notebooks/shallow_embeddings.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "d15b2796-90a7-4f8b-bebe-15d297f146e6",
   "metadata": {},
   "source": [
    "# Course 27666 AI-guided Protein Science\n",
    "\n",
    "# Graph Machine Learning\n",
    "\n",
    "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/Multiomics-Analytics-Group/course_graph_machine_learning/blob/main/notebooks/shallow_embeddings.ipynb)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e45c5870-15f7-4ad2-9707-ede33291676b",
   "metadata": {},
   "source": [
    "# Shallow Embeddings Methods\n",
    "## Node2Vec and DeepWalk with PyTorch Geometric (PyG)\n",
    "\n",
    "\n",
    "\n",
    "## Install PyG"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "05aae071-9691-4e97-a659-18f74c27e0c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Uncomment and run the following commands if PyG is not installed.\n",
    "#!pip install torch torch-geometric matplotlib torch-cluster scikit-learn"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d0cac26-880c-4978-a547-ee1927dc59a5",
   "metadata": {},
   "source": [
    "## 1. Helper Functions for Visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "c030c200-1d25-4760-97a8-4191923246f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import torch\n",
    "from torch_geometric.nn import Node2Vec\n",
    "import networkx as nx\n",
    "from torch_geometric.utils import from_networkx\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.manifold import TSNE\n",
    "\n",
    "def visualize_embeddings(embeddings, labels):\n",
    "    \"\"\"\n",
    "    Visualizes node embeddings using t-SNE.\n",
    "\n",
    "    Args:\n",
    "        embeddings (torch.Tensor): Node embeddings of shape [num_nodes, embedding_dim].\n",
    "        labels (torch.Tensor): Node labels of shape [num_nodes].\n",
    "    \"\"\"\n",
    "    \n",
    "    # Apply t-SNE to reduce dimensions to 2D\n",
    "    tsne = TSNE(n_components=2, random_state=42)\n",
    "    embeddings_2d = tsne.fit_transform(embeddings.detach().cpu().numpy())\n",
    "\n",
    "    # Scatter plot of embeddings\n",
    "    plt.figure(figsize=(8, 8))\n",
    "    scatter = plt.scatter(embeddings_2d[:, 0], embeddings_2d[:, 1], c=labels, cmap=\"Set1\", s=50)\n",
    "    plt.legend(*scatter.legend_elements(), title=\"Classes\")\n",
    "    plt.title(\"t-SNE Visualization of Node Embeddings\")\n",
    "    plt.show()\n",
    "\n",
    "def create_sample_graph():\n",
    "    \"\"\"\n",
    "    Creates a sample NetworkX graph and converts it to a PyTorch Geometric Data object.\n",
    "\n",
    "    Returns:\n",
    "        data (torch_geometric.data.Data): The PyG Data object containing the graph.\n",
    "        labels (torch.Tensor): Ground-truth node labels for visualization.\n",
    "    \"\"\"\n",
    "  \n",
    "    G = nx.karate_club_graph()\n",
    "\n",
    "    # Assign labels based on the club membership\n",
    "    labels = torch.tensor([G.nodes[n][\"club\"] == \"Mr. Hi\" for n in G.nodes], dtype=torch.long)\n",
    "\n",
    "    # Convert to PyG Data object\n",
    "    data = from_networkx(G)\n",
    "    data.y = labels  # Add labels to the graph data\n",
    "\n",
    "    return data, labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7f7ca457-4791-4a3e-84cd-2eda8b3ab2cb",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"Creating the graph...\")\n",
    "data, labels = create_sample_graph()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bc6082a9-c529-40d8-b7d0-81aa231059d6",
   "metadata": {},
   "source": [
    "## 2. Define Model Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "aa746a8c-58b8-4d08-a0b4-565cfcf47870",
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_dim=64 # Determines the size of the learned vector representations (embeddings) for each node in the graph\n",
    "walk_length=20 # This is a key parameter to control how deep the walks go in the graph.\n",
    "context_size=10 # Defines how many neighboring nodes (within a given distance from the current node) are used to predict the target node.\n",
    "walks_per_node=10 # Defines how many random walks are generated per node in the graph.\n",
