workbench 0.8.175__tar.gz → 0.8.177__tar.gz

{workbench-0.8.175 → workbench-0.8.177}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: workbench
-Version: 0.8.175
+Version: 0.8.177
 Summary: Workbench: A Dashboard and Python API for creating and deploying AWS SageMaker Model Pipelines
 Author-email: SuperCowPowers LLC <support@supercowpowers.com>
 License-Expression: MIT

{workbench-0.8.175 → workbench-0.8.177}/src/workbench/core/artifacts/artifact.py

@@ -238,8 +238,8 @@ class Artifact(ABC):
         """
 
         # Check for ReadOnly Role
-        if self.aws_account_clamp.read_only_role:
-            self.log.info("Cannot add metadata with a ReadOnly Role...")
+        if self.aws_account_clamp.read_only:
+            self.log.info("Cannot add metadata with a ReadOnly Permissions...")
             return
 
         # Sanity check

{workbench-0.8.175 → workbench-0.8.177}/src/workbench/core/artifacts/endpoint_core.py

@@ -32,11 +32,11 @@ from sagemaker import Predictor
 from workbench.core.artifacts.artifact import Artifact
 from workbench.core.artifacts import FeatureSetCore, ModelCore, ModelType
 from workbench.utils.endpoint_metrics import EndpointMetrics
-from workbench.utils.fast_inference import fast_inference
 from workbench.utils.cache import Cache
 from workbench.utils.s3_utils import compute_s3_object_hash
 from workbench.utils.model_utils import uq_metrics
 from workbench.utils.xgboost_model_utils import cross_fold_inference
+from workbench_bridges.endpoints.fast_inference import fast_inference
 
 
 class EndpointCore(Artifact):
@@ -1061,6 +1061,9 @@ if __name__ == "__main__":
     assert len(pred_results) == len(my_eval_df), "Predictions should match the number of sent rows"
 
     # Now we put in an invalid value
+    print("*" * 80)
+    print("NOW TESTING ERROR CONDITIONS...")
+    print("*" * 80)
     my_eval_df.at[42, "length"] = "invalid_value"
     pred_results = my_endpoint.inference(my_eval_df, drop_error_rows=True)
     print(f"Sent rows: {len(my_eval_df)}")

{workbench-0.8.175 → workbench-0.8.177}/src/workbench/core/artifacts/model_core.py

@@ -37,35 +37,6 @@ class ModelType(Enum):
     UNKNOWN = "unknown"
 
 
-# Deprecated Images
-"""
-        # US East 1 images
-        "py312-general-ml-training"
-        ("us-east-1", "training", "0.1", "x86_64"): (
-            "507740646243.dkr.ecr.us-east-1.amazonaws.com/aws-ml-images/py312-sklearn-xgb-training:0.1"
-        ),
-        ("us-east-1", "inference", "0.1", "x86_64"): (
-            "507740646243.dkr.ecr.us-east-1.amazonaws.com/aws-ml-images/py312-sklearn-xgb-inference:0.1"
-        ),
-
-        # US West 2 images
-        ("us-west-2", "training", "0.1", "x86_64"): (
-            "507740646243.dkr.ecr.us-west-2.amazonaws.com/aws-ml-images/py312-sklearn-xgb-training:0.1"
-        ),
-        ("us-west-2", "inference", "0.1", "x86_64"): (
-            "507740646243.dkr.ecr.us-west-2.amazonaws.com/aws-ml-images/py312-sklearn-xgb-inference:0.1"
-        ),
-
-        # ARM64 images
-        ("us-east-1", "inference", "0.1", "arm64"): (
-            "507740646243.dkr.ecr.us-east-1.amazonaws.com/aws-ml-images/py312-sklearn-xgb-inference:0.1-arm64"
-        ),
-        ("us-west-2", "inference", "0.1", "arm64"): (
-            "507740646243.dkr.ecr.us-west-2.amazonaws.com/aws-ml-images/py312-sklearn-xgb-inference:0.1-arm64"
-        ),
-"""
-
-
 class ModelImages:
     """Class for retrieving workbench inference images"""
 
@@ -890,6 +861,14 @@ class ModelCore(Artifact):
                 shap_data[key] = self.df_store.get(df_location)
             return shap_data or None
 
