dayhoff-tools 1.15.0__tar.gz → 1.15.1__tar.gz

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dayhoff-tools
-Version: 1.15.0
+Version: 1.15.1
 Summary: Common tools for all the repos at Dayhoff Labs
 Author: Daniel Martin-Alarcon
 Author-email: dma@dayhofflabs.com
@@ -40,8 +40,8 @@ Requires-Dist: sqlalchemy (>=2.0.40,<3.0.0) ; extra == "full"
 Requires-Dist: toml (>=0.10)
 Requires-Dist: tqdm (>=4.67.1) ; extra == "embedders"
 Requires-Dist: tqdm (>=4.67.1) ; extra == "full"
-Requires-Dist: transformers (==4.36.2) ; extra == "full"
 Requires-Dist: transformers (>=4.36.2) ; extra == "embedders"
+Requires-Dist: transformers (>=4.36.2) ; extra == "full"
 Requires-Dist: typer (>=0.9.0)
 Requires-Dist: tzdata (>=2025.2)
 Description-Content-Type: text/markdown

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/batch/workers/boltz.py

@@ -235,7 +235,7 @@ class BoltzProcessor:
         # Determine output directory
         # Boltz always creates boltz_results_{input_name} inside --out_dir
         input_base = os.path.splitext(os.path.basename(input_file))[0]
-        
+
         if output_dir is None:
             # No output_dir specified, boltz creates in current directory
             expected_output_dir = f"boltz_results_{input_base}"
@@ -244,7 +244,9 @@ class BoltzProcessor:
             # output_dir specified - use its parent for --out_dir
             # and expect boltz_results_{input_base} inside it
             parent_dir = os.path.dirname(output_dir)
-            expected_output_dir = os.path.join(parent_dir, f"boltz_results_{input_base}")
+            expected_output_dir = os.path.join(
+                parent_dir, f"boltz_results_{input_base}"
+            )
             out_dir_arg = parent_dir if parent_dir else None
 
         logger.info(f"Running Boltz prediction for {input_file}")
@@ -455,9 +457,7 @@ def main():
                 completed += 1
                 continue
 
-            logger.info(
-                f"[{file_idx + 1}/{len(my_files)}] Processing {file_stem}..."
-            )
+            logger.info(f"[{file_idx + 1}/{len(my_files)}] Processing {file_stem}...")
 
             try:
                 # Determine output directory

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/cli/batch/commands/cancel.py

@@ -152,9 +152,13 @@ def _cancel_retry_job(manifest, retry_id: str, force: bool, base_path: str):
             )
 
         click.echo()
-        click.echo(click.style(f"✓ Retry job {retry_id} cancelled successfully", fg="green"))
+        click.echo(
+            click.style(f"✓ Retry job {retry_id} cancelled successfully", fg="green")
+        )
         click.echo(f"Parent job: {manifest.job_id}")
 
     except BatchError as e:
-        click.echo(click.style(f"✗ Failed to cancel retry job: {e}", fg="red"), err=True)
+        click.echo(
+            click.style(f"✗ Failed to cancel retry job: {e}", fg="red"), err=True
+        )
         raise SystemExit(1)

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/cli/batch/commands/clean.py

@@ -20,7 +20,9 @@ from .status import format_time_ago, _aws_status_to_job_status
     default=7,
     help="Only clean jobs older than N days [default: 7]",
 )
-@click.option("--dry-run", is_flag=True, help="Show what would be cleaned without deleting")
+@click.option(
+    "--dry-run", is_flag=True, help="Show what would be cleaned without deleting"
+)
 @click.option("--force", is_flag=True, help="Delete without confirmation")
 @click.option("--base-path", default=BATCH_JOBS_BASE, help="Base path for job data")
 def clean(user, older_than, dry_run, force, base_path):
@@ -81,7 +83,9 @@ def clean(user, older_than, dry_run, force, base_path):
             live_statuses = client.get_job_statuses_batch(batch_job_ids)
         except BatchError as e:
             click.echo(f"Error: Could not fetch status from AWS Batch: {e}", err=True)
-            click.echo("Cannot safely clean jobs without knowing their status.", err=True)
+            click.echo(
+                "Cannot safely clean jobs without knowing their status.", err=True
+            )
             raise SystemExit(1)
 
