boltz2-python-client 0.5.0__tar.gz → 0.5.2.post1__tar.gz

boltz2_python_client-0.5.2.post1/CHANGELOG.md

@@ -0,0 +1,215 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+## [0.5.2.post1] - 2026-02-24
+
+PEP 440 post-release patch on top of `0.5.2`. **No public API changes**;
+fixes a long-standing packaging defect that has been present since
+`0.5.0` (and was carried into `0.5.2` unchanged). Users on a clean
+environment should upgrade to `0.5.2.post1`; users who already have
+`pandas` installed will see no behavior difference.
+
+### Fixed — packaging
+- **`import boltz2_client` now works on a clean install.** Previously
+  `boltz2_client/virtual_screening.py` did `import pandas as pd` at
+  module top, but `pandas` is only declared in the `[dev]` extra (not a
+  runtime dependency), so any clean `pip install boltz2-python-client`
+  followed by `import boltz2_client` raised
+  `ModuleNotFoundError: No module named 'pandas'`. The pandas import is
+  now deferred into the four methods that actually need it
+  (`CompoundLibrary.from_csv`, `ScreeningResults.to_dataframe`,
+  `ScreeningResults.get_top_hits`,
+  `ScreeningResults.get_statistics_by_group`); calling any of those
+  methods still requires pandas, but importing the package, the CLI,
+  and every other code path no longer does. Callers who use the virtual
+  screening DataFrame helpers should install with
+  `pip install "boltz2-python-client[dev]"` or add `pandas` to their
+  own environment.
+- A `from __future__ import annotations` was added to
+  `virtual_screening.py` so the `pd.DataFrame` return-type annotations
+  on those methods are evaluated lazily and do not require `pandas` at
+  import time. Static type checkers and IDEs continue to resolve the
+  annotations via a `TYPE_CHECKING` block.
+
+## [0.5.2] - 2026-02-24
+
+### Fixed — `MultiEndpointClient` reliability
+- **No longer cascades into "All endpoints failed" after a single endpoint glitch.**
+  `failed_requests` is now reset on every successful call across all
+  `predict_*` methods (`predict`, `predict_sync`, `predict_protein_structure`,
+  `predict_protein_ligand_complex`, `predict_covalent_complex`,
+  `predict_dna_protein_complex`, `predict_with_advanced_parameters`,
+  `predict_from_yaml_config`, `predict_from_yaml_file`, and the
+  corresponding `*_sync` variants). Previously a recovered endpoint could
+  remain marked unhealthy indefinitely after three transient failures.
+- **`_select_endpoint()` now honours the per-request `attempted_endpoints`
+  set.** The `LEAST_LOADED` and `RANDOM` strategies could otherwise re-pick
+  the same dead endpoint repeatedly inside a single dispatch loop, wasting
+  retries and (on `LEAST_LOADED`) starving healthy endpoints. Round-robin
+  also now skips already-attempted endpoints fairly without resetting the
+  global rotation counter.
+- **`health_check_sync()` now normalises `HealthStatus` to a boolean.**
+  Previously it stored the entire `HealthStatus` object on
+  `EndpointStatus.is_healthy`, which is always truthy and broke the sync
+  recovery + status-table paths.
+- **Background health-check loop runs an immediate probe at startup.**
+  Initial unreachability is now detected without waiting a full
+  `health_check_interval` (default 60 s).
+- **Synchronous `__exit__` no longer crashes when no event loop is running.**
+  `with MultiEndpointClient(...)` is now safe in plain synchronous code.
+
+### Fixed — CLI