+    def cross_folds(self) -> dict:
+        """Retrieve the cross-fold inference results(only works for XGBoost models)
+
+        Returns:
+            dict: Dictionary with the cross-fold inference results
+        """
+        return self.param_store.get(f"/workbench/models/{self.name}/inference/cross_fold")
+
     def supported_inference_instances(self) -> Optional[list]:
         """Retrieve the supported endpoint inference instance types
 

{workbench-0.8.175 → workbench-0.8.177}/src/workbench/core/cloud_platform/aws/aws_account_clamp.py

@@ -55,9 +55,10 @@ class AWSAccountClamp:
         # Check our Assume Role
         self.log.info("Checking Workbench Assumed Role...")
         role_info = self.aws_session.assumed_role_info()
+        self.log.info(f"Assumed Role: {role_info}")
 
-        # Check if the Role is a 'ReadOnly' role
-        self.read_only_role = "readonly" in role_info["AssumedRoleArn"].lower()
+        # Check if we have tag write permissions (if we don't, we are read-only)
+        self.read_only = not self.check_tag_permissions()
 
         # Check our Workbench API Key and Load the License
         self.log.info("Checking Workbench API License...")
@@ -141,6 +142,45 @@ class AWSAccountClamp:
         """
         return self.boto3_session.client("sagemaker")
 
+    def check_tag_permissions(self):
+        """Check if current role has permission to add tags to SageMaker endpoints.
+
+        Returns:
+            bool: True if AddTags is allowed, False otherwise
+        """
+        try:
+            sagemaker = self.boto3_session.client("sagemaker")
+
+            # Use a non-existent endpoint name
+            fake_endpoint = "workbench-permission-check-dummy-endpoint"
+
+            # Try to add tags to the non-existent endpoint
+            sagemaker.add_tags(
+                ResourceArn=f"arn:aws:sagemaker:{self.region}:{self.account_id}:endpoint/{fake_endpoint}",
+                Tags=[{"Key": "PermissionCheck", "Value": "Test"}],
+            )
+
+            # If we get here, we have permission (but endpoint doesn't exist)
+            return True
+
+        except ClientError as e:
+            error_code = e.response["Error"]["Code"]
+
+            # AccessDeniedException = no permission
+            if error_code == "AccessDeniedException":
+                self.log.debug("No AddTags permission (AccessDeniedException)")
+                return False
+
+            # ResourceNotFound = we have permission, but endpoint doesn't exist
+            elif error_code in ["ResourceNotFound", "ValidationException"]:
+                self.log.debug("AddTags permission verified (resource not found)")
+                return True
+
+            # Unexpected error, assume no permission for safety
+            else:
+                self.log.debug(f"Unexpected error checking permissions: {error_code}")
+                return False
+
 
 if __name__ == "__main__":
     """Exercise the AWS Account Clamp Class"""
@@ -165,3 +205,9 @@ if __name__ == "__main__":
     print("\n\n*** AWS Sagemaker Session/Client Check ***")
     sm_client = aws_account_clamp.sagemaker_client()
     print(sm_client.list_feature_groups()["FeatureGroupSummaries"])
+
+    print("\n\n*** AWS Tag Permission Check ***")
+    if aws_account_clamp.check_tag_permissions():
+        print("Tag Permission Check Success...")
+    else:
+        print("Tag Permission Check Failed...")

{workbench-0.8.175 → workbench-0.8.177}/src/workbench/model_scripts/custom_models/chem_info/mol_descriptors.py

@@ -91,16 +91,27 @@ import logging
 import pandas as pd
 import numpy as np
 import re
+import time
+from contextlib import contextmanager
 from rdkit import Chem
 from rdkit.Chem import Descriptors, rdCIPLabeler
 from rdkit.ML.Descriptors import MoleculeDescriptors
 from mordred import Calculator as MordredCalculator
 from mordred import AcidBase, Aromatic, Constitutional, Chi, CarbonTypes
 
+
 logger = logging.getLogger("workbench")
 logger.setLevel(logging.DEBUG)
 
 
+# Helper context manager for timing
+@contextmanager
+def timer(name):
+    start = time.time()
+    yield
+    print(f"{name}: {time.time() - start:.2f}s")
+
+
 def compute_stereochemistry_features(mol):
     """
     Compute stereochemistry descriptors using modern RDKit methods.
@@ -280,9 +291,11 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
                 descriptor_values.append([np.nan] * len(all_descriptors))
 