     # Find jobs that are safe to clean (SUCCEEDED or FAILED)

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/cli/batch/commands/embed_t5.py

@@ -21,7 +21,6 @@ from ..manifest import (
     save_manifest,
 )
 
-
 # Default settings for T5 embedding
 DEFAULT_QUEUE = "t4-1x-spot"
 DEFAULT_WORKERS = 50

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/cli/batch/commands/finalize.py

@@ -40,7 +40,9 @@ from ..manifest import (
     help="Skip deduplication step (use if input has no duplicates)",
 )
 @click.option("--base-path", default=BATCH_JOBS_BASE, help="Base path for job data")
-def finalize(job_id, output, force, keep_intermediates, full_output, skip_dedup, base_path):
+def finalize(
+    job_id, output, force, keep_intermediates, full_output, skip_dedup, base_path
+):
     """Combine results and clean up job intermediates.
 
     For embedding jobs, combines H5 files into a single output file.
@@ -238,14 +240,18 @@ def _finalize_embeddings(output_dir: Path, output_path: Path, skip_dedup: bool =
                 if skip_dedup:
                     # Skip dedup - optimize directly from combined
                     click.echo("Optimizing chunks...")
-                    optimize_protein_embedding_chunks(str(combined_path), str(output_path))
+                    optimize_protein_embedding_chunks(
+                        str(combined_path), str(output_path)
+                    )
                 else:
                     # Full pipeline: combine -> dedup -> optimize
                     deduped_path = Path(tmpdir) / "deduped.h5"
                     click.echo("Deduplicating...")
                     deduplicate_h5_file(str(combined_path), str(deduped_path))
                     click.echo("Optimizing chunks...")
-                    optimize_protein_embedding_chunks(str(deduped_path), str(output_path))
+                    optimize_protein_embedding_chunks(
+                        str(deduped_path), str(output_path)
+                    )
 
         click.echo(click.style("✓ H5 files combined successfully", fg="green"))
 
@@ -269,7 +275,7 @@ def _finalize_embeddings(output_dir: Path, output_path: Path, skip_dedup: bool =
 
 def _finalize_boltz(output_dir: Path, output_path: Path, full_output: bool = False):
     """Move Boltz output to destination.
-    
+
     Args:
         output_dir: Source directory containing boltz_results_* folders
         output_path: Destination directory for outputs
@@ -277,20 +283,24 @@ def _finalize_boltz(output_dir: Path, output_path: Path, full_output: bool = Fal
                     extract only essential files (CIF structures and confidence JSON).
     """
     # Find all output directories (one per complex)
-    complex_dirs = [d for d in output_dir.iterdir() if d.is_dir() and d.name.startswith("boltz_results_")]
+    complex_dirs = [
+        d
+        for d in output_dir.iterdir()
+        if d.is_dir() and d.name.startswith("boltz_results_")
+    ]
 
     if not complex_dirs:
         click.echo("No output directories found.", err=True)
         raise SystemExit(1)
 
     click.echo(f"Found {len(complex_dirs)} structure predictions")
-    
+
     if full_output:
         click.echo("Mode: Copying full output (all files)")
     else:
         click.echo("Mode: Extracting essential files only (CIF + confidence JSON)")
         click.echo("       Use --full-output to copy all files")
-    
+
     # Confirm before proceeding
     click.echo()
     if not click.confirm(f"Copy results to {output_path}?"):
@@ -302,16 +312,16 @@ def _finalize_boltz(output_dir: Path, output_path: Path, full_output: bool = Fal
 
     copied_count = 0
     skipped_count = 0
-    
+
     for complex_dir in complex_dirs:
         complex_name = complex_dir.name.replace("boltz_results_", "")
         dest = output_path / complex_name
-        
+
         if dest.exists():
             click.echo(f"  Skipping {complex_name} (already exists)")
             skipped_count += 1
             continue
-        
+
         if full_output:
             # Copy entire directory
             shutil.copytree(complex_dir, dest)
@@ -320,44 +330,53 @@ def _finalize_boltz(output_dir: Path, output_path: Path, full_output: bool = Fal
             # Extract only essential files
             _extract_essential_boltz_files(complex_dir, dest, complex_name)
             click.echo(f"  Extracted {complex_name} (essential files)")
-        
+
         copied_count += 1
 