+- `boltz2 msa-search`, `boltz2 msa-predict`, and `boltz2 msa-ligand` now
+  exit with a non-zero status when the underlying call fails (previously
+  they printed the error and exited cleanly, masking failures in pipelines).
+- `boltz2 screen` now reports a clear error if client initialization fails
+  or if `--pocket-residues` is malformed, instead of crashing with a stack
+  trace before the progress UI starts.
+- Removed a redundant local `import json` inside
+  `boltz2 multimer-msa --save-all`; the module-level import is used.
+
+### Fixed — virtual screening
+- `VirtualScreening` no longer raises `IndexError` when a prediction
+  response comes back with empty `structures`, `confidence_scores`, or
+  affinity score lists; missing fields are reported as `None` and the
+  campaign continues.
+
+### Fixed — async client
+- `Boltz2Client._sagemaker_predict` now uses `asyncio.get_running_loop()`
+  instead of the deprecated `asyncio.get_event_loop()` (silences the
+  Python 3.10+ `DeprecationWarning` and avoids a future-version hard
+  error).
+
+### Added
+- **`boltz2 --version` / `boltz2 -V`** prints the installed package
+  version (previously the CLI had no version flag).
+
+### Removed
+- Dead `_HAS_VISUALIZATION` / `_HAS_ANALYSIS` optional-import scaffolding
+  in `boltz2_client/__init__.py` referencing modules that have never
+  shipped. The package public surface is unchanged: no name was ever
+  successfully exported from those blocks.
+
+### Docs
+- Updated the README tagline to accurately describe what ships in the
+  package (multi-endpoint load balancing + notebook examples that
+  visualize structures with `py3Dmol` and Molstar) instead of claiming a
+  non-existent "built-in 3D visualization" module.
+
+### Tests
+- Added `tests/test_multi_endpoint_reliability.py` with 11 regression
+  tests pinning each of the fixes above. Each test fails on 0.5.1 and
+  passes on 0.5.2; the full suite (excluding live-endpoint tests) is now
+  134 passing.
+
+## [0.5.1] - 2026-04-07
+
+### Fixed
+- CLI `metadata` command now shows a clear error for SageMaker endpoints instead of a cryptic `NoneType` error
+- SageMaker endpoint exception handling cleanup in CLI
+
+## [0.5.0] - 2026-04-07
+
+### Added
+- AWS SageMaker endpoint support (`EndpointType.SAGEMAKER`, CLI `--endpoint-type sagemaker`, `--sagemaker-endpoint-name`, `--sagemaker-region`)
+
+### Changed
+- **PocketConstraint API** now uses `Contact` objects (binder + contacts)
+- **Parameter ranges:** `recycling_steps` 1–10, `diffusion_samples` 1–25
+- **`write_full_pde`** support for full predicted distance error matrix output
+- **`Ligand.id`** is optional
+- Compatibility updates for **Boltz-2 NIM v1.6.0**
+
+### Fixed
+- Comprehensive test suite fixes and live endpoint tests
+
+## [0.4.0] - 2025-11-01
+
+### Added
+- A3M to multimer MSA conversion utilities, `multimer-msa` and `convert-msa` CLI commands
+- Example script `13_a3m_to_multimer_csv.py` and guides (`docs/a3m_to_multimer_msa.md`)
+- Multimer notebooks (`01_multimer_prediction.ipynb`, `03_colabfold_a3m_to_multimer.ipynb`) and related example scripts
+- `08_affinity_prediction_simple.py` as the maintained affinity example (with `kinase_y7w_affinity.json`)
+
+### Changed
+- Multi-endpoint client default load-balancing strategy documented as **least-loaded**; expanded multi-endpoint coverage across prediction APIs
+- README and guides updated for v1.5 NIM parameter limits (recycling/diffusion), PAE/PDE output, and PDB export