     # Create RDKit features DataFrame
-    rdkit_features_df = pd.DataFrame(descriptor_values, columns=calc.GetDescriptorNames(), index=result.index)
+    rdkit_features_df = pd.DataFrame(descriptor_values, columns=calc.GetDescriptorNames())
 
     # Add RDKit features to result
+    # Remove any columns from result that exist in rdkit_features_df
+    result = result.drop(columns=result.columns.intersection(rdkit_features_df.columns))
     result = pd.concat([result, rdkit_features_df], axis=1)
 
     # Compute Mordred descriptors
@@ -299,7 +312,7 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
 
         # Compute Mordred descriptors
         valid_mols = [mol if mol is not None else Chem.MolFromSmiles("C") for mol in molecules]
-        mordred_df = calc.pandas(valid_mols, nproc=1)  # For serverless, use nproc=1
+        mordred_df = calc.pandas(valid_mols, nproc=1)  # Endpoint multiprocessing will fail with nproc>1
 
         # Replace values for invalid molecules with NaN
         for i, mol in enumerate(molecules):
@@ -310,10 +323,9 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
         for col in mordred_df.columns:
             mordred_df[col] = pd.to_numeric(mordred_df[col], errors="coerce")
 
-        # Set index to match result DataFrame
-        mordred_df.index = result.index
-
         # Add Mordred features to result
+        # Remove any columns from result that exist in mordred
+        result = result.drop(columns=result.columns.intersection(mordred_df.columns))
         result = pd.concat([result, mordred_df], axis=1)
 
     # Compute stereochemistry features if requested
@@ -326,9 +338,10 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
             stereo_features.append(stereo_dict)
 
         # Create stereochemistry DataFrame
-        stereo_df = pd.DataFrame(stereo_features, index=result.index)
+        stereo_df = pd.DataFrame(stereo_features)
 
         # Add stereochemistry features to result
+        result = result.drop(columns=result.columns.intersection(stereo_df.columns))
         result = pd.concat([result, stereo_df], axis=1)
 
         logger.info(f"Added {len(stereo_df.columns)} stereochemistry descriptors")
@@ -357,7 +370,6 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
 
 
 if __name__ == "__main__":
-    import time
     from mol_standardize import standardize
     from workbench.api import DataSource
 

{workbench-0.8.175/src/workbench/utils/chem_utils → workbench-0.8.177/src/workbench/model_scripts/custom_models/chem_info}/mol_standardize.py

@@ -81,6 +81,8 @@ Usage:
 import logging
 from typing import Optional, Tuple
 import pandas as pd
+import time
+from contextlib import contextmanager
 from rdkit import Chem
 from rdkit.Chem import Mol
 from rdkit.Chem.MolStandardize import rdMolStandardize
@@ -90,6 +92,14 @@ log = logging.getLogger("workbench")
 RDLogger.DisableLog("rdApp.warning")
 
 
+# Helper context manager for timing
+@contextmanager
+def timer(name):
+    start = time.time()
+    yield
+    print(f"{name}: {time.time() - start:.2f}s")
+
+
 class MolStandardizer:
     """
     Streamlined molecular standardizer for ADMET preprocessing
@@ -116,6 +126,7 @@ class MolStandardizer:
         Pipeline:
         1. Cleanup (remove Hs, disconnect metals, normalize)
         2. Get largest fragment (optional - only if remove_salts=True)
+           2a. Extract salt information BEFORE further modifications
         3. Neutralize charges
         4. Canonicalize tautomer (optional)
 
@@ -130,18 +141,24 @@ class MolStandardizer:
 
         try:
             # Step 1: Cleanup
-            mol = rdMolStandardize.Cleanup(mol, self.params)
-            if mol is None:
+            cleaned_mol = rdMolStandardize.Cleanup(mol, self.params)
+            if cleaned_mol is None:
                 return None, None
 
+            # If not doing any transformations, return early
+            if not self.remove_salts and not self.canonicalize_tautomer:
+                return cleaned_mol, None
+
             salt_smiles = None
+            mol = cleaned_mol
 
