rcsb-embedding-model 0.0.20__tar.gz → 0.0.22__tar.gz

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rcsb-embedding-model
-Version: 0.0.20
+Version: 0.0.22
 Summary: Protein Embedding Model for Structure Search
 Project-URL: Homepage, https://github.com/rcsb/rcsb-embedding-model
 Project-URL: Issues, https://github.com/rcsb/rcsb-embedding-model/issues
@@ -17,7 +17,7 @@ Description-Content-Type: text/markdown
 
 # RCSB Embedding Model
 
-**Version** 0.0.20
+**Version** 0.0.22
 
 
 ## Overview
@@ -48,7 +48,7 @@ If you are interested in training the model with a new dataset, visit the [rcsb-
 **Requirements:**
 
 - Python ≥ 3.10
-- ESM == 3.1.1
+- ESM >= 3.2.0
 - Lightning ≥ 2.5.0 
 - Typer ≥ 0.15.0
 

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/README.md

@@ -1,6 +1,6 @@
 # RCSB Embedding Model
 
-**Version** 0.0.20
+**Version** 0.0.22
 
 
 ## Overview
@@ -31,7 +31,7 @@ If you are interested in training the model with a new dataset, visit the [rcsb-
 **Requirements:**
 
 - Python ≥ 3.10
-- ESM == 3.1.1
+- ESM >= 3.2.0
 - Lightning ≥ 2.5.0 
 - Typer ≥ 0.15.0
 

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "rcsb-embedding-model"
-version = "0.0.20"
+version = "0.0.22"
 authors = [
     { name="Joan Segura", email="joan.segura@rcsb.org" },
 ]

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/src/rcsb_embedding_model/cli/inference.py

@@ -8,6 +8,9 @@ from rcsb_embedding_model.types.api_types import StructureFormat, Accelerator, S
     StructureLocation, SrcAssemblyFrom, SrcTensorFrom, OutFormat
 from rcsb_embedding_model.utils.data import adapt_csv_to_embedding_chain_stream
 
+import os
+os.environ["TOKENIZERS_PARALLELISM"] = "false"
+
 app = typer.Typer(
     add_completion=False
 )
@@ -322,26 +325,36 @@ def complete_embedding(
             resolve_path=True,
             help='CSV file 4 columns: Structure Name | Structure File Path or URL (switch structure-location) | Assembly Id | Output embedding name.'
         )],
-        output_path: Annotated[typer.FileText, typer.Option(
+        output_res_path: Annotated[typer.FileText, typer.Option(
             exists=True,
             file_okay=False,
             dir_okay=True,
             resolve_path=True,
-            help='Output path to store predictions. Embeddings are stored as a single DataFrame file (see output_name).'
+            help='Output path to store residue embeddings. Residue embeddings are stored in separated files'
         )],
-        res_embedding_location: Annotated[typer.FileText, typer.Option(
+        output_chain_path: Annotated[typer.FileText, typer.Option(
+            exists=True,
+            file_okay=False,
+            dir_okay=True,
+            resolve_path=True,
+            help='Output path to store chain embeddings. Embeddings are stored as a single JSON file (see output_chain_name).'
+        )],
+        output_assembly_path: Annotated[typer.FileText, typer.Option(
             exists=True,
             file_okay=False,
             dir_okay=True,
             resolve_path=True,
-            help='Output path to store ESM predictions.'
+            help='Output path to store assembly embeddings. Embeddings are stored as a single JSON file (see output_assembly_name).'
         )],
         output_format: Annotated[OutFormat, typer.Option(
             help='Format of the output. Options: separated (predictions are stored in single files) or grouped (predictions are stored in a single JSON file).'
         )] = OutFormat.separated,
-        output_name: Annotated[str, typer.Option(
-            help='File name for storing embeddings as a single JSON file. Used when output-format=grouped.'
-        )] = 'inference',
+        output_chain_name: Annotated[str, typer.Option(
+            help='File name for storing chain embeddings as a single JSON file. Used when output-format=grouped.'
+        )] = 'chain-inference',
+        output_assembly_name: Annotated[str, typer.Option(
+            help='File name for storing chain embeddings as a single JSON file. Used when output-format=grouped.'
+        )] = 'chain-inference',
         structure_location: Annotated[StructureLocation, typer.Option(
             help='Structure file location.'
         )] = StructureLocation.local,
@@ -351,10 +364,22 @@ def complete_embedding(
         min_res_n: Annotated[int, typer.Option(
             help='When using all chains in a structure, consider only chains with more than <min_res_n> residues.'
         )] = 0,
-        batch_size: Annotated[int, typer.Option(
+        batch_size_res: Annotated[int, typer.Option(
             help='Number of samples processed together in one iteration.'
         )] = 1,
-        num_workers: Annotated[int, typer.Option(
+        num_workers_res: Annotated[int, typer.Option(
+            help='Number of subprocesses to use for data loading.'
+        )] = 0,
+        batch_size_chain: Annotated[int, typer.Option(
+            help='Number of samples processed together in one iteration.'
+        )] = 1,
+        num_workers_chain: Annotated[int, typer.Option(
+            help='Number of subprocesses to use for data loading.'
+        )] = 0,
+        batch_size_assembly: Annotated[int, typer.Option(
+            help='Number of samples processed together in one iteration.'
+        )] = 1,
+        num_workers_assembly: Annotated[int, typer.Option(
             help='Number of subprocesses to use for data loading.'
         )] = 0,
         num_nodes: Annotated[int, typer.Option(
@@ -369,43 +394,43 @@ def complete_embedding(
 ):
     residue_embedding(
         src_file=src_chain_file,
-        output_path=res_embedding_location,
+        output_path=output_res_path,
         output_format=OutFormat.separated,
         structure_location=structure_location,
         structure_format=structure_format,
         min_res_n=min_res_n,
-        batch_size=batch_size,
-        num_workers=num_workers,
+        batch_size=batch_size_res,
+        num_workers=num_workers_res,
         num_nodes=num_nodes,
         accelerator=accelerator,
         devices=devices,
     )
     chain_embedding(
         src_file=src_chain_file,
-        output_path=output_path,
+        output_path=output_chain_path,
         output_format=output_format,
-        output_name=f"{output_name}-chain",
-        res_embedding_location=res_embedding_location,
+        output_name=output_chain_name,
+        res_embedding_location=output_res_path,
         structure_location=structure_location,
         structure_format=structure_format,
         min_res_n=min_res_n,
-        batch_size=batch_size,
-        num_workers=num_workers,
+        batch_size=batch_size_chain,
+        num_workers=num_workers_chain,
         num_nodes=num_nodes,
         accelerator=accelerator,
         devices=devices
     )
     assembly_embedding(
         src_file=src_assembly_file,
-        output_path=output_path,
+        output_path=output_assembly_path,
         output_format=output_format,
-        output_name=f"{output_name}-assembly",
-        res_embedding_location=res_embedding_location,
+        output_name=output_assembly_name,
+        res_embedding_location=output_res_path,
         structure_location=structure_location,
         structure_format=structure_format,
         min_res_n=min_res_n,
-        batch_size=batch_size,
-        num_workers=num_workers,
+        batch_size=batch_size_assembly,
+        num_workers=num_workers_assembly,
         num_nodes=num_nodes,
         accelerator=accelerator,
         devices=devices