+
+### Documentation
+- Multi-endpoint virtual screening guide and examples (`comprehensive_multi_endpoint_demo.py`, `multi_endpoint_screening.py`)
+
+## [0.3.0] - 2025-09-11
+
+### Added
+- **GPU-accelerated MSA Search NIM Integration**:
+  - New `MSASearchClient` for direct MSA Search NIM interaction
+  - `MSAFormatConverter` for A3M, FASTA, and Stockholm format support
+  - `MSASearchIntegration` for Boltz-2 workflow integration
+- **New Client Methods**:
+  - `configure_msa_search()` - Configure MSA Search NIM endpoint
+  - `search_msa()` - Standalone MSA search
+  - `predict_with_msa_search()` - Integrated MSA + structure prediction
+  - `predict_ligand_with_msa_search()` - MSA + ligand + affinity prediction
+  - `batch_msa_search()` - Batch processing for multiple sequences
+- **CLI Commands**:
+  - `boltz2 msa-search` - Search and save MSA alignments
+  - `boltz2 msa-predict` - Combined MSA search + structure prediction
+  - `boltz2 msa-ligand` - MSA-guided ligand affinity prediction
+  - `--msa-file` option for existing MSA files
+- **Examples**:
+  - `10_msa_search_integration.py` - Comprehensive MSA integration demo
+  - `11_msa_search_large_protein.py` - Large protein optimization
+  - `12_msa_affinity_prediction.py` - MSA-guided affinity prediction
+- **Documentation**:
+  - New MSA Search Guide with complete usage examples
+  - Updated Affinity Prediction Guide with MSA-guided section
+  - Enhanced README with MSA integration examples
+
+### Fixed
+- Parameter naming consistency: `max_hits` → `max_msa_sequences`, `e_value_threshold` → `e_value`
+- MSA file handling in `predict_protein_structure()` method
+- Documentation inconsistencies and outdated examples
+
+### Improved
+- Better error handling for MSA Search API responses
+- Retry logic for MSA Search requests
+- More comprehensive test coverage
+
+## [0.2.1] - 2025-08-14
+
+### Added
+- Complete multi-endpoint support for ALL Boltz2 NIM functionalities:
+  - `predict_protein_structure()` with MSA support
+  - `predict_protein_ligand_complex()` with affinity prediction
+  - `predict_covalent_complex()`
+  - `predict_dna_protein_complex()`
+  - `predict_with_advanced_parameters()`
+  - `predict_from_yaml_config()` and `predict_from_yaml_file()`
+- New `msa` parameter in `predict_protein_structure()` for direct MSA input
+- Comprehensive demo notebook (`examples/boltz2_comprehensive_demo.ipynb`) with:
+  - Single and multi-endpoint examples
+  - Both Python API and CLI demonstrations
+  - Advanced features and visualization
+- `AlignmentFormat` export in `__init__.py`
+
+### Fixed
+- MSA support now correctly uses nested dictionary format `{database: {format: AlignmentFileRecord}}`
+- CLI multi-endpoint initialization (removed incorrect `is_async` parameter)
+- Multi-endpoint client MSA parameter passing
+- YAML configuration MSA format handling
+
+### Changed
+- Updated README.md examples (removed `is_async=True` from MultiEndpointClient)
+- Improved error handling in multi-endpoint operations
+- Better organization of test scripts into `test_scripts/` directory
+
+### Documentation
+- New comprehensive demo notebook with complete examples
+- Updated all guides for consistency
+- Added troubleshooting tips for multi-endpoint usage
+
+## [0.2.0] - Previous release
+- Initial multi-endpoint support for virtual screening
+- Basic affinity prediction features
+- Virtual screening capabilities