    "num_epochs=100 "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a8fe060c-61c8-4dcf-b082-4f1a4b35d441",
   "metadata": {},
   "source": [
    "## 3. Node2Vec Implementation\n",
    "\n",
    "#### Instantiate the Node2Vec model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "a58b9264-8bbb-4b13-9984-8ccf9269dc80",
   "metadata": {},
   "outputs": [],
   "source": [
    "# p - # Return parameter controls the likelihood of immediately revisiting the previous node during a random walk. \n",
    "# q In-out\" parameter controls the likelihood of exploring nodes that are close or far from the current node. \n",
    "model = Node2Vec(\n",
    "    edge_index=data.edge_index,\n",
    "    embedding_dim=embedding_dim,\n",
    "    walk_length=walk_length,\n",
    "    context_size=context_size,\n",
    "    walks_per_node=walks_per_node,\n",
    "    num_negative_samples=1,\n",
    "    p=0.2,\n",
    "    q=0.5,\n",
    "    sparse=True, #This parameter indicates whether to store the embeddings in sparse format or not (True - computationally efficient)\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "709a3e31-4ac8-4ca7-9c83-b4c224af4b5a",
   "metadata": {},
   "source": [
    "### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "9d3294b8-2744-4cf7-b8ab-c731fd22c2ab",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0, Loss: 6.942193984985352\n",
      "Epoch 10, Loss: 5.5647382736206055\n",
      "Epoch 20, Loss: 4.2929840087890625\n",
      "Epoch 30, Loss: 3.0488717555999756\n",
      "Epoch 40, Loss: 2.8861241340637207\n",
      "Epoch 50, Loss: 2.5792930126190186\n",
      "Epoch 60, Loss: 2.5770885944366455\n",
      "Epoch 70, Loss: 2.2524054050445557\n",
      "Epoch 80, Loss: 2.141613006591797\n",
      "Epoch 90, Loss: 2.279317855834961\n"
     ]
    }
   ],
   "source": [
    "optimizer = torch.optim.SparseAdam(list(model.parameters()), lr=0.01)\n",
    "model.train()\n",
    "for epoch in range(num_epochs):\n",
    "    optimizer.zero_grad()\n",
    "    \n",
    "    # Sample random walks for positive and negative samples using a specific node batch\n",
    "    batch_size = 32  # Set the batch size for sampling\n",
    "    num_batches = (data.num_nodes + batch_size - 1) // batch_size  # Calculate the number of batches\n",
    "    \n",
    "    # Train on each batch of nodes\n",
    "    for i in range(num_batches):\n",
    "        # Get the current batch of nodes\n",
    "        batch_nodes = torch.arange(i * batch_size, min((i + 1) * batch_size, data.num_nodes))\n",
    "        \n",
    "        # Sample random walks for the current batch of nodes\n",
    "        # pos_rw would contain the positive samples (the node itself and its neighbors)\n",
    "        #neg_rw contains the negative samples (nodes that are not neighbors, usually sampled randomly)\n",
    "        pos_rw, neg_rw = model.sample(batch_nodes)  # Pass a batch of nodes for sampling\n",
    "        \n",
    "        # Calculate the loss for the current batch\n",
    "        loss = model.loss(pos_rw, neg_rw)\n",
    "        loss.backward()\n",
    "    \n",
    "    # Update model parameters after each batch\n",
    "    optimizer.step()\n",
    "    \n",
    "    # Print the loss at every 10th epoch\n",
    "    if epoch % 10 == 0:\n",
    "        print(f\"Epoch {epoch}, Loss: {loss.item()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "19bae470-7cd4-4829-a070-4d9332659654",
   "metadata": {},
   "source": [
    "### Get the embeddings after training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "b2266ffb-a80b-4645-9946-1dc5f22b54a7",
   "metadata": {},
   "outputs": [],
   "source": [
    "node2vec_embeddings = model.embedding.weight"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "83719d42-c856-4da5-b7c6-3e2a20affcf7",
   "metadata": {},
   "source": [
    "## 4. Visualising Node2Vec Embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "25c67197-bc4c-465d-945b-56d203b8d439",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Training Node2Vec...\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x800 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualize_embeddings(node2vec_embeddings, labels)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f0db4e06-cf8f-4a56-9560-785ab9d8598d",
   "metadata": {},
   "source": [
