import gradio as gr
import joblib
import shap
import numpy as np
import pandas as pd
import json

model = joblib.load("model_v6.pkl")

with open("feature_order.json") as f:
    features = json.load(f)

explainer = shap.TreeExplainer(model)

def predict(log_AF, hydro_diff, protein_pos, charge_change, grantham):

    X = pd.DataFrame([{
        "log_AF": log_AF,
        "Hydro_diff": hydro_diff,
        "Protein_pos_norm": protein_pos,
        "Charge_change": charge_change,
        "Grantham": grantham
    }])

    prob = model.predict_proba(X)[0][1]
    shap_vals = explainer.shap_values(X)[0]

    explanation = ""
    for i, feat in enumerate(features):
        explanation += f"{feat}: value={X.iloc[0][feat]:.4f}, SHAP={shap_vals[i]:+.4f}\n"

    return f"Pathogenic Probability: {prob:.4f}", explanation

demo = gr.Interface(
    fn=predict,
    inputs=[
        gr.Number(label="log_AF"),
        gr.Number(label="Hydrophobicity Difference"),
        gr.Number(label="Protein Position Normalized (0–1)"),
        gr.Number(label="Charge Change (0 or 1)"),
        gr.Number(label="Grantham Score")
    ],
    outputs=[
        gr.Textbox(label="Prediction"),
        gr.Textbox(label="Feature Contributions")
    ],
    title="Explainable Mutation Pathogenicity Model v6",
    description="XGBoost + SHAP Explainable Variant Predictor"
)

demo.launch()