{boltz2_python_client-0.5.0/boltz2_python_client.egg-info → boltz2_python_client-0.5.2.post1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: boltz2-python-client
-Version: 0.5.0
+Version: 0.5.2.post1
 Summary: Python client for Boltz-2 protein structure prediction API with covalent complex and multi-endpoint support
 Author-email: NVIDIA Corporation <bionemo-support@nvidia.com>
 Maintainer-email: NVIDIA Corporation <bionemo-support@nvidia.com>
@@ -52,7 +52,7 @@ Copyright (c) 2025-2026, NVIDIA CORPORATION. All rights reserved.
 [![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 
-A comprehensive Python client for NVIDIA's Boltz-2 biomolecular structure prediction service. This package provides both synchronous and asynchronous interfaces, a rich CLI, and built-in 3D visualization capabilities.
+A comprehensive Python client for NVIDIA's Boltz-2 biomolecular structure prediction service. This package provides both synchronous and asynchronous interfaces, a rich CLI, multi-endpoint load balancing, and notebook examples that visualize predicted structures with `py3Dmol` and Molstar.
 
 ## Features
 
@@ -167,18 +167,43 @@ client = Boltz2Client(
 ### AWS SageMaker Endpoint
 
 ```python
-client = Boltz2Client(
+from boltz2_client import Boltz2SyncClient
+
+# Requires: pip install "boltz2-python-client[sagemaker]"
+# AWS credentials must be configured (e.g. via AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY
+# environment variables, ~/.aws/credentials, or an IAM role)
+
+client = Boltz2SyncClient(
     endpoint_type="sagemaker",
     sagemaker_endpoint_name="my-boltz2-endpoint",
     sagemaker_region="us-east-1",
 )
-```
 
-Requires `pip install "boltz2-python-client[sagemaker]"` and AWS credentials.
+# Health check (calls describe_endpoint under the hood)
+health = client.health_check()
+print(f"Endpoint status: {health.status}")
+
+# Predict protein structure — same API as local/hosted endpoints
+result = client.predict_protein_structure(
+    sequence="MKTVRQERLKSIVRILERSKEPVSGAQLAEELSVSRQVIVQDIAYLRSLGYNIVATPRGYVLAGG",
+    recycling_steps=3,
+    sampling_steps=200,
+)
+print(f"Confidence: {result.confidence_scores[0]:.3f}")
+
+# Protein-ligand with affinity prediction
+result = client.predict_protein_ligand_complex(
+    protein_sequence="MKTVRQERLK...",
+    ligand_smiles="CC(=O)OC1=CC=CC=C1C(=O)O",
+    predict_affinity=True,
+)
+```
 
 ```bash
-# CLI
+# CLI — all commands work with SageMaker by adding the endpoint flags
 boltz2 --endpoint-type sagemaker --sagemaker-endpoint-name my-boltz2-endpoint health
+boltz2 --endpoint-type sagemaker --sagemaker-endpoint-name my-boltz2-endpoint \
+    protein "MKTVRQERLKSIVRILERSKEPVSGAQLAEELSVSRQVIVQDIAYLRSLGYNIVATPRGYVLAGG"
 ```
 
 ## Supported Prediction Types

{boltz2_python_client-0.5.0 → boltz2_python_client-0.5.2.post1}/README.md

@@ -6,7 +6,7 @@ Copyright (c) 2025-2026, NVIDIA CORPORATION. All rights reserved.
 [![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 
-A comprehensive Python client for NVIDIA's Boltz-2 biomolecular structure prediction service. This package provides both synchronous and asynchronous interfaces, a rich CLI, and built-in 3D visualization capabilities.
+A comprehensive Python client for NVIDIA's Boltz-2 biomolecular structure prediction service. This package provides both synchronous and asynchronous interfaces, a rich CLI, multi-endpoint load balancing, and notebook examples that visualize predicted structures with `py3Dmol` and Molstar.
 
 ## Features
 
@@ -121,18 +121,43 @@ client = Boltz2Client(
 ### AWS SageMaker Endpoint
 
 ```python
-client = Boltz2Client(
+from boltz2_client import Boltz2SyncClient
+
+# Requires: pip install "boltz2-python-client[sagemaker]"
+# AWS credentials must be configured (e.g. via AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY
+# environment variables, ~/.aws/credentials, or an IAM role)
+
+client = Boltz2SyncClient(
     endpoint_type="sagemaker",
     sagemaker_endpoint_name="my-boltz2-endpoint",
     sagemaker_region="us-east-1",
 )
-```
 
-Requires `pip install "boltz2-python-client[sagemaker]"` and AWS credentials.
+# Health check (calls describe_endpoint under the hood)
+health = client.health_check()
+print(f"Endpoint status: {health.status}")
+
+# Predict protein structure — same API as local/hosted endpoints
+result = client.predict_protein_structure(
+    sequence="MKTVRQERLKSIVRILERSKEPVSGAQLAEELSVSRQVIVQDIAYLRSLGYNIVATPRGYVLAGG",
+    recycling_steps=3,
+    sampling_steps=200,
+)
+print(f"Confidence: {result.confidence_scores[0]:.3f}")
+
+# Protein-ligand with affinity prediction
+result = client.predict_protein_ligand_complex(
+    protein_sequence="MKTVRQERLK...",
+    ligand_smiles="CC(=O)OC1=CC=CC=C1C(=O)O",
+    predict_affinity=True,
+)
+```
 
 ```bash
-# CLI
+# CLI — all commands work with SageMaker by adding the endpoint flags
 boltz2 --endpoint-type sagemaker --sagemaker-endpoint-name my-boltz2-endpoint health
+boltz2 --endpoint-type sagemaker --sagemaker-endpoint-name my-boltz2-endpoint \
+    protein "MKTVRQERLKSIVRILERSKEPVSGAQLAEELSVSRQVIVQDIAYLRSLGYNIVATPRGYVLAGG"
 ```
 
