DEMO/lactamase_Protein_Property.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is the tutorial for protein property function. The dataset that we use is called beta lactamase."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First we import all necessary packages."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import sys\n",
    "import argparse\n",
    "import torch\n",
    "import wandb\n",
    "sys.path.append(os.path.abspath('..'))\n",
    "\n",
    "from DeepProtein.dataset import *\n",
    "import DeepProtein.utils as utils\n",
    "import DeepProtein.ProteinPred as models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next we fix some hyperparamaters. Here we make wandb local."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Failed to detect the name of this notebook, you can set it manually with the WANDB_NOTEBOOK_NAME environment variable to enable code saving.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "Tracking run with wandb version 0.17.4"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "W&B syncing is set to <code>`offline`<code> in this directory.  <br/>Run <code>`wandb online`<code> or set <code>WANDB_MODE=online<code> to enable cloud syncing."
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<button onClick=\"this.nextSibling.style.display='block';this.style.display='none';\">Display W&B run</button><iframe src='?jupyter=true' style='border:none;width:100%;height:420px;display:none;'></iframe>"
      ],
      "text/plain": [
       "<wandb.sdk.wandb_run.Run at 0x16177a13c10>"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "wandb.init(mode='offline', name = \"PPI_Affinity\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "lr = 0.0001\n",
    "epochs = 20\n",
    "batch_size = 32\n",
    "seed = 7\n",
    "target_encoding = \"CNN\" # test only"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next we load our dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_beta = Beta_lactamase('../DeepProtein/data', 'train')\n",
    "valid_beta = Beta_lactamase('../DeepProtein/data', 'valid')\n",
    "test_beta = Beta_lactamase('../DeepProtein/data', 'test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_protein_processed, train_target, train_protein_idx = collate_fn(train_beta)\n",
    "valid_protein_processed, valid_target, valid_protein_idx = collate_fn(valid_beta)\n",
    "test_protein_processed, test_target, test_protein_idx = collate_fn(test_beta)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Protein Function Prediction Mode...\n",
      "in total: 4158 proteins\n",
      "encoding protein...\n",
      "unique target sequence: 4158\n",
      "Done.\n",
      "Protein Function Prediction Mode...\n",
      "in total: 520 proteins\n",
      "encoding protein...\n",
      "unique target sequence: 520\n",
      "Done.\n",
      "Protein Function Prediction Mode...\n",
      "in total: 520 proteins\n",
      "encoding protein...\n",
      "unique target sequence: 520\n",
      "Done.\n"
     ]
    }
   ],
   "source": [
    "train, _, _ = utils.data_process(X_target=train_protein_processed, y=train_target, target_encoding=target_encoding,\n",
    "                                    # drug_encoding= drug_encoding,\n",
    "                                    split_method='random', frac=[0.99998, 1e-5, 1e-5],\n",
    "                                    random_seed=1)\n",
    "\n",
    "_, val, _ = utils.data_process(X_target=valid_protein_processed, y=valid_target, target_encoding=target_encoding,\n",
    "                                # drug_encoding= drug_encoding,\n",
    "                                split_method='random', frac=[1e-5, 0.99998, 1e-5],\n",
    "                                random_seed=1)\n",
    "\n",
    "_, _, test = utils.data_process(X_target=test_protein_processed, y=test_target, target_encoding=target_encoding,\n",
    "                                # drug_encoding= drug_encoding,\n",
    "                                split_method='random', frac=[1e-5, 1e-5, 0.99998],\n",
    "                                random_seed=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load configuration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "config = generate_config(target_encoding=target_encoding,\n",
