NauRaa/Gene_Expression_Clustering_App

1import gradio as gr
2import pandas as pd
3import joblib
4import numpy as np
5
6# ===========================
7# Load Saved Model & Objects
8# ===========================
9model = joblib.load("best_cluster_model.joblib")
10pca = joblib.load("pca_transform.joblib")
11scaler = joblib.load("scaler.joblib")
12selector = joblib.load("selector.joblib")
13
14# ===========================
15# Prediction Function
16# ===========================
17def predict_clusters(file):
18    try:
19        df = pd.read_csv(file.name)
20    except Exception as e:
21        return f"❌ Error loading file: {e}"
22
23    # Drop non-numeric columns
24    df = df.select_dtypes(include=[np.number])
25    
26    if df.empty:
27        return "❌ No numeric columns found in the dataset."
28
29    # 🔧 Remove unseen columns (like 'Cluster')
30    df = df.loc[:, [c for c in df.columns if c in selector.feature_names_in_]]
31
32    # Feature selection
33    X_selected = selector.transform(df)
34    
35    # Scaling
36    X_scaled = scaler.transform(X_selected)
37    
38    # PCA transformation
39    X_pca = pca.transform(X_scaled)
40    
41    # Prediction
42    try:
43        labels = model.predict(X_pca)
44    except:
45        labels = model.fit_predict(X_pca)
46    
47    # Create output dataframe
48    df_out = pd.DataFrame({
49        "PC1": X_pca[:, 0],
50        "PC2": X_pca[:, 1],
51        "Cluster": labels
52    })
53    
54    return df_out
55
56# ===========================
57# Gradio Interface
58# ===========================
59interface = gr.Interface(
60    fn=predict_clusters,
61    inputs=gr.File(label="📂 Upload your gene expression CSV file"),
62    outputs=gr.Dataframe(label="🧬 Clustered Results (PCA + Cluster)"),
63    title="🧫 Gene Expression Clustering App",
64    description="Upload a gene expression CSV file. The app will perform scaling, PCA, and clustering using the trained unsupervised model."
65)
66
67if __name__ == "__main__":
68    interface.launch()
69