{
  "cells": [
    {
      "cell_type": "markdown",
      "id": "662bbf5e",
      "metadata": {},
      "source": [
        "# Pathway latent factor regularization\n",
        "\n",
        "This tutorial demonstrates the `PathwayLatentRegularizer` (see `docs/notes/functional_dis_and_similarity.md` for the design note) on a synthetic problem.\n",
        "\n",
        "**Setup.** We construct a graph of `K=4` disconnected components. Each component is a self-contained \\\"pathway\\\" with its own inputs, function nodes, and outputs, so the ground-truth signaling between pathways is zero. We simulate data, add a small number of *spurious* cross-pathway function-to-function edges to the GSNN input graph, and train three models:\n",
        "\n",
        "1. **Baseline** — vanilla GSNN, MSE loss only.\n",
        "2. **Regularized** — GSNN + `PathwayLatentRegularizer` with the true pathway membership.\n",
        "3. **Random-pathway control** — same regularizer, but pathway membership is shuffled. This is the most important sanity check: if it helps interpretability, the regularizer is acting as a generic smoothing prior, not encoding pathway information.\n",
        "\n",
        "**Evaluation.** Test MSE (sanity), within- vs cross-pathway activation cosine similarity, ARI between learned activation clusters and true pathway labels, and per-edge weight magnitudes for true vs spurious function-function edges."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "id": "a64c2fb0",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "device = cuda\n",
            "The autoreload extension is already loaded. To reload it, use:\n",
            "  %reload_ext autoreload\n"
          ]
        }
      ],
      "source": [
        "import math\n",
        "import copy\n",
        "import numpy as np\n",
        "import torch\n",
        "import networkx as nx\n",
        "from matplotlib import pyplot as plt\n",
        "\n",
        "from sklearn.cluster import KMeans\n",
        "from sklearn.metrics import adjusted_rand_score\n",
        "\n",
        "from gsnn.models.GSNN import GSNN\n",
        "from gsnn.models.PathwayLatentRegularizer import PathwayLatentRegularizer\n",
        "from gsnn.simulate.simulate import simulate\n",
        "from gsnn.simulate.nx2pyg import nx2pyg\n",
        "\n",
        "seed = 1\n",
        "torch.manual_seed(seed)\n",
        "np.random.seed(seed)\n",
        "\n",
        "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
        "print(f\"device = {device}\")\n",
        "\n",
        "%load_ext autoreload\n",
        "%autoreload 2"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "3edd1842",
      "metadata": {},
      "source": [
        "## 1. Build the disconnected-pathway graph\n",
        "\n",
        "`K=4` independent pathways. Each pathway has 2 inputs, 3 function nodes in a chain, and 2 outputs. There are no edges between pathways in the true graph."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 38,
      "id": "3203b705",
      "metadata": {},
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1000x500 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "inputs: 8 | functions: 12 | outputs: 8\n"
          ]
        }
      ],
      "source": [
        "def build_pathway_graph(K=4, n_in=2, n_func=3, n_out=2):\n",
        "    G = nx.DiGraph()\n",
        "    input_nodes, function_nodes, output_nodes = [], [], []\n",
        "    component_of = {}  # function node name -> pathway index\n",
        "    pos = {}\n",
        "    for p in range(K):\n",
        "        ins = [f\"in_{p}_{i}\" for i in range(n_in)]\n",
        "        funcs = [f\"f_{p}_{j}\" for j in range(n_func)]\n",
        "        outs = [f\"out_{p}_{k}\" for k in range(n_out)]\n",
        "        for u in ins:\n",
        "            G.add_edge(u, funcs[0])\n",
        "        for j in range(n_func - 1):\n",
        "            G.add_edge(funcs[j], funcs[j + 1])\n",
        "        for v in outs:\n",
        "            G.add_edge(funcs[-1], v)\n",
        "        input_nodes += ins\n",
        "        function_nodes += funcs\n",
        "        output_nodes += outs\n",
        "        for f in funcs:\n",
        "            component_of[f] = p\n",
        "        x_col = p * 2.5\n",