    "## 5. DeepWalk as a Special Case of Node2Vec\n",
    "\n",
    "### Initialize Node2Vec with p=1 and q=1 (DeepWalk settings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "6f6c1551-dbdc-421e-b05c-80f44deeacfd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# p - # Return parameter controls the likelihood of immediately revisiting the previous node during a random walk. \n",
    "# A value of 1.0 means there's no bias towards revisiting nodes, making the walk more random.\n",
    "# q In-out\" parameter controls the likelihood of exploring nodes that are close or far from the current node. \n",
    "# A value of 1.0 means that there's no bias towards either direction, and the walk is more balanced.\n",
    "model = Node2Vec(\n",
    "    edge_index=data.edge_index,\n",
    "    embedding_dim=embedding_dim,\n",
    "    walk_length=walk_length,\n",
    "    context_size=context_size,\n",
    "    walks_per_node=walks_per_node,\n",
    "    num_negative_samples=1,\n",
    "    p=1.0,  # Return parameter\n",
    "    q=1.0,  # In-out parameter\n",
    "    sparse=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "417cbc92-a913-4af4-8a29-1ba7a46a2bf9",
   "metadata": {},
   "source": [
    "### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "91ed0c57-05f6-4934-849c-d256ea35c801",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0, Loss: 2.217787504196167\n",
      "Epoch 10, Loss: 1.9960956573486328\n",
      "Epoch 20, Loss: 1.953965425491333\n",
      "Epoch 30, Loss: 2.0914649963378906\n",
      "Epoch 40, Loss: 1.9845225811004639\n",
      "Epoch 50, Loss: 2.122910737991333\n",
      "Epoch 60, Loss: 2.0817089080810547\n",
      "Epoch 70, Loss: 2.1663856506347656\n",
      "Epoch 80, Loss: 2.1824443340301514\n",
      "Epoch 90, Loss: 2.147550582885742\n"
     ]
    }
   ],
   "source": [
    "optimizer = torch.optim.SparseAdam(list(model.parameters()), lr=0.01)\n",
    "\n",
    "model.train()\n",
    "for epoch in range(num_epochs):\n",
    "    optimizer.zero_grad()\n",
    "    \n",
    "    # Sample random walks for positive and negative samples using a specific node batch\n",
    "    batch_size = 32  # Set the batch size for sampling\n",
    "    num_batches = (data.num_nodes + batch_size - 1) // batch_size  # Calculate the number of batches\n",
    "    \n",
    "    # Train on each batch of nodes\n",
    "    for i in range(num_batches):\n",
    "        # Get the current batch of nodes\n",
    "        batch_nodes = torch.arange(i * batch_size, min((i + 1) * batch_size, data.num_nodes))\n",
    "        \n",
    "        # Sample random walks for the current batch of nodes\n",
    "        pos_rw, neg_rw = model.sample(batch_nodes)  # Pass a batch of nodes for sampling\n",
    "        \n",
    "        # Calculate the loss for the current batch\n",
    "        loss = model.loss(pos_rw, neg_rw)\n",
    "        loss.backward()\n",
    "    \n",
    "    # Update model parameters after each batch\n",
    "    optimizer.step()\n",
    "    \n",
    "    # Print the loss at every 10th epoch\n",
    "    if epoch % 10 == 0:\n",
    "        print(f\"Epoch {epoch}, Loss: {loss.item()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2384e88a-84d0-4ee3-bc87-547c28b5e566",
   "metadata": {},
   "source": [
    "### Get the embeddings after training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "86333431-a7ed-4c9e-bd53-1e9f6937baf2",
   "metadata": {},
   "outputs": [],
   "source": [
    "deepwalk_embeddings = model.embedding.weight  # Corrected line to access learned embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6d5a50ee-ef31-4751-8957-d5b5ecd7fb7e",
   "metadata": {},
   "source": [
    "## 4. Running and Visualizing the Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "4edc4358-1708-41b2-83e3-3ad6cabdcf31",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Visualizing DeepWalk Embeddings...\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x800 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"\\nVisualizing DeepWalk Embeddings...\")\n",
    "visualize_embeddings(deepwalk_embeddings, labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0ca71964-8419-4796-ac8a-d5677cead1cd",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}