 ## Supported Prediction Types

{boltz2_python_client-0.5.0 → boltz2_python_client-0.5.2.post1}/boltz2_client/__init__.py

@@ -26,7 +26,7 @@ Example:
     >>> print(f"Confidence: {result.confidence_scores[0]:.3f}")
 """
 
-__version__ = "0.5.0"
+__version__ = "0.5.2.post1"
 __author__ = "NVIDIA Corporation"
 __email__ = "bionemo-support@nvidia.com"
 
@@ -103,28 +103,6 @@ from .utils import (
     convert_pdb_to_cif,
 )
 
-# Optional imports for visualization
-try:
-    from .visualization import (
-        StructureVisualizer,
-        visualize_structure,
-        create_multi_view,
-    )
-    _HAS_VISUALIZATION = True
-except ImportError:
-    _HAS_VISUALIZATION = False
-
-# Optional imports for analysis
-try:
-    from .analysis import (
-        StructureAnalyzer,
-        calculate_rmsd,
-        analyze_contacts,
-    )
-    _HAS_ANALYSIS = True
-except ImportError:
-    _HAS_ANALYSIS = False
-
 __all__ = [
     # Core client classes
     "Boltz2Client",
@@ -201,21 +179,6 @@ __all__ = [
     "SPECIES_TO_TAXID",
 ]
 
-# Add visualization exports if available
-if _HAS_VISUALIZATION:
-    __all__.extend([
-        "StructureVisualizer",
-        "visualize_structure", 
-        "create_multi_view",
-    ])
-
-# Add analysis exports if available
-if _HAS_ANALYSIS:
-    __all__.extend([
-        "StructureAnalyzer",
-        "calculate_rmsd",
-        "analyze_contacts",
-    ])
 
 def get_version() -> str:
     """Get the current version of the package."""

{boltz2_python_client-0.5.0 → boltz2_python_client-0.5.2.post1}/boltz2_client/cli/__init__.py

@@ -22,6 +22,7 @@ from rich.table import Table
 from rich.progress import Progress, SpinnerColumn, TextColumn, TimeElapsedColumn, BarColumn
 import yaml as pyyaml
 
+from .. import __version__ as _pkg_version
 from ..client import Boltz2Client
 from ..models import (
     PredictionRequest, Polymer, Ligand, BondConstraint,
@@ -53,6 +54,8 @@ def print_warning(message: str):
 
 
 @click.group()
+@click.version_option(_pkg_version, "-V", "--version", prog_name="boltz2",
+                      message="%(prog)s %(version)s")
 @click.option('--base-url', default='http://localhost:8000', help='Service base URL (can be comma-separated for multiple endpoints)')
 @click.option('--api-key', help='API key for NVIDIA hosted endpoints (or set NVIDIA_API_KEY env var)')
 @click.option('--endpoint-type',

{boltz2_python_client-0.5.0 → boltz2_python_client-0.5.2.post1}/boltz2_client/cli/info.py

@@ -78,6 +78,11 @@ def metadata(ctx):
     """Get service metadata and model information."""
     async def get_metadata():
         try:
+            if ctx.obj.get('endpoint_type') == 'sagemaker':
+                print_error("Metadata endpoint is not available for SageMaker endpoints.")
+                print_error("Use 'boltz2 ... health' to check SageMaker endpoint status.")
+                raise click.Abort()
+
             client = create_client(ctx)
             
             with Progress(
@@ -108,6 +113,8 @@ def metadata(ctx):
                 
                 console.print(table)
                 
+        except click.Abort:
+            raise
         except Exception as e:
             print_error(f"Failed to get metadata: {e}")
             raise click.Abort()

{boltz2_python_client-0.5.0 → boltz2_python_client-0.5.2.post1}/boltz2_client/cli/msa.py

@@ -115,8 +115,12 @@ def msa_search_command(ctx, sequence: str, endpoint: str, databases: List[str],
             
         except Exception as e:
             print_error(f"MSA search failed: {e}")
+            raise
     