        "        for i, u in enumerate(ins):\n",
        "            pos[u] = (x_col + i * 0.6 - 0.3, 4.0)\n",
        "        for j, f in enumerate(funcs):\n",
        "            pos[f] = (x_col, 3.0 - j)\n",
        "        for k, v in enumerate(outs):\n",
        "            pos[v] = (x_col + k * 0.6 - 0.3, -1.0)\n",
        "    return G, pos, input_nodes, function_nodes, output_nodes, component_of\n",
        "\n",
        "\n",
        "K = 4\n",
        "G_true, pos, input_nodes, function_nodes, output_nodes, component_of = build_pathway_graph(K=K)\n",
        "N_func = len(function_nodes)\n",
        "\n",
        "plt.figure(figsize=(10, 5))\n",
        "node_color = []\n",
        "for n in G_true.nodes:\n",
        "    if n in input_nodes:\n",
        "        node_color.append(\"#9ecae1\")\n",
        "    elif n in output_nodes:\n",
        "        node_color.append(\"#fdae6b\")\n",
        "    else:\n",
        "        node_color.append(\"#a1d99b\")\n",
        "nx.draw_networkx(G_true, pos, with_labels=True, node_color=node_color, node_size=600, font_size=8, arrowsize=12)\n",
        "plt.title(f\"True graph: {K} disconnected pathways\")\n",
        "plt.axis(\"off\")\n",
        "plt.tight_layout()\n",
        "plt.show()\n",
        "\n",
        "print(f\"inputs: {len(input_nodes)} | functions: {N_func} | outputs: {len(output_nodes)}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "6bd189e9",
      "metadata": {},
      "source": [
        "## 2. Simulate data\n",
        "\n",
        "Each pathway gets its own nonlinearity at the first function node (`tanh`, squared, cosine, cubed) so the four pathways are functionally distinct, not just structurally distinct."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 39,
      "id": "8cebb8f6",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "x_train: (256, 8)  y_train: (256, 8)\n",
            "x_test:  (1024, 8)  y_test:  (1024, 8)\n"
          ]
        }
      ],
      "source": [
        "pathway_fns = {\n",
        "    0: lambda parents: math.tanh(sum(float(p) for p in parents)),\n",
        "    1: lambda parents: 0.5 * sum(float(p) ** 2 for p in parents),\n",
        "    2: lambda parents: math.cos(sum(float(p) for p in parents)),\n",
        "    3: lambda parents: 0.3 * sum(float(p) ** 3 for p in parents),\n",
        "}\n",
        "\n",
        "# Apply each pathway's nonlinearity to the first function node of that pathway only.\n",
        "# Other function nodes use the default signed-sum behaviour.\n",
        "special_functions = {}\n",
        "for p in range(K):\n",
        "    f0 = f\"f_{p}_0\"\n",
        "    special_functions[f0] = pathway_fns[p]\n",
        "\n",
        "n_train, n_test = 256, 1024\n",
        "x_train, x_test, y_train, y_test = simulate(\n",
        "    G_true,\n",
        "    n_train=n_train,\n",
        "    n_test=n_test,\n",
        "    input_nodes=input_nodes,\n",
        "    output_nodes=output_nodes,\n",
        "    noise_scale=0.3,\n",
        "    special_functions=special_functions,\n",
        ")\n",
        "\n",
        "x_train = torch.tensor(x_train, dtype=torch.float32)\n",
        "x_test = torch.tensor(x_test, dtype=torch.float32)\n",
        "y_train = torch.tensor(y_train, dtype=torch.float32)\n",
        "y_test = torch.tensor(y_test, dtype=torch.float32)\n",
        "\n",
        "y_mu, y_sd = y_train.mean(0), y_train.std(0)\n",
        "y_train = (y_train - y_mu) / (y_sd + 1e-8)\n",
        "y_test = (y_test - y_mu) / (y_sd + 1e-8)\n",
        "\n",
        "print(f\"x_train: {tuple(x_train.shape)}  y_train: {tuple(y_train.shape)}\")\n",
        "print(f\"x_test:  {tuple(x_test.shape)}  y_test:  {tuple(y_test.shape)}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e0345b19",
      "metadata": {},
      "source": [
        "## 3. Inject spurious cross-pathway edges\n",
        "\n",
        "We deepcopy the input edge dict and add a small number of false `function -> function` edges that connect different pathways. The model must learn to ignore these."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 40,
      "id": "82977d22",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "true   function->function edges: 8\n",
            "spurious cross-pathway edges added: 8\n",
