index = CPI_models\n",
    "print(df_performance_on_Human)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e965ccaf",
   "metadata": {},
   "source": [
    "## 4. Ablation experiments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "16f8987c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ablation_models = ['TDNet','BUNet','LSM','HSM','HFS']\n",
    "color_list = ['royalblue','steelblue','lightsalmon','tomato','firebrick']\n",
    "ablation_MSE_list = []\n",
    "ablation_CI_list = []\n",
    "\n",
    "for model in ablation_models:\n",
    "    url = f'https://zenodo.org/record/8418121/files/predicted_value_of_{model}_on_davis .csv'\n",
    "    url = quote(url, safe=':/')\n",
    "    df = pd.read_csv(url)\n",
    "    G = np.array(list(df['label']))\n",
    "    P = np.array(list(df['predicted value']))\n",
    "    ablation_MSE_list.append(eval(\"{:.3f}\".format(mse(G,P))))\n",
    "    ablation_CI_list.append(eval(\"{:.3f}\".format(concordance_index(G,P))))\n",
    "    \n",
    "plt.figure(1,(8,4))\n",
    "ax1 = plt.subplot(1,2,1)\n",
    "ax2 = plt.subplot(1,2,2)\n",
    "\n",
    "for i in range(len(ablation_models)):\n",
    "    ax1.bar(ablation_models[i], ablation_MSE_list[i],width=0.5,color=color_list[i])\n",
    "ax1_y_ticks = np.arange(0.185,0.206,0.005)\n",
    "ax1.set_ylim((0.185, 0.206))\n",
    "ax1.set_yticks(ax1_y_ticks)\n",
    "ax1.set_title('MSE')\n",
    "\n",
    "for i in range(len(ablation_models)):\n",
    "    ax2.bar(ablation_models[i], ablation_CI_list[i],width=0.5,color=color_list[i])\n",
    "ax2_y_ticks = np.arange(0.892,0.909,0.004)\n",
    "ax2.set_ylim((0.892, 0.909))\n",
    "ax2.set_yticks(ax2_y_ticks)\n",
    "ax2.set_title('CI')\n",
    "\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}