    "                            cls_hidden_dims=[1024, 1024],\n",
    "                            train_epoch=epochs,\n",
    "                            LR=lr,\n",
    "                            batch_size=batch_size ,\n",
    "                            )\n",
    "config['multi'] = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Set seed"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<torch._C.Generator at 0x161342c9670>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.manual_seed(seed)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load model and train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Let's use 1 GPU!\n",
      "--- Data Preparation ---\n",
      "--- Go for Training ---\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\anaconda3\\envs\\graph\\lib\\site-packages\\torch\\autograd\\graph.py:744: UserWarning: Plan failed with a cudnnException: CUDNN_BACKEND_EXECUTION_PLAN_DESCRIPTOR: cudnnFinalize Descriptor Failed cudnn_status: CUDNN_STATUS_NOT_SUPPORTED (Triggered internally at C:\\actions-runner\\_work\\pytorch\\pytorch\\builder\\windows\\pytorch\\aten\\src\\ATen\\native\\cudnn\\Conv_v8.cpp:919.)\n",
      "  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training at Epoch 1 iteration 0 with loss 0.60325. Total time 0.00611 hours\n",
      "Training at Epoch 1 iteration 100 with loss 0.14836. Total time 0.00722 hours\n",
      "Validation at Epoch 1 , MAE: 0.28039 , MSE: 0.10288 , Spearman Correlation: 0.03946 with p-value: 3.69E-01 , Concordance Index: 0.51339\n",
      "Training at Epoch 2 iteration 0 with loss 0.11994. Total time 0.01972 hours\n",
      "Training at Epoch 2 iteration 100 with loss 0.11785. Total time 0.02111 hours\n",
      "Validation at Epoch 2 , MAE: 0.25531 , MSE: 0.09920 , Spearman Correlation: 0.07149 with p-value: 1.03E-01 , Concordance Index: 0.52361\n",
      "Training at Epoch 3 iteration 0 with loss 0.09509. Total time 0.03361 hours\n",
      "Training at Epoch 3 iteration 100 with loss 0.15386. Total time 0.035 hours\n",
      "Validation at Epoch 3 , MAE: 0.25639 , MSE: 0.09914 , Spearman Correlation: 0.19788 with p-value: 5.45E-06 , Concordance Index: 0.56621\n",
      "Training at Epoch 4 iteration 0 with loss 0.12765. Total time 0.0475 hours\n",
      "Training at Epoch 4 iteration 100 with loss 0.08382. Total time 0.04888 hours\n",
      "Validation at Epoch 4 , MAE: 0.28163 , MSE: 0.10321 , Spearman Correlation: 0.25584 with p-value: 3.24E-09 , Concordance Index: 0.58742\n",
      "Training at Epoch 5 iteration 0 with loss 0.06673. Total time 0.06166 hours\n",
      "Training at Epoch 5 iteration 100 with loss 0.06847. Total time 0.06305 hours\n",
      "Validation at Epoch 5 , MAE: 0.26916 , MSE: 0.10017 , Spearman Correlation: 0.30720 with p-value: 7.93E-13 , Concordance Index: 0.60520\n",
      "Training at Epoch 6 iteration 0 with loss 0.09371. Total time 0.07583 hours\n",
      "Training at Epoch 6 iteration 100 with loss 0.08309. Total time 0.07722 hours\n",
      "Validation at Epoch 6 , MAE: 0.26100 , MSE: 0.09913 , Spearman Correlation: 0.33356 with p-value: 5.58E-15 , Concordance Index: 0.61490\n",
      "Training at Epoch 7 iteration 0 with loss 0.09480. Total time 0.09 hours\n",
      "Training at Epoch 7 iteration 100 with loss 0.08611. Total time 0.09138 hours\n",
      "Validation at Epoch 7 , MAE: 0.24481 , MSE: 0.10022 , Spearman Correlation: 0.35635 with p-value: 5.15E-17 , Concordance Index: 0.62251\n",
      "Training at Epoch 8 iteration 0 with loss 0.10713. Total time 0.10388 hours\n",
      "Training at Epoch 8 iteration 100 with loss 0.11166. Total time 0.10527 hours\n",
      "Validation at Epoch 8 , MAE: 0.28947 , MSE: 0.10581 , Spearman Correlation: 0.35991 with p-value: 2.39E-17 , Concordance Index: 0.62389\n",
      "Training at Epoch 9 iteration 0 with loss 0.09195. Total time 0.11805 hours\n",
      "Training at Epoch 9 iteration 100 with loss 0.09761. Total time 0.11944 hours\n",
      "Validation at Epoch 9 , MAE: 0.26807 , MSE: 0.09973 , Spearman Correlation: 0.39742 with p-value: 4.02E-21 , Concordance Index: 0.63637\n",