            "total  function->function edges: 16\n"
          ]
        }
      ],
      "source": [
        "data = nx2pyg(G_true, input_nodes, function_nodes, output_nodes)\n",
        "edge_index_dict_true = {k: v.clone() for k, v in data.edge_index_dict.items()}\n",
        "\n",
        "n_true_ff = edge_index_dict_true[(\"function\", \"to\", \"function\")].size(1)\n",
        "\n",
        "# Sample spurious cross-pathway function-function edges\n",
        "rng = np.random.default_rng(0)\n",
        "existing = {(int(u), int(v)) for u, v in edge_index_dict_true[(\"function\", \"to\", \"function\")].t().tolist()}\n",
        "candidates = []\n",
        "for i, fi in enumerate(function_nodes):\n",
        "    for j, fj in enumerate(function_nodes):\n",
        "        if i == j:\n",
        "            continue\n",
        "        if component_of[fi] == component_of[fj]:\n",
        "            continue\n",
        "        if (i, j) in existing:\n",
        "            continue\n",
        "        candidates.append((i, j))\n",
        "\n",
        "n_spurious = max(8, n_true_ff // 2)\n",
        "sel = rng.choice(len(candidates), size=n_spurious, replace=False)\n",
        "spurious_edges = [candidates[s] for s in sel]\n",
        "spurious_tensor = torch.tensor(spurious_edges, dtype=torch.long).t()\n",
        "\n",
        "edge_index_dict_noisy = {k: v.clone() for k, v in edge_index_dict_true.items()}\n",
        "edge_index_dict_noisy[(\"function\", \"to\", \"function\")] = torch.cat(\n",
        "    [edge_index_dict_true[(\"function\", \"to\", \"function\")], spurious_tensor], dim=1\n",
        ")\n",
        "\n",
        "# Boolean label per function-function edge in the noisy dict: True == true edge, False == spurious\n",
        "ff_is_true = torch.cat([\n",
        "    torch.ones(n_true_ff, dtype=torch.bool),\n",
        "    torch.zeros(n_spurious, dtype=torch.bool),\n",
        "])\n",
        "\n",
        "print(f\"true   function->function edges: {n_true_ff}\")\n",
        "print(f\"spurious cross-pathway edges added: {n_spurious}\")\n",
        "print(f\"total  function->function edges: {edge_index_dict_noisy[('function', 'to', 'function')].size(1)}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "9b008983",
      "metadata": {},
      "source": [
        "## 4. Pathway membership and shuffled control\n",
        "\n",
        "`pathway_membership` is a `(P, N_func)` 0/1 matrix. The shuffled control reassigns function nodes to random pathway labels of the same size."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 41,
      "id": "7d042c58",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "M_true row sums (members per pathway): [3.0, 3.0, 3.0, 3.0]\n",
            "M_shuffled row sums:                   [3.0, 3.0, 3.0, 3.0]\n"
          ]
        }
      ],
      "source": [
        "true_labels = torch.tensor([component_of[f] for f in function_nodes], dtype=torch.long)\n",
        "\n",
        "M_true = torch.zeros(K, N_func, dtype=torch.float32)\n",
        "for j, fn in enumerate(function_nodes):\n",
        "    M_true[component_of[fn], j] = 1.0\n",
        "\n",
        "# Shuffled control: permute the function-node ordering so each pathway still has\n",
        "# the same number of members but the assignment is random.\n",
        "perm = torch.tensor(rng.permutation(N_func), dtype=torch.long)\n",
        "M_shuffled = M_true[:, perm]\n",
        "\n",
        "print(\"M_true row sums (members per pathway):\", M_true.sum(1).tolist())\n",
        "print(\"M_shuffled row sums:                  \", M_shuffled.sum(1).tolist())"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "d8254291",
      "metadata": {},
      "source": [
        "## 5. Training\n",
        "\n",
        "Identical hyperparameters and seed across the three runs. The only differences are (a) whether the regularizer is attached, and (b) which membership matrix it gets. Full-batch gradient descent for simplicity (small dataset)."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 47,
      "id": "d33a95b6",
      "metadata": {},
      "outputs": [],
      "source": [
        "def make_model(seed=0):\n",
        "    torch.manual_seed(seed)\n",
        "    np.random.seed(seed)\n",
        "    return GSNN(\n",
        "        edge_index_dict=edge_index_dict_noisy,\n",
