Upload 3 files

app.py

@@ -0,0 +1,55 @@
+# app.py
+import gradio as gr
+import joblib
+import numpy as np
+from collections import Counter
+from typing import List
+
+# helper: k-mer extraction / vectorize (k=3)
+def kmer_counts(seq: str, k=3):
+    seq = seq.strip().upper()
+    counts = Counter()
+    if len(seq) < k:
+        return counts
+    for i in range(len(seq)-k+1):
+        counts[seq[i:i+k]] += 1
+    return counts
+
+def vectorize_single(seq: str, vocab: List[str], k=3):
+    x = np.zeros((1, len(vocab)), dtype=float)
+    c = kmer_counts(seq, k)
+    for j,kmer in enumerate(vocab):
+        x[0,j] = c.get(kmer, 0)
+    return x
+
+# load model+vocab (mutation_model.joblib must be uploaded too)
+model, vocab = joblib.load("mutation_model.joblib")
+
+def predict_sequence(sequence: str):
+    if not sequence or len(sequence.strip()) < 3:
+        return {"error":"sequence too short"}
+    X = vectorize_single(sequence, vocab=vocab, k=3)
+    pred = model.predict(X)[0]
+    prob = float(model.predict_proba(X).max()) if hasattr(model, "predict_proba") else None
+    return {
+        "sequence": sequence,
+        "mutation_detected": bool(pred),
+        "confidence": prob
+    }
+
+# Gradio UI
+with gr.Blocks() as demo:
+    gr.Markdown("# DNA Mutation Detector (Quick Space)")
+    seq_in = gr.Textbox(label="DNA sequence", placeholder="ATGCGTACGTTAGC...")
+    btn = gr.Button("Analyze")
+    out = gr.JSON()
+    btn.click(fn=predict_sequence, inputs=seq_in, outputs=out)
+
+# Expose a simple inference API endpoint (Gradio provides /api/predict automatically)
+# but we also expose a programmatic function name for convenience:
+def api_predict(payload: dict):
+    seq = payload.get("sequence", "")
+    return predict_sequence(seq)
+
+if __name__ == "__main__":
+    demo.launch()

mutation_model.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cecba933e452e5a139a2f7f7cb85b35976128c42f20e48e26013f3cabfc56d75
+size 305695

requirements.txt

@@ -0,0 +1,4 @@
+gradio>=3.0
+joblib
+numpy
+scikit-learn