             # Step 2: Fragment handling (conditional based on remove_salts)
             if self.remove_salts:
-                # Get parent molecule and extract salt information
-                parent_mol = rdMolStandardize.FragmentParent(mol, self.params)
+                # Get parent molecule
+                parent_mol = rdMolStandardize.FragmentParent(cleaned_mol, self.params)
                 if parent_mol:
-                    salt_smiles = self._extract_salt(mol, parent_mol)
+                    # Extract salt BEFORE any modifications to parent
+                    salt_smiles = self._extract_salt(cleaned_mol, parent_mol)
                     mol = parent_mol
                 else:
                     return None, None
@@ -153,7 +170,7 @@ class MolStandardizer:
                 if mol is None:
                     return None, salt_smiles
 
-            # Step 4: Canonicalize tautomer
+            # Step 4: Canonicalize tautomer (LAST STEP)
             if self.canonicalize_tautomer:
                 mol = self.tautomer_enumerator.Canonicalize(mol)
 
@@ -172,13 +189,22 @@ class MolStandardizer:
         - Mixtures: multiple large neutral organic fragments
 
         Args:
-            orig_mol: Original molecule (before FragmentParent)
-            parent_mol: Parent molecule (after FragmentParent)
+            orig_mol: Original molecule (after Cleanup, before FragmentParent)
+            parent_mol: Parent molecule (after FragmentParent, before tautomerization)
 
         Returns:
             SMILES string of salt components or None if no salts/mixture detected
         """
         try:
+            # Quick atom count check
+            if orig_mol.GetNumAtoms() == parent_mol.GetNumAtoms():
+                return None
+
+            # Quick heavy atom difference check
+            heavy_diff = orig_mol.GetNumHeavyAtoms() - parent_mol.GetNumHeavyAtoms()
+            if heavy_diff <= 0:
+                return None
+
             # Get all fragments from original molecule
             orig_frags = Chem.GetMolFrags(orig_mol, asMols=True)
 
@@ -268,7 +294,7 @@ def standardize(
     if "orig_smiles" not in result.columns:
         result["orig_smiles"] = result[smiles_column]
 
-    # Initialize standardizer with salt removal control
+    # Initialize standardizer
     standardizer = MolStandardizer(canonicalize_tautomer=canonicalize_tautomer, remove_salts=extract_salts)
 
     def process_smiles(smiles: str) -> pd.Series:
@@ -286,6 +312,11 @@ def standardize(
             log.error("Encountered missing or empty SMILES string")
             return pd.Series({"smiles": None, "salt": None})
 
+        # Early check for unreasonably long SMILES
+        if len(smiles) > 1000:
+            log.error(f"SMILES too long ({len(smiles)} chars): {smiles[:50]}...")
+            return pd.Series({"smiles": None, "salt": None})
+
         # Parse molecule
         mol = Chem.MolFromSmiles(smiles)
         if mol is None:
@@ -299,7 +330,9 @@ def standardize(
         if std_mol is not None:
             # Check if molecule is reasonable
             if std_mol.GetNumAtoms() == 0 or std_mol.GetNumAtoms() > 200:  # Arbitrary limits
-                log.error(f"Unusual molecule size: {std_mol.GetNumAtoms()} atoms")
+                log.error(f"Rejecting molecule size: {std_mol.GetNumAtoms()} atoms")
+                log.error(f"Original SMILES: {smiles}")
+                return pd.Series({"smiles": None, "salt": salt_smiles})
 
         if std_mol is None:
             return pd.Series(
@@ -325,8 +358,11 @@ def standardize(
 
 
 if __name__ == "__main__":
-    import time
-    from workbench.api import DataSource
+
+    # Pandas display options for better readability
+    pd.set_option("display.max_columns", None)
+    pd.set_option("display.width", 1000)
+    pd.set_option("display.max_colwidth", 100)
 
     # Test with DataFrame including various salt forms
     test_data = pd.DataFrame(
@@ -362,67 +398,53 @@ if __name__ == "__main__":
     )
 
     # General test
+    print("Testing standardization with full dataset...")
     standardize(test_data)
 
     # Remove the last two rows to avoid errors with None and INVALID
     test_data = test_data.iloc[:-2].reset_index(drop=True)
 
