PathoPred / app.py
rahafahmad89's picture
Update app.py
a1db08b verified
Raw History Blame Contribute Delete
8.07 kB
import os
import json
import traceback
import joblib
import numpy as np
import pandas as pd
import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import shap
import gradio as gr
# ============================================================
# PATCH: Fix gradio_client boolean JSON-schema crash
# ============================================================
try:
import gradio_client.utils as _gcu
if hasattr(_gcu, "_json_schema_to_python_type"):
_orig = _gcu._json_schema_to_python_type
def _patched(schema, defs=None):
# JSON Schema allows boolean schemas (True / False)
if isinstance(schema, bool):
return "Any"
return _orig(schema, defs)
_gcu._json_schema_to_python_type = _patched
except Exception:
pass
# ============================================================
# Artifacts
# ============================================================
MODEL_PATH = "model_extratrees_ds4.joblib"
FEATS_PATH = "feature_names.json"
model = joblib.load(MODEL_PATH)
with open(FEATS_PATH) as f:
FEATURES = json.load(f)
LABEL_NAME = {0: "Benign", 1: "Pathogenic"}
THRESHOLD = 0.7
ID_CANDIDATES = [
"#Uploaded_Variation", "#Uploaded_variation", "uploaded_variation",
"Uploaded_Variation", "Uploaded_variation", "variant_id",
"Variant_ID", "variation_id", "Variant", "ID", "HGVSc", "HGVSp"
]
# ============================================================
# Helpers
# ============================================================
def _clean_headers(df):
df.columns = (
df.columns.astype(str)
.str.replace(r"[\u200b\ufeff]", "", regex=True)
.str.strip()
)
return df
def _read_table(file):
name = getattr(file, "name", str(file))
if name.lower().endswith((".xlsx", ".xls")):
df = pd.read_excel(name, engine="openpyxl")
else:
df = pd.read_csv(
name,
engine="python",
sep=None,
dtype=str,
na_values=["", "NA", "N/A"],
quoting=3,
on_bad_lines="skip",
encoding_errors="ignore",
)
return _clean_headers(df)
def _coerce_numeric(df):
for c in df.columns:
df[c] = pd.to_numeric(df[c], errors="coerce")
return df
def _prepare(df):
for c in FEATURES:
if c not in df.columns:
df[c] = np.nan
return _coerce_numeric(df[FEATURES].copy())
def _validate_columns(df):
return [c for c in FEATURES if c not in df.columns]
def _pick_id_series(df):
norm = {c.lower(): c for c in df.columns}
for cand in ID_CANDIDATES:
if cand.lower() in norm:
col = norm[cand.lower()]
return df[col].astype(str)
return pd.Series(df.index, name="row_index")
# ============================================================
# Prediction
# ============================================================
def predict_csv(file):
try:
if file is None:
return None, "โŒ No file uploaded."
df = _read_table(file)
missing = _validate_columns(df)
if missing:
return None, f"โŒ Missing {len(missing)} features. Example: {missing[:10]}"
X = _prepare(df)
proba = model.predict_proba(X)[:, 1]
pred = (proba > THRESHOLD).astype(int)
ids = _pick_id_series(df)
out = pd.DataFrame({
"Variant ID": ids,
"Predicted probability (pathogenic)": proba,
"Predicted class": np.where(
pred == 1, "๐Ÿ”ด Pathogenic", "๐ŸŸข Benign"
),
})
summary = (
f"โœ… Processed {len(out)} variants \n"
f"๐Ÿ”ด Pathogenic: {(pred == 1).sum()} | "
f"๐ŸŸข Benign: {(pred == 0).sum()} \n"
f"Threshold = {THRESHOLD}"
)
return out, summary
except Exception:
return None, "โŒ ERROR:\n```" + traceback.format_exc() + "```"
# ============================================================
# SHAP Explanation
# ============================================================
def explain_first_row(file):
try:
if file is None:
return None, "โŒ No file uploaded."
df = _read_table(file)
missing = _validate_columns(df)
if missing:
return None, "โŒ Missing required features for explanation."
X = _prepare(df)
if len(X) == 0:
return None, "โŒ No rows found."
explainer = shap.Explainer(model.predict_proba, X)
shap_values = explainer(X.iloc[[0]])
vals = shap_values.values[0, :, 1]
s = pd.Series(vals, index=FEATURES).sort_values(
key=np.abs, ascending=False
).head(15)
fig = plt.figure(figsize=(7, 5))
s.iloc[::-1].plot(kind="barh")
plt.title("Top feature contributions (Sample #1)")
plt.tight_layout()
return fig, (
"๐Ÿ” **Why this prediction?** \n"
"Positive bars increase the probability of pathogenicity; "
"negative bars decrease it. \n"
"These signals are *supportive only* and must be interpreted "
"within ACMG/AMP guidelines."
)
except Exception:
return None, "โŒ ERROR:\n```" + traceback.format_exc() + "```"
# ============================================================
# UI
# ============================================================
with gr.Blocks(
title="Pathogenicity Predictor",
theme=gr.themes.Soft(primary_hue="rose", secondary_hue="slate"),
) as demo:
gr.Markdown("""
# ๐Ÿงฌ Pathogenicity Predictor
**Research prototype for variant pathogenicity assessment**
Upload a CSV or Excel file containing the trained feature columns.
This tool provides *supportive* predictions and explanations.
""")
with gr.Row():
with gr.Column(scale=1):
gr.Markdown("### ๐Ÿ“‚ Input")
inp = gr.File(
label="Upload CSV / XLSX",
file_types=[".csv", ".xlsx", ".xls"]
)
gr.Markdown("""
**Requirements**
- One row = one variant
- Must include all trained feature columns
- Results are **not** clinical diagnoses
""")
with gr.Column(scale=2):
gr.Markdown("### ๐Ÿ“Š Results")
with gr.Row():
btn_pred = gr.Button("๐Ÿ”ฎ Predict", variant="primary")
btn_expl = gr.Button("๐Ÿ” Explain first variant")
out_df = gr.Dataframe(
label="Predictions",
interactive=False,
wrap=True,
)
out_txt = gr.Markdown()
gr.Markdown("### ๐Ÿ”ฌ Why this prediction?")
out_plot = gr.Plot(label="Feature contribution (SHAP)")
btn_pred.click(predict_csv, inp, [out_df, out_txt], show_progress=True)
btn_expl.click(explain_first_row, inp, [out_plot, out_txt])
# ========================================================
# Feedback section (email)
# ========================================================
gr.Markdown("""
---
### ๐Ÿ’ฌ Feedback
We would love to hear your thoughts, suggestions, or bug reports.
๐Ÿ“ง **[Send feedback via email](mailto:dr.rahafahmad@gmail.com?subject=Pathogenicity%20Predictor%20Feedback)**
Click on the link or email us directly at **dr.rahafahmad@gmail.com, 700039113@uaeu.ac.ae, bassam.ali@uaeu.ac.ae**
""")
gr.Markdown("""
---
โš ๏ธ **Research prototype**
This tool is for research and educational use only.
Not intended for clinical diagnosis or medical decision-making.
ยฉ 2026
""")
# ============================================================
# Launch (Spaces-safe)
# ============================================================
if __name__ == "__main__":
port = int(os.environ.get("PORT", 7860))
demo.queue()
demo.launch(server_name="0.0.0.0", server_port=port, debug=True)