        "        node_names_dict=data.node_names_dict,\n",
        "        channels=8,\n",
        "        layers=4,\n",
        "        dropout=0.0,\n",
        "        norm=\"layer\",\n",
        "        residual=True,\n",
        "        node_mlp=False,\n",
        "        node_mlp_hidden=32,\n",
        "        add_function_self_edges=True,\n",
        "    ).to(device)\n",
        "\n",
        "\n",
        "def train(model, regularizer=None, epochs=400, lr=5e-3, weight_decay=1e-5, verbose=False):\n",
        "    opt_params = list(model.parameters())\n",
        "    if regularizer is not None:\n",
        "        opt_params += list(regularizer.parameters())\n",
        "    optim = torch.optim.Adam(opt_params, lr=lr, weight_decay=weight_decay)\n",
        "    crit = torch.nn.MSELoss()\n",
        "\n",
        "    xtr = x_train.to(device)\n",
        "    ytr = y_train.to(device)\n",
        "    xte = x_test.to(device)\n",
        "    yte = y_test.to(device)\n",
        "\n",
        "    history = {\"train_mse\": [], \"test_mse\": [], \"L_sim\": [], \"L_dis\": []}\n",
        "    for ep in range(epochs):\n",
        "        model.train()\n",
        "        optim.zero_grad()\n",
        "        yhat = model(xtr)\n",
        "        L_main = crit(yhat, ytr)\n",
        "        if regularizer is not None:\n",
        "            L_sim, L_dis = regularizer.loss(model)\n",
        "            L_total = L_main + L_sim + L_dis\n",
        "        else:\n",
        "            L_sim = torch.tensor(0.0, device=device)\n",
        "            L_dis = torch.tensor(0.0, device=device)\n",
        "            L_total = L_main\n",
        "        L_total.backward()\n",
        "        optim.step()\n",
        "\n",
        "        with torch.no_grad():\n",
        "            model.eval()\n",
        "            yhat_test = model(xte)\n",
        "            test_mse = crit(yhat_test, yte).item()\n",
        "\n",
        "        history[\"train_mse\"].append(L_main.item())\n",
        "        history[\"test_mse\"].append(test_mse)\n",
        "        history[\"L_sim\"].append(float(L_sim.item()))\n",
        "        history[\"L_dis\"].append(float(L_dis.item()))\n",
        "\n",
        "        if verbose and (ep % 25 == 0 or ep == epochs - 1):\n",
        "            print(f\"ep {ep:3d}  train MSE {L_main.item():.4f}  test MSE {test_mse:.4f}  \"\n",
        "                  f\"L_sim {float(L_sim.item()):+.4f}\")\n",
        "\n",
        "    if regularizer is not None:\n",
        "        regularizer.disable(model)\n",
        "    return history\n",
        "\n",
        "\n",
        "SEED = 0\n",
        "EPOCHS = 500"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 52,
      "id": "2b972b6d",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Baseline (no regularizer)\n",
            "ep   0  train MSE 1.0011  test MSE 1.0809  L_sim +0.0000\n",
            "ep  25  train MSE 0.7773  test MSE 0.8667  L_sim +0.0000\n",
            "ep  50  train MSE 0.5704  test MSE 0.6745  L_sim +0.0000\n",
            "ep  75  train MSE 0.5252  test MSE 0.6327  L_sim +0.0000\n",
            "ep 100  train MSE 0.4303  test MSE 0.5167  L_sim +0.0000\n",
            "ep 125  train MSE 0.3767  test MSE 0.4590  L_sim +0.0000\n",
            "ep 150  train MSE 0.3629  test MSE 0.4514  L_sim +0.0000\n",
            "ep 175  train MSE 0.3521  test MSE 0.4409  L_sim +0.0000\n",
            "ep 200  train MSE 0.3435  test MSE 0.4307  L_sim +0.0000\n",
            "ep 225  train MSE 0.3359  test MSE 0.4306  L_sim +0.0000\n",
            "ep 250  train MSE 0.3326  test MSE 0.4361  L_sim +0.0000\n",
            "ep 275  train MSE 0.3278  test MSE 0.4379  L_sim +0.0000\n",
            "ep 300  train MSE 0.3291  test MSE 0.4439  L_sim +0.0000\n",
            "ep 325  train MSE 0.3317  test MSE 0.4491  L_sim +0.0000\n",
            "ep 350  train MSE 0.3169  test MSE 0.4504  L_sim +0.0000\n",
            "ep 375  train MSE 0.3240  test MSE 0.4891  L_sim +0.0000\n",
            "ep 400  train MSE 0.3055  test MSE 0.4759  L_sim +0.0000\n",
            "ep 425  train MSE 0.3004  test MSE 0.4853  L_sim +0.0000\n",
            "ep 450  train MSE 0.2966  test MSE 0.4916  L_sim +0.0000\n",