     click.echo()
     if skipped_count > 0:
-        click.echo(f"Copied {copied_count} predictions, skipped {skipped_count} existing")
+        click.echo(
+            f"Copied {copied_count} predictions, skipped {skipped_count} existing"
+        )
     else:
-        click.echo(click.style(f"✓ Copied {copied_count} structure predictions successfully", fg="green"))
+        click.echo(
+            click.style(
+                f"✓ Copied {copied_count} structure predictions successfully",
+                fg="green",
+            )
+        )
 
 
 def _extract_essential_boltz_files(source_dir: Path, dest_dir: Path, complex_name: str):
     """Extract only essential files from Boltz output.
-    
+
     Essential files are:
     - predictions/*/*.cif (structure files)
     - predictions/*/confidence_*.json (confidence metrics)
-    
+
     Args:
         source_dir: Source boltz_results_* directory
         dest_dir: Destination directory to create
         complex_name: Name of the complex (for better error messages)
     """
     dest_dir.mkdir(parents=True, exist_ok=True)
-    
+
     predictions_dir = source_dir / "predictions"
     if not predictions_dir.exists():
-        click.echo(f"    Warning: No predictions directory found for {complex_name}", err=True)
+        click.echo(
+            f"    Warning: No predictions directory found for {complex_name}", err=True
+        )
         return
-    
+
     # Find all subdirectories in predictions/ (usually just one named after the complex)
     for pred_subdir in predictions_dir.iterdir():
         if not pred_subdir.is_dir():
             continue
-        
+
         # Copy CIF files (structures)
         for cif_file in pred_subdir.glob("*.cif"):
             shutil.copy2(cif_file, dest_dir / cif_file.name)
-        
+
         # Copy confidence JSON files
         for json_file in pred_subdir.glob("confidence_*.json"):
             shutil.copy2(json_file, dest_dir / json_file.name)

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/cli/batch/commands/status.py

@@ -264,7 +264,9 @@ def _show_job_details(job_id: str, base_path: str):
             reslice_info = ""
             if retry.reslice_prefix:
                 reslice_info = f" (resliced to {retry.reslice_count} chunks)"
-            click.echo(f"  - {retry.retry_id}: {len(retry.indices)} indices{reslice_info}")
+            click.echo(
+                f"  - {retry.retry_id}: {len(retry.indices)} indices{reslice_info}"
+            )
             click.echo(f"    Indices: {retry.indices}")
             if retry.batch_job_id:
                 # Show brief status for retry job
@@ -273,7 +275,9 @@ def _show_job_details(job_id: str, base_path: str):
                     array_status = client.get_array_job_status(retry.batch_job_id)
                     if array_status.is_complete:
                         pct = array_status.success_rate * 100
-                        color = "green" if pct == 100 else "yellow" if pct > 90 else "red"
+                        color = (
+                            "green" if pct == 100 else "yellow" if pct > 90 else "red"
+                        )
                         click.echo(
                             f"    Status: Complete - {click.style(f'{pct:.0f}%', fg=color)} "
                             f"({array_status.succeeded}/{array_status.total} succeeded)"
@@ -364,7 +368,9 @@ def _show_retry_details(manifest, retry_id: str):
     click.echo("Retry Config:")
     click.echo(f"  Indices:   {retry_info.indices}")
     if retry_info.reslice_prefix:
-        click.echo(f"  Reslice:   {retry_info.reslice_prefix} ({retry_info.reslice_count} chunks)")
+        click.echo(
+            f"  Reslice:   {retry_info.reslice_prefix} ({retry_info.reslice_count} chunks)"
+        )
     else:
         click.echo(f"  Reslice:   No (retrying original chunks)")
 

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/cli/batch/commands/submit.py

@@ -18,7 +18,6 @@ from ..manifest import (
     save_manifest,
 )
 
-
 # Default job definition for generic jobs
 DEFAULT_JOB_DEFINITION = "dayhoff-batch-base"
 DEFAULT_QUEUE = "t4-1x-spot"