      "Training at Epoch 10 iteration 0 with loss 0.09650. Total time 0.13222 hours\n",
      "Training at Epoch 10 iteration 100 with loss 0.10099. Total time 0.13333 hours\n",
      "Validation at Epoch 10 , MAE: 0.25779 , MSE: 0.09849 , Spearman Correlation: 0.37107 with p-value: 2.02E-18 , Concordance Index: 0.62692\n",
      "Training at Epoch 11 iteration 0 with loss 0.09160. Total time 0.14583 hours\n",
      "Training at Epoch 11 iteration 100 with loss 0.09940. Total time 0.14722 hours\n",
      "Validation at Epoch 11 , MAE: 0.27325 , MSE: 0.10046 , Spearman Correlation: 0.38376 with p-value: 1.09E-19 , Concordance Index: 0.63148\n",
      "Training at Epoch 12 iteration 0 with loss 0.07503. Total time 0.15972 hours\n",
      "Training at Epoch 12 iteration 100 with loss 0.12077. Total time 0.16111 hours\n",
      "Validation at Epoch 12 , MAE: 0.26834 , MSE: 0.09918 , Spearman Correlation: 0.39354 with p-value: 1.04E-20 , Concordance Index: 0.63525\n",
      "Training at Epoch 13 iteration 0 with loss 0.09036. Total time 0.17416 hours\n",
      "Training at Epoch 13 iteration 100 with loss 0.08257. Total time 0.17527 hours\n",
      "Validation at Epoch 13 , MAE: 0.26196 , MSE: 0.09754 , Spearman Correlation: 0.37684 with p-value: 5.43E-19 , Concordance Index: 0.62920\n",
      "Training at Epoch 14 iteration 0 with loss 0.12204. Total time 0.18805 hours\n",
      "Training at Epoch 14 iteration 100 with loss 0.09680. Total time 0.18944 hours\n",
      "Validation at Epoch 14 , MAE: 0.27246 , MSE: 0.09869 , Spearman Correlation: 0.38846 with p-value: 3.55E-20 , Concordance Index: 0.63367\n",
      "Training at Epoch 15 iteration 0 with loss 0.07557. Total time 0.20194 hours\n",
      "Training at Epoch 15 iteration 100 with loss 0.11436. Total time 0.20333 hours\n",
      "Validation at Epoch 15 , MAE: 0.24921 , MSE: 0.09432 , Spearman Correlation: 0.40281 with p-value: 1.05E-21 , Concordance Index: 0.63862\n",
      "Training at Epoch 16 iteration 0 with loss 0.13546. Total time 0.21611 hours\n",
      "Training at Epoch 16 iteration 100 with loss 0.05881. Total time 0.2175 hours\n",
      "Validation at Epoch 16 , MAE: 0.24987 , MSE: 0.09116 , Spearman Correlation: 0.39981 with p-value: 2.22E-21 , Concordance Index: 0.63725\n",
      "Training at Epoch 17 iteration 0 with loss 0.08335. Total time 0.23 hours\n",
      "Training at Epoch 17 iteration 100 with loss 0.10740. Total time 0.23111 hours\n",
      "Validation at Epoch 17 , MAE: 0.28126 , MSE: 0.09792 , Spearman Correlation: 0.41559 with p-value: 3.92E-23 , Concordance Index: 0.64271\n",
      "Training at Epoch 18 iteration 0 with loss 0.09387. Total time 0.24388 hours\n",
      "Training at Epoch 18 iteration 100 with loss 0.05974. Total time 0.245 hours\n",
      "Validation at Epoch 18 , MAE: 0.23498 , MSE: 0.08397 , Spearman Correlation: 0.42465 with p-value: 3.50E-24 , Concordance Index: 0.64564\n",
      "Training at Epoch 19 iteration 0 with loss 0.06589. Total time 0.25777 hours\n",
      "Training at Epoch 19 iteration 100 with loss 0.02869. Total time 0.25916 hours\n",
      "Validation at Epoch 19 , MAE: 0.25475 , MSE: 0.08466 , Spearman Correlation: 0.49870 with p-value: 4.77E-34 , Concordance Index: 0.67182\n",
      "Training at Epoch 20 iteration 0 with loss 0.08787. Total time 0.27194 hours\n",
      "Training at Epoch 20 iteration 100 with loss 0.17954. Total time 0.27305 hours\n",
      "Validation at Epoch 20 , MAE: 0.22293 , MSE: 0.07817 , Spearman Correlation: 0.50819 with p-value: 1.70E-35 , Concordance Index: 0.67529\n",
      "--- Go for Testing ---\n",
      "Testing MSE: 0.07588887643926069 , MAE: 0.2166960984938599 , Spearman Correlation: 0.5595528455006911 with p-value: 3.55E-44 , Concordance Index: 0.69479532120367\n",
      "--- Training Finished ---\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model = models.model_initialize(**config)\n",
    "model.train(train, val, test, compute_pos_enc = False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "graph",
   "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": "3.9.18"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}