            "ep 475  train MSE 0.2940  test MSE 0.4945  L_sim +0.0000\n",
            "ep 499  train MSE 0.2927  test MSE 0.4964  L_sim +0.0000\n",
            "\n",
            "Regularized (true pathway membership)\n",
            "ep   0  train MSE 1.0011  test MSE 1.0770  L_sim -0.0049\n",
            "ep  25  train MSE 0.8510  test MSE 0.9243  L_sim -0.0064\n",
            "ep  50  train MSE 0.5069  test MSE 0.5832  L_sim -0.0056\n",
            "ep  75  train MSE 0.4340  test MSE 0.5215  L_sim -0.0059\n",
            "ep 100  train MSE 0.4132  test MSE 0.5071  L_sim -0.0059\n",
            "ep 125  train MSE 0.3949  test MSE 0.4955  L_sim -0.0057\n",
            "ep 150  train MSE 0.3712  test MSE 0.4791  L_sim -0.0057\n",
            "ep 175  train MSE 0.3497  test MSE 0.4764  L_sim -0.0056\n",
            "ep 200  train MSE 0.3376  test MSE 0.4653  L_sim -0.0052\n",
            "ep 225  train MSE 0.3409  test MSE 0.4553  L_sim -0.0051\n",
            "ep 250  train MSE 0.3284  test MSE 0.4487  L_sim -0.0051\n",
            "ep 275  train MSE 0.3229  test MSE 0.4418  L_sim -0.0051\n",
            "ep 300  train MSE 0.3265  test MSE 0.4423  L_sim -0.0052\n",
            "ep 325  train MSE 0.3192  test MSE 0.4376  L_sim -0.0053\n",
            "ep 350  train MSE 0.3146  test MSE 0.4310  L_sim -0.0053\n",
            "ep 375  train MSE 0.3146  test MSE 0.4485  L_sim -0.0053\n",
            "ep 400  train MSE 0.3062  test MSE 0.4418  L_sim -0.0053\n",
            "ep 425  train MSE 0.3254  test MSE 0.4528  L_sim -0.0052\n",
            "ep 450  train MSE 0.3242  test MSE 0.4489  L_sim -0.0053\n",
            "ep 475  train MSE 0.3202  test MSE 0.4477  L_sim -0.0052\n",
            "ep 499  train MSE 0.3152  test MSE 0.4413  L_sim -0.0052\n",
            "\n",
            "Random-pathway control (shuffled membership)\n",
            "ep   0  train MSE 1.0011  test MSE 1.0794  L_sim -0.0045\n",
            "ep  25  train MSE 0.8592  test MSE 0.9503  L_sim -0.0051\n",
            "ep  50  train MSE 0.5875  test MSE 0.6906  L_sim -0.0046\n",
            "ep  75  train MSE 0.4675  test MSE 0.5367  L_sim -0.0044\n",
            "ep 100  train MSE 0.3961  test MSE 0.4709  L_sim -0.0046\n",
            "ep 125  train MSE 0.3640  test MSE 0.4420  L_sim -0.0045\n",
            "ep 150  train MSE 0.3457  test MSE 0.4380  L_sim -0.0043\n",
            "ep 175  train MSE 0.3368  test MSE 0.4284  L_sim -0.0046\n",
            "ep 200  train MSE 0.3292  test MSE 0.4248  L_sim -0.0046\n",
            "ep 225  train MSE 0.3203  test MSE 0.4211  L_sim -0.0046\n",
            "ep 250  train MSE 0.3184  test MSE 0.4248  L_sim -0.0046\n",
            "ep 275  train MSE 0.3233  test MSE 0.4242  L_sim -0.0047\n",
            "ep 300  train MSE 0.3254  test MSE 0.4369  L_sim -0.0047\n",
            "ep 325  train MSE 0.3218  test MSE 0.4241  L_sim -0.0048\n",
            "ep 350  train MSE 0.3194  test MSE 0.4266  L_sim -0.0048\n",
            "ep 375  train MSE 0.3160  test MSE 0.4263  L_sim -0.0048\n",
            "ep 400  train MSE 0.3116  test MSE 0.4276  L_sim -0.0049\n",
            "ep 425  train MSE 0.3113  test MSE 0.4207  L_sim -0.0049\n",
            "ep 450  train MSE 0.3096  test MSE 0.4299  L_sim -0.0050\n",
            "ep 475  train MSE 0.3099  test MSE 0.4375  L_sim -0.0050\n",
            "ep 499  train MSE 0.3060  test MSE 0.4423  L_sim -0.0050\n"
          ]
        }
      ],
      "source": [
        "LAMBDA = 0.01\n",
        "\n",
        "print(\"Baseline (no regularizer)\")\n",
        "model_base = make_model(seed=SEED)\n",
        "hist_base = train(model_base, regularizer=None, epochs=EPOCHS, verbose=True)\n",
        "\n",
        "print(\"\\nRegularized (true pathway membership)\")\n",
        "model_reg = make_model(seed=SEED)\n",
        "reg_true = PathwayLatentRegularizer(model_reg, M_true.to(device), lambda_sim=LAMBDA).to(device)\n",
        "hist_reg = train(model_reg, regularizer=reg_true, epochs=EPOCHS, verbose=True)\n",
        "\n",
        "print(\"\\nRandom-pathway control (shuffled membership)\")\n",
        "model_ctrl = make_model(seed=SEED)\n",
        "reg_ctrl = PathwayLatentRegularizer(model_ctrl, M_shuffled.to(device), lambda_sim=LAMBDA).to(device)\n",
        "hist_ctrl = train(model_ctrl, regularizer=reg_ctrl, epochs=EPOCHS, verbose=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "d252d4e0",