     # Test WITHOUT salt removal (keeps full molecule)
-    print("\nStandardization KEEPING salts (extract_salts=False):")
-    print("This preserves the full molecule including counterions")
+    print("\nStandardization KEEPING salts (extract_salts=False) Tautomerization: True")
     result_keep = standardize(test_data, extract_salts=False, canonicalize_tautomer=True)
-    display_cols = ["compound_id", "orig_smiles", "smiles", "salt"]
-    print(result_keep[display_cols].to_string())
+    display_order = ["compound_id", "orig_smiles", "smiles", "salt"]
+    print(result_keep[display_order])
 
     # Test WITH salt removal
     print("\n" + "=" * 70)
     print("Standardization REMOVING salts (extract_salts=True):")
-    print("This extracts parent molecule and records salt information")
     result_remove = standardize(test_data, extract_salts=True, canonicalize_tautomer=True)
-    print(result_remove[display_cols].to_string())
+    print(result_remove[display_order])
 
-    # Test WITHOUT tautomerization (keeping salts)
+    # Test with problematic cases specifically
     print("\n" + "=" * 70)
-    print("Standardization KEEPING salts, NO tautomerization:")
-    result_no_taut = standardize(test_data, extract_salts=False, canonicalize_tautomer=False)
-    print(result_no_taut[display_cols].to_string())
+    print("Testing specific problematic cases:")
+    problem_cases = pd.DataFrame(
+        {
+            "smiles": [
+                "CC(=O)O.CCN",  # Should extract CC(=O)O as salt
+                "CCO.CC",  # Should return CC as salt
+            ],
+            "compound_id": ["TEST_C002", "TEST_C005"],
+        }
+    )
+
+    problem_result = standardize(problem_cases, extract_salts=True, canonicalize_tautomer=True)
+    print(problem_result[display_order])
+
+    # Performance test with larger dataset
+    from workbench.api import DataSource
 
-    # Show the difference for salt-containing molecules
-    print("\n" + "=" * 70)
-    print("Comparison showing differences:")
-    for idx, row in result_keep.iterrows():
-        keep_smiles = row["smiles"]
-        remove_smiles = result_remove.loc[idx, "smiles"]
-        no_taut_smiles = result_no_taut.loc[idx, "smiles"]
-        salt = result_remove.loc[idx, "salt"]
-
-        # Show differences when they exist
-        if keep_smiles != remove_smiles or keep_smiles != no_taut_smiles:
-            print(f"\n{row['compound_id']} ({row['orig_smiles']}):")
-            if keep_smiles != no_taut_smiles:
-                print(f"  With salt + taut:    {keep_smiles}")
-                print(f"  With salt, no taut:  {no_taut_smiles}")
-            if keep_smiles != remove_smiles:
-                print(f"  Parent only + taut:  {remove_smiles}")
-            if salt:
-                print(f"  Extracted salt:      {salt}")
-
-    # Summary statistics
     print("\n" + "=" * 70)
-    print("Summary:")
-    print(f"Total molecules: {len(result_remove)}")
-    print(f"Molecules with salts: {result_remove['salt'].notna().sum()}")
-    unique_salts = result_remove["salt"].dropna().unique()
-    print(f"Unique salts found: {unique_salts[:5].tolist()}")
 
-    # Get a real dataset from Workbench and time the standardization
     ds = DataSource("aqsol_data")
-    df = ds.pull_dataframe()[["id", "smiles"]]
-    start_time = time.time()
-    std_df = standardize(df, extract_salts=True, canonicalize_tautomer=True)
-    end_time = time.time()
-    print(f"\nStandardized {len(std_df)} molecules from Workbench in {end_time - start_time:.2f} seconds")
-    print(std_df.head())
-    print(f"Molecules with salts: {std_df['salt'].notna().sum()}")
-    unique_salts = std_df["salt"].dropna().unique()
-    print(f"Unique salts found: {unique_salts[:5].tolist()}")
+    df = ds.pull_dataframe()[["id", "smiles"]][:1000]
+
+    for tautomer in [True, False]:
+        for extract in [True, False]:
+            print(f"Performance test with AQSol dataset: tautomer={tautomer} extract_salts={extract}:")
+            start_time = time.time()
+            std_df = standardize(df, canonicalize_tautomer=tautomer, extract_salts=extract)
+            elapsed = time.time() - start_time
+            mol_per_sec = len(df) / elapsed
+            print(f"{elapsed:.2f}s ({mol_per_sec:.0f} mol/s)")

