app.py

import torch
import torch.nn.functional as F
from torch_geometric.nn import GCNConv, global_mean_pool
from torch_geometric.utils import from_networkx
from torch_geometric.data import Data

import networkx as nx
import gradio as gr
import matplotlib.pyplot as plt
from PIL import Image
import io

# ---------- MODELO ----------

class GCN(torch.nn.Module):
    def __init__(self, hidden_channels=64):
        super().__init__()
        self.conv1 = GCNConv(7, hidden_channels)
        self.conv2 = GCNConv(hidden_channels, hidden_channels)
        self.lin = torch.nn.Linear(hidden_channels, 2)

    def forward(self, x, edge_index, batch):
        x = self.conv1(x, edge_index)
        x = F.relu(x)
        x = self.conv2(x, edge_index)
        x = F.relu(x)
        x = global_mean_pool(x, batch)
        return self.lin(x)

# ---------- CARGA DEL MODELO ENTRENADO ----------

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model = GCN().to(device)
model.load_state_dict(torch.load("model_gcn.pth", map_location=device))
model.eval()

# ---------- FUNCIONES AUXILIARES ----------

def parse_input(num_nodes, edges_str, node_features_str):
    G = nx.Graph()

    try:
        # Añadir nodos
        for i in range(num_nodes):
            G.add_node(i)

        # Parsear aristas
        edges = eval(edges_str)
        G.add_edges_from(edges)

        # Parsear características
        node_features = eval(node_features_str)
        if len(node_features) != num_nodes or any(len(f) != 7 for f in node_features):
            raise ValueError("Cada nodo debe tener exactamente 7 características.")

        nx.set_node_attributes(G, {i: node_features[i] for i in range(num_nodes)}, "x")
        return G

    except Exception as e:
        raise gr.Error(f"Error en los datos del grafo: {e}")

def draw_graph(G, pred_label):
    pos = nx.spring_layout(G)
    node_colors = ['lightgreen' if pred_label == 1 else 'lightcoral'] * G.number_of_nodes()

    plt.figure(figsize=(4, 4))
    nx.draw(G, pos, with_labels=True, node_color=node_colors, edge_color='gray', node_size=800)
    plt.title("Grafo de entrada")

    buf = io.BytesIO()
    plt.savefig(buf, format='png')
    plt.close()
    buf.seek(0)
    return Image.open(buf)

def predict_graph(num_nodes, edges_str, node_features_str):
    G = parse_input(num_nodes, edges_str, node_features_str)

    data = from_networkx(G)
    data.x = torch.tensor([v for v in nx.get_node_attributes(G, "x").values()], dtype=torch.float)
    data.edge_index = data.edge_index
    data.batch = torch.tensor([0] * data.num_nodes)

    data = data.to(device)
    with torch.no_grad():
        out = model(data.x, data.edge_index, data.batch)
        pred = out.argmax(dim=1).item()

    label_text = "Mutagénico ✅" if pred == 1 else "No mutagénico ❌"
    return label_text, draw_graph(G, pred)

# ---------- INTERFAZ GRADIO ----------

description = """
Clasificador molecular basado en **Redes Neuronales en Grafo (GNN)** entrenado sobre el dataset **MUTAG**.

✏️ Puedes modificar el grafo:  
- Número de nodos  
- Aristas (formato: `[(0,1),(1,2)]`)  
- Características de cada nodo (7 valores binarios por nodo)  
"""

inputs = [
    gr.Number(label="Número de nodos", value=3, precision=0),
    gr.Textbox(label="Aristas", value="[(0,1),(1,2)]"),
    gr.Textbox(label="Características por nodo", value="[[1,0,0,1,0,1,0],[0,1,1,0,1,0,1],[1,1,0,0,1,0,1]]")
]

outputs = [
    gr.Text(label="Predicción"),
    gr.Image(label="Visualización del grafo")
]

demo = gr.Interface(
    fn=predict_graph,
    inputs=inputs,
    outputs=outputs,
    title="🧪 Clasificador Molecular con GCN",
    description=description,
    examples=[
        [3, "[(0,1),(1,2)]", "[[1,0,0,1,0,1,0],[0,1,1,0,1,0,1],[1,1,0,0,1,0,1]]"],
        [4, "[(0,1),(1,2),(2,3)]", "[[1,1,0,1,0,0,1],[0,0,1,1,1,0,0],[1,0,1,0,1,1,0],[0,1,0,1,1,0,1]]"]
    ]
)

demo.launch(show_error=True)