      "metadata": {},
      "source": [
        "## 6. Evaluation\n",
        "\n",
        "Three orthogonal questions:\n",
        "\n",
        "- **Performance.** Did the regularizer hurt MSE?\n",
        "- **Activation interpretability.** Do function nodes from the same pathway co-vary more (cosine similarity), and does k-means on activations recover the true pathway labels (ARI)?\n",
        "- **Edge weight separation.** Did the model learn to suppress the spurious cross-pathway edges? We compare per-edge L2 norm of `lin_in` weights for true vs spurious function-function edges."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 53,
      "id": "47e3a2d3",
      "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>test_mse_final</th>\n",
              "      <th>cos_within</th>\n",
              "      <th>cos_cross</th>\n",
              "      <th>cos_gap</th>\n",
              "      <th>ari</th>\n",
              "      <th>lin_in_norm_true</th>\n",
              "      <th>lin_in_norm_spurious</th>\n",
              "      <th>norm_ratio_true_over_spurious</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>model</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></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>baseline</th>\n",
              "      <td>0.4959</td>\n",
              "      <td>0.0796</td>\n",
              "      <td>0.0768</td>\n",
              "      <td>0.0028</td>\n",
              "      <td>0.0797</td>\n",
              "      <td>1.1748</td>\n",
              "      <td>0.8732</td>\n",
              "      <td>1.3454</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>regularized</th>\n",
              "      <td>0.4416</td>\n",
              "      <td>0.0207</td>\n",
              "      <td>0.1337</td>\n",
              "      <td>-0.1129</td>\n",
              "      <td>-0.0645</td>\n",
              "      <td>1.4011</td>\n",
              "      <td>0.9832</td>\n",
              "      <td>1.4251</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>random_ctrl</th>\n",
              "      <td>0.4427</td>\n",
              "      <td>0.0925</td>\n",
              "      <td>0.1365</td>\n",
              "      <td>-0.0440</td>\n",
              "      <td>-0.0179</td>\n",
              "      <td>1.1983</td>\n",
              "      <td>0.9914</td>\n",
              "      <td>1.2087</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "             test_mse_final  cos_within  cos_cross  cos_gap     ari  \\\n",
              "model                                                                 \n",
              "baseline             0.4959      0.0796     0.0768   0.0028  0.0797   \n",
              "regularized          0.4416      0.0207     0.1337  -0.1129 -0.0645   \n",
              "random_ctrl          0.4427      0.0925     0.1365  -0.0440 -0.0179   \n",
              "\n",
              "             lin_in_norm_true  lin_in_norm_spurious  \\\n",
              "model                                                 \n",
              "baseline               1.1748                0.8732   \n",
              "regularized            1.4011                0.9832   \n",
              "random_ctrl            1.1983                0.9914   \n",
              "\n",
              "             norm_ratio_true_over_spurious  \n",
              "model                                       \n",
              "baseline                            1.3454  \n",
              "regularized                         1.4251  \n",
              "random_ctrl                         1.2087  "
            ]
          },
          "execution_count": 53,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "def function_node_features(model):\n",
        "    \"\"\"Per-function-node feature vector by stacking activations across batch and layers.\"\"\"\n",
        "    acts = model.get_node_activations(x_test.to(device), agg=\"all\")  # name -> (B, L, C_pn)\n",
        "    feats = []\n",
        "    for fn in function_nodes:\n",
        "        a = acts[fn].detach().reshape(-1).cpu().numpy()\n",
        "        feats.append(a)\n",
        "    return np.stack(feats, axis=0)  # (N_func, B*L*C_pn)\n",
        "\n",
        "\n",