-    asyncio.run(run_msa_search())
+    try:
+        asyncio.run(run_msa_search())
+    except Exception:
+        raise click.Abort()
 
 
 @cli.command(name='msa-predict')
@@ -233,8 +237,12 @@ def msa_predict_command(ctx, sequence: str, endpoint: str, databases: List[str],
             
         except Exception as e:
             print_error(f"MSA prediction failed: {e}")
+            raise
     
-    asyncio.run(run_msa_predict())
+    try:
+        asyncio.run(run_msa_predict())
+    except Exception:
+        raise click.Abort()
 
 
 @cli.command(name='msa-ligand')
@@ -380,8 +388,12 @@ def msa_ligand_command(ctx, protein_sequence: str, smiles: Optional[str], ccd: O
             
         except Exception as e:
             print_error(f"MSA-ligand prediction failed: {e}")
+            raise
     
-    asyncio.run(run_msa_ligand())
+    try:
+        asyncio.run(run_msa_ligand())
+    except Exception:
+        raise click.Abort()
 
 
 @cli.command(name='convert-msa')
@@ -779,8 +791,6 @@ def multimer_msa_command(ctx, a3m_files: Tuple[str, ...], chain_ids: str,
             
             # Save all outputs if requested
             if save_all:
-                import json
-                
                 # Collect all scores and metrics
                 scores = {
                     'confidence_scores': response.confidence_scores,

{boltz2_python_client-0.5.0 → boltz2_python_client-0.5.2.post1}/boltz2_client/cli/screen.py

@@ -44,11 +44,15 @@ def screen(ctx, target_sequence, compounds_file, target_name, output_dir, no_aff
         boltz2 screen "MKTVRQERLK..." compounds.csv -o results/
         boltz2 screen target.fasta library.json --pocket-residues "10,15,20,25"
     """
-    client = create_client(ctx)
-    
     # Import here to avoid circular imports
     from ..virtual_screening import VirtualScreening, CompoundLibrary
     
+    try:
+        client = create_client(ctx)
+    except Exception as e:
+        print_error(f"Failed to initialize client: {e}")
+        raise click.Abort()
+    
     console.print(f"\n[bold cyan]🧬 Virtual Screening Campaign[/bold cyan]")
     console.print(f"Target: {target_name}")
     console.print(f"Compounds: {compounds_file}")
@@ -64,7 +68,11 @@ def screen(ctx, target_sequence, compounds_file, target_name, output_dir, no_aff
     # Parse pocket residues
     pocket_residues_list = None
     if pocket_residues:
-        pocket_residues_list = [int(x.strip()) for x in pocket_residues.split(',')]
+        try:
+            pocket_residues_list = [int(x.strip()) for x in pocket_residues.split(',')]
+        except ValueError as e:
+            print_error(f"Invalid --pocket-residues value (expected comma-separated ints): {e}")
+            raise click.Abort()
         console.print(f"Pocket constraint: {len(pocket_residues_list)} residues")
     
     # Create screener

{boltz2_python_client-0.5.0 → boltz2_python_client-0.5.2.post1}/boltz2_client/client.py

@@ -243,8 +243,6 @@ class Boltz2Client:
         self, request_dict: dict, progress_callback: Optional[Callable] = None
     ) -> dict:
         """Invoke a SageMaker endpoint and return the parsed JSON response."""
-        import asyncio
-
         def _invoke():
             return self._sm_runtime.invoke_endpoint(
                 EndpointName=self._sm_endpoint_name,
@@ -256,7 +254,10 @@ class Boltz2Client:
         if progress_callback:
             progress_callback(f"Invoking SageMaker endpoint {self._sm_endpoint_name}...")
 
-        loop = asyncio.get_event_loop()
+        # Always run the blocking boto3 call in a worker thread. Use the
+        # currently running loop (asyncio.get_event_loop is deprecated for
+        # this purpose in Python 3.10+ and emits a DeprecationWarning).
+        loop = asyncio.get_running_loop()
         response = await loop.run_in_executor(None, _invoke)
 
         http_status = response["ResponseMetadata"]["HTTPStatusCode"]