{workbench-0.8.175 → workbench-0.8.177}/src/workbench/model_scripts/custom_models/uq_models/mapie.template

@@ -472,9 +472,9 @@ def predict_fn(df, models) -> pd.DataFrame:
     # Add median (q_50) from XGBoost prediction
     df["q_50"] = df["prediction"]
 
-    # Calculate uncertainty metrics based on 95% interval
-    interval_width = df["q_975"] - df["q_025"]
-    df["prediction_std"] = interval_width / 3.92
+    # Calculate uncertainty metrics based on 50% interval
+    interval_width = df["q_75"] - df["q_25"]
+    df["prediction_std"] = interval_width / 1.348
 
     # Reorder the quantile columns for easier reading
     quantile_cols = ["q_025", "q_05", "q_10", "q_25", "q_75", "q_90", "q_95", "q_975"]

{workbench-0.8.175 → workbench-0.8.177}/src/workbench/model_scripts/xgb_model/generated_model_script.py

@@ -28,11 +28,11 @@ from typing import List, Tuple
 
 # Template Parameters
 TEMPLATE_PARAMS = {
-    "model_type": "classifier",
-    "target": "class",
+    "model_type": "regressor",
+    "target": "udm_asy_res_value",
     "features": ['chi2v', 'fr_sulfone', 'chi1v', 'bcut2d_logplow', 'fr_piperzine', 'kappa3', 'smr_vsa1', 'slogp_vsa5', 'fr_ketone_topliss', 'fr_sulfonamd', 'fr_imine', 'fr_benzene', 'fr_ester', 'chi2n', 'labuteasa', 'peoe_vsa2', 'smr_vsa6', 'bcut2d_chglo', 'fr_sh', 'peoe_vsa1', 'fr_allylic_oxid', 'chi4n', 'fr_ar_oh', 'fr_nh0', 'fr_term_acetylene', 'slogp_vsa7', 'slogp_vsa4', 'estate_vsa1', 'vsa_estate4', 'numbridgeheadatoms', 'numheterocycles', 'fr_ketone', 'fr_morpholine', 'fr_guanido', 'estate_vsa2', 'numheteroatoms', 'fr_nitro_arom_nonortho', 'fr_piperdine', 'nocount', 'numspiroatoms', 'fr_aniline', 'fr_thiophene', 'slogp_vsa10', 'fr_amide', 'slogp_vsa2', 'fr_epoxide', 'vsa_estate7', 'fr_ar_coo', 'fr_imidazole', 'fr_nitrile', 'fr_oxazole', 'numsaturatedrings', 'fr_pyridine', 'fr_hoccn', 'fr_ndealkylation1', 'numaliphaticheterocycles', 'fr_phenol', 'maxpartialcharge', 'vsa_estate5', 'peoe_vsa13', 'minpartialcharge', 'qed', 'fr_al_oh', 'slogp_vsa11', 'chi0n', 'fr_bicyclic', 'peoe_vsa12', 'fpdensitymorgan1', 'fr_oxime', 'molwt', 'fr_dihydropyridine', 'smr_vsa5', 'peoe_vsa5', 'fr_nitro', 'hallkieralpha', 'heavyatommolwt', 'fr_alkyl_halide', 'peoe_vsa8', 'fr_nhpyrrole', 'fr_isocyan', 'bcut2d_chghi', 'fr_lactam', 'peoe_vsa11', 'smr_vsa9', 'tpsa', 'chi4v', 'slogp_vsa1', 'phi', 'bcut2d_logphi', 'avgipc', 'estate_vsa11', 'fr_coo', 'bcut2d_mwhi', 'numunspecifiedatomstereocenters', 'vsa_estate10', 'estate_vsa8', 'numvalenceelectrons', 'fr_nh2', 'fr_lactone', 'vsa_estate1', 'estate_vsa4', 'numatomstereocenters', 'vsa_estate8', 'fr_para_hydroxylation', 'peoe_vsa3', 'fr_thiazole', 'peoe_vsa10', 'fr_ndealkylation2', 'slogp_vsa12', 'peoe_vsa9', 'maxestateindex', 'fr_quatn', 'smr_vsa7', 'minestateindex', 'numaromaticheterocycles', 