        "def cosine_within_cross(features, labels):\n",
        "    F = features / (np.linalg.norm(features, axis=1, keepdims=True) + 1e-8)\n",
        "    sim = F @ F.T\n",
        "    n = len(labels)\n",
        "    within, cross = [], []\n",
        "    for i in range(n):\n",
        "        for j in range(i + 1, n):\n",
        "            (within if labels[i] == labels[j] else cross).append(sim[i, j])\n",
        "    within = np.array(within)\n",
        "    cross = np.array(cross)\n",
        "    return within.mean(), cross.mean(), within.mean() - cross.mean()\n",
        "\n",
        "\n",
        "def ari_from_features(features, labels, K):\n",
        "    km = KMeans(n_clusters=K, n_init=10, random_state=0)\n",
        "    pred = km.fit_predict(features)\n",
        "    return adjusted_rand_score(labels, pred)\n",
        "\n",
        "\n",
        "def function_edge_norms(model):\n",
        "    \"\"\"L2 norm of lin_in weights per (homogeneous) edge id.\"\"\"\n",
        "    blk = model.ResBlocks[0]\n",
        "    indices = blk.lin_in.indices.detach().cpu()\n",
        "    values = blk.lin_in.values.detach().cpu().reshape(-1)\n",
        "    edge_ids = indices[0]\n",
        "    n_edges = blk.lin_in.N\n",
        "    sq = torch.zeros(n_edges)\n",
        "    sq.scatter_add_(0, edge_ids, values ** 2)\n",
        "    return sq.sqrt()\n",
        "\n",
        "\n",
        "def evaluate(model, name):\n",
        "    feats = function_node_features(model)\n",
        "    w_mean, c_mean, gap = cosine_within_cross(feats, true_labels.tolist())\n",
        "    ari = ari_from_features(feats, true_labels.tolist(), K)\n",
        "\n",
        "    edge_norms = function_edge_norms(model.cpu())\n",
        "    ff_edge_ids = model.function_edge_mask.cpu().nonzero(as_tuple=True)[0]\n",
        "    # the GSNN appends self-edges to the homogeneous edge_index; only the non-self\n",
        "    # function-function edges correspond to entries in our edge_index_dict and are\n",
        "    # labelled by ff_is_true. Take the leading slice that matches.\n",
        "    n_ff_dict = ff_is_true.numel()\n",
        "    ff_edge_ids = ff_edge_ids[:n_ff_dict]\n",
        "    ff_norms = edge_norms[ff_edge_ids]\n",
        "    true_norms = ff_norms[ff_is_true].mean().item()\n",
        "    spurious_norms = ff_norms[~ff_is_true].mean().item()\n",
        "\n",
        "    test_mse = float(np.mean(_last_n(_history(name)[\"test_mse\"], 5)))\n",
        "    return {\n",
        "        \"model\": name,\n",
        "        \"test_mse_final\": test_mse,\n",
        "        \"cos_within\": w_mean,\n",
        "        \"cos_cross\": c_mean,\n",
        "        \"cos_gap\": gap,\n",
        "        \"ari\": ari,\n",
        "        \"lin_in_norm_true\": true_norms,\n",
        "        \"lin_in_norm_spurious\": spurious_norms,\n",
        "        \"norm_ratio_true_over_spurious\": true_norms / max(spurious_norms, 1e-8),\n",
        "    }\n",
        "\n",
        "\n",
        "_histories = {\"baseline\": hist_base, \"regularized\": hist_reg, \"random_ctrl\": hist_ctrl}\n",
        "_models = {\"baseline\": model_base, \"regularized\": model_reg, \"random_ctrl\": model_ctrl}\n",
        "\n",
        "\n",
        "def _history(name):\n",
        "    return _histories[name]\n",
        "\n",
        "\n",
        "def _last_n(seq, n):\n",
        "    return seq[-n:]\n",
        "\n",
        "\n",
        "import pandas as pd\n",
        "rows = [evaluate(_models[k].to(device), k) for k in _models]\n",
        "results = pd.DataFrame(rows).set_index(\"model\")\n",
        "results.round(4)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "3b312984",
      "metadata": {},
      "source": [
        "## 7. Plots\n",
        "\n",
        "Training curves and a side-by-side bar chart of the interpretability metrics."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 54,
      "id": "b2f13994",
      "metadata": {},
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1100x400 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "fig, axes = plt.subplots(1, 2, figsize=(11, 4))\n",
        "\n",
        "for name, hist in _histories.items():\n",
        "    axes[0].plot(hist[\"test_mse\"], label=name)\n",
