Pre-download protenix weights from HF Hub; ByteDance CDN is geo-blocked

protenix.tos-cn-beijing.volces.com returns 403 from HF Space IPs.
TMF001/protenix-v2-weights on HF Hub has the same checkpoint (1.77 GB).
Download via hf_hub_download before running inference, skip if already
present. PROTENIX_ROOT_DIR controls where protenix looks for checkpoints.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

app.py

@@ -398,6 +398,34 @@ def toggle_secondary_structure(choice):
 PROTENIX_MODEL = os.environ.get("PROTENIX_MODEL", "protenix-v2")
 PROTENIX_TIMEOUT = int(os.environ.get("PROTENIX_TIMEOUT", "1200"))  # first run: JIT compile + weight download
 
+# Checkpoint directory protenix uses: $PROTENIX_ROOT_DIR/checkpoint (default ~/checkpoint)
+_PROTENIX_CKPT_DIR = Path(os.environ.get("PROTENIX_ROOT_DIR", Path.home())) / "checkpoint"
+
+# Map of model_name → HF Hub repo/file that mirrors the ByteDance CDN
+# (CDN protenix.tos-cn-beijing.volces.com is geo-blocked from HF Spaces)
+_HF_WEIGHT_SOURCES = {
+    "protenix-v2": ("TMF001/protenix-v2-weights", "protenix-v2.pt"),
+}
+
+def _ensure_protenix_weights(model_name: str) -> str | None:
+    """Download model weights from HF Hub if not already present. Returns error str or None."""
+    ckpt_path = _PROTENIX_CKPT_DIR / f"{model_name}.pt"
+    if ckpt_path.exists():
+        return None
+    if model_name not in _HF_WEIGHT_SOURCES:
+        return f"No HF Hub mirror known for model '{model_name}'; ByteDance CDN is geo-blocked from HF Spaces"
+    repo_id, filename = _HF_WEIGHT_SOURCES[model_name]
+    print(f"[protenix-weights] Downloading {filename} from {repo_id} …")
+    try:
+        from huggingface_hub import hf_hub_download
+        _PROTENIX_CKPT_DIR.mkdir(parents=True, exist_ok=True)
+        tmp = hf_hub_download(repo_id=repo_id, filename=filename)
+        shutil.copy(tmp, ckpt_path)
+        print(f"[protenix-weights] Saved to {ckpt_path}")
+        return None
+    except Exception as e:
+        return f"Weight download failed: {e}"
+
 # pLDDT confidence colour bands (AlphaFold convention, B-factor scale 0-100)
 _PLDDT_COLORS = [
     (90, "#0053D6"),   # very high — dark blue
@@ -496,6 +524,11 @@ def fold_with_protenix(sequence, msa_file, model_name):
             output_dir = work_dir / "output"
             output_dir.mkdir()
 
+            # Ensure model weights are present (ByteDance CDN is geo-blocked; use HF Hub mirror)
+            weight_err = _ensure_protenix_weights(model_name)
+            if weight_err:
+                return None, None, None, f"Weight download error: {weight_err}"
+
             # Invoke protenix via Python directly — bypasses the entry-point binary
             # which is sometimes not created during startup installs.
             # Equivalent to: protenix pred -i <input> -o <output> -n <model>