'numrotatablebonds', 'fr_ar_nh', 'fr_ether', 'exactmolwt', 'fr_phenol_noorthohbond', 'slogp_vsa3', 'fr_ar_n', 'sps', 'fr_c_o_nocoo', 'bertzct', 'peoe_vsa7', 'slogp_vsa8', 'numradicalelectrons', 'molmr', 'fr_tetrazole', 'numsaturatedcarbocycles', 'bcut2d_mrhi', 'kappa1', 'numamidebonds', 'fpdensitymorgan2', 'smr_vsa8', 'chi1n', 'estate_vsa6', 'fr_barbitur', 'fr_diazo', 'kappa2', 'chi0', 'bcut2d_mrlow', 'balabanj', 'peoe_vsa4', 'numhacceptors', 'fr_sulfide', 'chi3n', 'smr_vsa2', 'fr_al_oh_notert', 'fr_benzodiazepine', 'fr_phos_ester', 'fr_aldehyde', 'fr_coo2', 'estate_vsa5', 'fr_prisulfonamd', 'numaromaticcarbocycles', 'fr_unbrch_alkane', 'fr_urea', 'fr_nitroso', 'smr_vsa10', 'fr_c_s', 'smr_vsa3', 'fr_methoxy', 'maxabspartialcharge', 'slogp_vsa9', 'heavyatomcount', 'fr_azide', 'chi3v', 'smr_vsa4', 'mollogp', 'chi0v', 'fr_aryl_methyl', 'fr_nh1', 'fpdensitymorgan3', 'fr_furan', 'fr_hdrzine', 'fr_arn', 'numaromaticrings', 'vsa_estate3', 'fr_azo', 'fr_halogen', 'estate_vsa9', 'fr_hdrzone', 'numhdonors', 'fr_alkyl_carbamate', 'fr_isothiocyan', 'minabspartialcharge', 'fr_al_coo', 'ringcount', 'chi1', 'estate_vsa7', 'fr_nitro_arom', 'vsa_estate9', 'minabsestateindex', 'maxabsestateindex', 'vsa_estate6', 'estate_vsa10', 'estate_vsa3', 'fr_n_o', 'fr_amidine', 'fr_thiocyan', 'fr_phos_acid', 'fr_c_o', 'fr_imide', 'numaliphaticrings', 'peoe_vsa6', 'vsa_estate2', 'nhohcount', 'numsaturatedheterocycles', 'slogp_vsa6', 'peoe_vsa14', 'fractioncsp3', 'bcut2d_mwlow', 'numaliphaticcarbocycles', 'fr_priamide', 'nacid', 'nbase', 'naromatom', 'narombond', 'sz', 'sm', 'sv', 'sse', 'spe', 'sare', 'sp', 'si', 'mz', 'mm', 'mv', 'mse', 'mpe', 'mare', 'mp', 'mi', 'xch_3d', 'xch_4d', 'xch_5d', 'xch_6d', 'xch_7d', 'xch_3dv', 'xch_4dv', 'xch_5dv', 'xch_6dv', 'xch_7dv', 'xc_3d', 'xc_4d', 'xc_5d', 'xc_6d', 'xc_3dv', 'xc_4dv', 'xc_5dv', 'xc_6dv', 'xpc_4d', 'xpc_5d', 'xpc_6d', 'xpc_4dv', 'xpc_5dv', 'xpc_6dv', 'xp_0d', 'xp_1d', 'xp_2d', 'xp_3d', 'xp_4d', 'xp_5d', 'xp_6d', 'xp_7d', 'axp_0d', 'axp_1d', 'axp_2d', 'axp_3d', 'axp_4d', 'axp_5d', 'axp_6d', 'axp_7d', 'xp_0dv', 'xp_1dv', 'xp_2dv', 'xp_3dv', 'xp_4dv', 'xp_5dv', 'xp_6dv', 'xp_7dv', 'axp_0dv', 'axp_1dv', 'axp_2dv', 'axp_3dv', 'axp_4dv', 'axp_5dv', 'axp_6dv', 'axp_7dv', 'c1sp1', 'c2sp1', 'c1sp2', 'c2sp2', 'c3sp2', 'c1sp3', 'c2sp3', 'c3sp3', 'c4sp3', 'hybratio', 'fcsp3', 'num_stereocenters', 'num_unspecified_stereocenters', 'num_defined_stereocenters', 'num_r_centers', 'num_s_centers', 'num_stereobonds', 'num_e_bonds', 'num_z_bonds', 'stereo_complexity', 'frac_defined_stereo'],
     "compressed_features": [],
-    "model_metrics_s3_path": "s3://ideaya-sageworks-bucket/models/sol-class-f1-100/training",
+    "model_metrics_s3_path": "s3://ideaya-sageworks-bucket/models/pka-a1-reg-0-nightly-100-test/training",
     "train_all_data": True
 }
 