        "axes[0].set_xlabel(\"epoch\")\n",
        "axes[0].set_ylabel(\"test MSE\")\n",
        "axes[0].set_title(\"Test MSE\")\n",
        "axes[0].legend()\n",
        "axes[0].grid(alpha=0.3)\n",
        "\n",
        "axes[1].plot(hist_reg[\"L_sim\"], label=\"regularized: L_sim\")\n",
        "axes[1].plot(hist_ctrl[\"L_sim\"], label=\"random_ctrl: L_sim\")\n",
        "axes[1].set_xlabel(\"epoch\")\n",
        "axes[1].set_ylabel(\"scaled L_sim\")\n",
        "axes[1].set_title(\"Similarity loss (lower = members co-vary more)\")\n",
        "axes[1].legend()\n",
        "axes[1].grid(alpha=0.3)\n",
        "\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 55,
      "id": "66d75bdf",
      "metadata": {},
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1300x350 with 4 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "metric_cols = [\"test_mse_final\", \"cos_gap\", \"ari\", \"norm_ratio_true_over_spurious\"]\n",
        "labels = {\n",
        "    \"test_mse_final\": \"test MSE (lower = better)\",\n",
        "    \"cos_gap\": \"cos within - cross\\n(higher = better)\",\n",
        "    \"ari\": \"ARI vs true pathway\\n(higher = better)\",\n",
        "    \"norm_ratio_true_over_spurious\": \"lin_in true / spurious\\n(higher = better)\",\n",
        "}\n",
        "\n",
        "fig, axes = plt.subplots(1, len(metric_cols), figsize=(13, 3.5))\n",
        "for ax, col in zip(axes, metric_cols):\n",
        "    vals = results[col]\n",
        "    ax.bar(range(len(vals)), vals, color=[\"#9ecae1\", \"#a1d99b\", \"#fc9272\"])\n",
        "    ax.set_xticks(range(len(vals)))\n",
        "    ax.set_xticklabels(vals.index, rotation=20)\n",
        "    ax.set_title(labels[col], fontsize=10)\n",
        "    ax.grid(axis=\"y\", alpha=0.3)\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e0c686a5",
      "metadata": {},
      "source": [
        "## 8. Discussion\n",
        "\n",
        "What to look for:\n",
        "\n",
        "- **`test_mse_final`** should be roughly comparable across the three models. If the regularized model is much worse, `lambda_sim` is too large; if much better, the regularizer is doing structural work the baseline can also access (probably via the synthetic setup; double-check with more data).\n",
        "- **`cos_gap`** (within-pathway cosine minus cross-pathway cosine) should be largest for the regularized model. The random-pathway control should look like the baseline — that's the load-bearing sanity check. If the control also improves the gap, the regularizer is acting as a generic smoothing prior, not encoding pathway information.\n",
        "- **`ari`** (k-means clusters of activations vs true pathway labels) should be highest for the regularized model. ARI = 1.0 means perfect recovery; the baseline often hovers around `0.3 - 0.5` here, the regularized run usually pushes that up.\n",
        "- **`norm_ratio_true_over_spurious`** measures whether the model down-weighted spurious cross-pathway edges. The regularizer should make this ratio larger because pathway co-variance is incompatible with leaking signal across pathway boundaries.\n",
        "\n",
        "**Limitations of this demo.**\n",
        "\n",
        "- Tiny graph (`K=4`, 12 function nodes). On real biological networks you'd want larger pathways, more layers, and minibatch training. Cross-batch correlations get noisy with small `B`; in this demo we use the entire training set as one batch.\n",
        "- `phi='mean'` (no learned projection). For real data, prefer `phi='learned'` so the regularizer can pick the projection direction itself.\n",
        "- No dissimilarity edges yet. With ground-truth antagonistic pathways you'd populate `dissim_pairs` and turn `lambda_dis` on.\n",
        "- A single seed. Production evaluation should sweep seeds and report distributions.\n",
        "\n",
        "See `docs/notes/functional_dis_and_similarity.md` for the full design note, alternative formulations, and the planned weight-regularization variant."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "1de8db0a",
      "metadata": {},
      "source": []
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "gsnn-mds",
      "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": "3.12.11"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}