app.py

import numpy as np
import pandas as pd
import gradio as gr
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score, classification_report
import joblib

# Generate more realistic synthetic genetic data
def generate_realistic_genetic_data(n_samples=1000):
    np.random.seed(42)

    # Define genetic markers associated with different conditions
    # Marker ranges based on typical genetic variation patterns
    genetic_data = {
        'BRCA1_mutation': np.random.choice([0, 1], size=n_samples, p=[0.95, 0.05]),  # BRCA1 mutation
        'P53_variation': np.random.normal(0.5, 0.1, n_samples),  # P53 tumor suppressor variation
        'APOE_allele': np.random.choice([2, 3, 4], size=n_samples, p=[0.1, 0.7, 0.2]),  # APOE allele types
        'DNA_methylation': np.random.beta(2, 5, n_samples),  # DNA methylation levels
        'telomere_length': np.random.normal(6000, 1000, n_samples),  # Telomere length
        'CYP2D6_activity': np.random.gamma(2, 2, n_samples),  # CYP2D6 enzyme activity
        'inflammatory_markers': np.random.exponential(2, n_samples),  # Inflammatory markers
        'glucose_metabolism': np.random.normal(100, 15, n_samples),  # Glucose metabolism
        'oxidative_stress': np.random.gamma(3, 1, n_samples),  # Oxidative stress levels
        'immune_response': np.random.normal(0.7, 0.1, n_samples)  # Immune response strength
    }

    # Create DataFrame
    df = pd.DataFrame(genetic_data)

    # Generate disease status based on complex interactions
    disease_probability = (
        0.3 * genetic_data['BRCA1_mutation'] +
        0.2 * (genetic_data['P53_variation'] > 0.7) +
        0.15 * (genetic_data['APOE_allele'] == 4) +
        0.1 * (genetic_data['DNA_methylation'] > 0.6) +
        0.05 * (genetic_data['telomere_length'] < 5000) +
        0.1 * (genetic_data['CYP2D6_activity'] > 5) +
        0.05 * (genetic_data['inflammatory_markers'] > 3) +
        0.05 * (genetic_data['glucose_metabolism'] > 120)
    )

    df['disease'] = (disease_probability > 0.5).astype(int)

    return df

# Data preprocessing
def preprocess_data(data):
    X = data.drop('disease', axis=1)
    y = data['disease']

    scaler = StandardScaler()
    X_scaled = scaler.fit_transform(X)

    return X_scaled, y, scaler

# Train and evaluate model
def train_and_evaluate_model():
    # Generate and preprocess data
    print("Generating synthetic genetic data...")
    data = generate_realistic_genetic_data(1500)
    print("\nData Sample:")
    print(data.head())
    print("\nData Statistics:")
    print(data.describe())

    # Preprocess data
    X_scaled, y, scaler = preprocess_data(data)

    # Split data
    X_train, X_test, y_train, y_test = train_test_split(
        X_scaled, y, test_size=0.2, random_state=42
    )

    # Train model
    print("\nTraining Random Forest model...")
    model = RandomForestClassifier(
        n_estimators=100,
        max_depth=5,
        random_state=42
    )
    model.fit(X_train, y_train)

    # Evaluate model
    y_pred = model.predict(X_test)
    accuracy = accuracy_score(y_test, y_pred)
    print("\nModel Evaluation:")
    print(f"Accuracy: {accuracy:.2f}")
    print("\nClassification Report:")
    print(classification_report(y_test, y_pred))

    # Feature importance
    feature_importance = pd.DataFrame({
        'feature': data.drop('disease', axis=1).columns,
        'importance': model.feature_importances_
    })
    print("\nFeature Importance:")
    print(feature_importance.sort_values('importance', ascending=False))

    return model, scaler

# Prediction function
def predict_disease(
    brca1_mutation, p53_variation, apoe_allele, dna_methylation,
    telomere_length, cyp2d6_activity, inflammatory_markers,
    glucose_metabolism, oxidative_stress, immune_response
):
    # Create input array
    input_data = np.array([
        brca1_mutation, p53_variation, apoe_allele, dna_methylation,
        telomere_length, cyp2d6_activity, inflammatory_markers,
        glucose_metabolism, oxidative_stress, immune_response
    ]).reshape(1, -1)

    # Scale input data
    scaled_data = scaler.transform(input_data)

    # Make prediction
    prediction = model.predict_proba(scaled_data)[0]

    return {
        "No Disease": float(prediction[0]),
        "Disease": float(prediction[1])
    }

# Train model and get scaler
print("Training model and preparing interface...")
model, scaler = train_and_evaluate_model()

# Create Gradio interface
iface = gr.Interface(
    fn=predict_disease,
    inputs=[
        gr.Number(label="BRCA1 Mutation (0 or 1)", value=0),
        gr.Number(label="P53 Variation (typically 0.3-0.7)", value=0.5),
        gr.Number(label="APOE Allele (2, 3, or 4)", value=3),
        gr.Number(label="DNA Methylation (0-1)", value=0.4),
        gr.Number(label="Telomere Length (typically 4000-8000)", value=6000),
        gr.Number(label="CYP2D6 Activity (typically 0-10)", value=4),
        gr.Number(label="Inflammatory Markers (typically 0-10)", value=2),
        gr.Number(label="Glucose Metabolism (typically 70-130)", value=100),
        gr.Number(label="Oxidative Stress (typically 0-10)", value=3),
        gr.Number(label="Immune Response (typically 0.5-0.9)", value=0.7)
    ],
    outputs=gr.Label(label="Disease Prediction"),
    title="Genetic Disease Prediction System",
    description="""This system predicts genetic disease risk based on various genetic markers and biological indicators.
                   Please input values within the suggested ranges for accurate predictions.""",
    examples=[
        # High-risk example
        [1, 0.8, 4, 0.7, 4800, 6, 4, 125, 5, 0.6],
        # Low-risk example
        [0, 0.4, 3, 0.3, 6500, 3, 1, 95, 2, 0.8],
        # Moderate-risk example
        [0, 0.6, 3, 0.5, 5500, 4, 2, 110, 3, 0.7]
    ]
)

# Launch the interface
iface.launch(share=True)