{workbench-0.8.175 → workbench-0.8.177}/src/workbench/utils/chem_utils/mol_descriptors.py

@@ -91,16 +91,27 @@ import logging
 import pandas as pd
 import numpy as np
 import re
+import time
+from contextlib import contextmanager
 from rdkit import Chem
 from rdkit.Chem import Descriptors, rdCIPLabeler
 from rdkit.ML.Descriptors import MoleculeDescriptors
 from mordred import Calculator as MordredCalculator
 from mordred import AcidBase, Aromatic, Constitutional, Chi, CarbonTypes
 
+
 logger = logging.getLogger("workbench")
 logger.setLevel(logging.DEBUG)
 
 
+# Helper context manager for timing
+@contextmanager
+def timer(name):
+    start = time.time()
+    yield
+    print(f"{name}: {time.time() - start:.2f}s")
+
+
 def compute_stereochemistry_features(mol):
     """
     Compute stereochemistry descriptors using modern RDKit methods.
@@ -280,9 +291,11 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
                 descriptor_values.append([np.nan] * len(all_descriptors))
 
     # Create RDKit features DataFrame
-    rdkit_features_df = pd.DataFrame(descriptor_values, columns=calc.GetDescriptorNames(), index=result.index)
+    rdkit_features_df = pd.DataFrame(descriptor_values, columns=calc.GetDescriptorNames())
 
     # Add RDKit features to result
+    # Remove any columns from result that exist in rdkit_features_df
+    result = result.drop(columns=result.columns.intersection(rdkit_features_df.columns))
     result = pd.concat([result, rdkit_features_df], axis=1)
 
     # Compute Mordred descriptors
@@ -299,7 +312,7 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
 
         # Compute Mordred descriptors
         valid_mols = [mol if mol is not None else Chem.MolFromSmiles("C") for mol in molecules]
-        mordred_df = calc.pandas(valid_mols, nproc=1)  # For serverless, use nproc=1
+        mordred_df = calc.pandas(valid_mols, nproc=1)  # Endpoint multiprocessing will fail with nproc>1
 
         # Replace values for invalid molecules with NaN
         for i, mol in enumerate(molecules):
@@ -310,10 +323,9 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
         for col in mordred_df.columns:
             mordred_df[col] = pd.to_numeric(mordred_df[col], errors="coerce")
 
-        # Set index to match result DataFrame
-        mordred_df.index = result.index
-
         # Add Mordred features to result
+        # Remove any columns from result that exist in mordred
+        result = result.drop(columns=result.columns.intersection(mordred_df.columns))
         result = pd.concat([result, mordred_df], axis=1)
 
     # Compute stereochemistry features if requested
@@ -326,9 +338,10 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
             stereo_features.append(stereo_dict)
 
         # Create stereochemistry DataFrame
-        stereo_df = pd.DataFrame(stereo_features, index=result.index)
+        stereo_df = pd.DataFrame(stereo_features)
 
         # Add stereochemistry features to result
+        result = result.drop(columns=result.columns.intersection(stereo_df.columns))
         result = pd.concat([result, stereo_df], axis=1)
 
         logger.info(f"Added {len(stereo_df.columns)} stereochemistry descriptors")
@@ -357,7 +370,6 @@ def compute_descriptors(df: pd.DataFrame, include_mordred: bool = True, include_
 
 
 if __name__ == "__main__":
-    import time
     from mol_standardize import standardize
     from workbench.api import DataSource