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/cli/utility_commands.py

@@ -258,5 +258,3 @@ def build_and_upload_wheel(bump_part: str = "patch"):
                     print(f"Warning: Could not find version {new_version} to revert.")
             except Exception as revert_e:
                 print(f"Warning: Failed to revert version change: {revert_e}")
-
-

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/embedders.py

@@ -179,8 +179,8 @@ class H5Reformatter(Processor):
     def embedding_file_to_df(self, file_name: str) -> pd.DataFrame:
         with h5py.File(file_name, "r") as f:
             gene_names = list(f.keys())
-            Xg = [f[key][()] for key in gene_names]  # type:ignore
-        return pd.DataFrame(np.asmatrix(Xg), index=gene_names)  # type:ignore
+            Xg = [f[key][()] for key in gene_names]  # type: ignore
+        return pd.DataFrame(np.asmatrix(Xg), index=gene_names)  # type: ignore
 
     def write_df_to_h5(self, df: pd.DataFrame, filename: str, description: str) -> None:
         """

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/fasta.py

@@ -857,14 +857,12 @@ def fasta_to_sqlite(fasta_file: str, db_file: str, batch_size: int = 1000) -> No
     # Create the SQLite database and table
     print("Creating SQLite database...")
     with sqlite3.connect(db_file) as conn:
-        conn.execute(
-            """
+        conn.execute("""
             CREATE TABLE IF NOT EXISTS proteins (
                 protein_id TEXT PRIMARY KEY,
                 sequence TEXT NOT NULL
             )
-        """
-        )
+        """)
     print("Database created successfully.")
 
     # Estimate number of records for progress bar

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/intake/kegg.py

@@ -25,9 +25,7 @@ def get_ko2gene_df(db: str, ko: str | list[str] | None = None) -> pd.DataFrame:
         query = (
             f"SELECT gene,ko FROM gene_to_ko WHERE ko IN ({','.join('?' * len(ko))})"
         )
-        result_df = pd.read_sql_query(
-            query, conn, params=ko  # type:ignore
-        )
+        result_df = pd.read_sql_query(query, conn, params=ko)  # type: ignore
     else:
         query = f"SELECT gene,ko FROM gene_to_ko"
         result_df = pd.read_sql_query(query, conn)

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/dayhoff_tools/intake/structure.py

@@ -409,10 +409,10 @@ class PDBFolderProcessor:
     def _get_pdb_files(self) -> list[str]:
         """
         Get a list of PDB files in the specified directory, optionally filtered by ID set.
-        Files are sorted by creation time to ensure consistent processing order.
+        Files are sorted alphabetically to ensure consistent, reproducible processing order.
 
         Returns:
-            List of PDB file names sorted by creation time.
+            List of PDB file names sorted alphabetically.
         """
         print("Scanning directory for PDB files...")
         pdb_files = [
@@ -424,8 +424,8 @@ class PDBFolderProcessor:
                 f for f in pdb_files if self._extract_id_from_filename(f) in self.id_set
             ]
 
-        # Sort files by creation time
-        pdb_files.sort(key=lambda f: os.path.getctime(os.path.join(self.pdb_dir, f)))
+        # Sort files alphabetically for deterministic, reproducible order
+        pdb_files.sort()
 
         print(f"Found {len(pdb_files)} PDB files")
         return pdb_files

{dayhoff_tools-1.15.0 → dayhoff_tools-1.15.1}/pyproject.toml

@@ -11,7 +11,7 @@ build-backend = "poetry.core.masonry.api"
 
 [project]
 name = "dayhoff-tools"
-version = "1.15.0"
+version = "1.15.1"
 description = "Common tools for all the repos at Dayhoff Labs"
 authors = [
     {name = "Daniel Martin-Alarcon", email = "dma@dayhofflabs.com"}
@@ -43,7 +43,7 @@ full = [
     "sentencepiece>=0.2.0",
     "sqlalchemy>=2.0.40,<3.0.0",
     "tqdm>=4.67.1",
-    "transformers==4.36.2",
+    "transformers>=4.36.2",  # Relaxed: exact pin broke conda envs with huggingface-hub>=1.0
 ]
 
 # Embedding models (requires torch - user must install separately for their platform)