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/src/rcsb_embedding_model/inference/chain_inference.py

@@ -7,6 +7,7 @@ from rcsb_embedding_model.modules.chain_module import ChainModule
 from rcsb_embedding_model.types.api_types import Accelerator, Devices, OptionalPath, FileOrStreamTuple, SrcLocation, \
     SrcTensorFrom, StructureLocation, StructureFormat, OutFormat
 from rcsb_embedding_model.utils.data import collate_seq_embeddings
+from rcsb_embedding_model.utils.model import get_aggregator_model
 from rcsb_embedding_model.writer.batch_writer import CsvBatchWriter, JsonStorage
 
 
@@ -52,13 +53,17 @@ def predict(
         )
     )
 
-    module = ChainModule()
+    aggregator_model = get_aggregator_model()
+    module = ChainModule(
+        model=aggregator_model
+    )
     inference_writer = (JsonStorage(out_path, out_name) if out_format == OutFormat.grouped else CsvBatchWriter(out_path)) if out_path is not None else None
     trainer = Trainer(
         callbacks=[inference_writer] if inference_writer is not None else None,
         num_nodes=num_nodes,
         accelerator=accelerator,
         devices=devices,
+        strategy="ddp",
         logger=False
     )
 

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/src/rcsb_embedding_model/inference/esm_inference.py

@@ -6,6 +6,7 @@ from rcsb_embedding_model.dataset.esm_prot_from_chain import EsmProtFromChain
 from rcsb_embedding_model.modules.esm_module import EsmModule
 from rcsb_embedding_model.types.api_types import StructureFormat, Accelerator, Devices, OptionalPath, StructureLocation, \
     SrcProteinFrom, FileOrStreamTuple, SrcLocation, OutFormat
+from rcsb_embedding_model.utils.model import get_residue_model
 from rcsb_embedding_model.writer.batch_writer import TensorBatchWriter, JsonStorage
 
 
@@ -46,13 +47,17 @@ def predict(
         collate_fn=lambda _: _
     )
 
-    module = EsmModule()
+    esm_model = get_residue_model()
+    module = EsmModule(
+        model=esm_model
+    )
     inference_writer = (JsonStorage(out_path, out_name) if out_format == OutFormat.grouped else TensorBatchWriter(out_path)) if out_path is not None else None
     trainer = Trainer(
         callbacks=[inference_writer] if inference_writer is not None else None,
         num_nodes=num_nodes,
         accelerator=accelerator,
         devices=devices,
+        strategy="ddp",
         logger=False
     )
 

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/src/rcsb_embedding_model/inference/structure_inference.py

@@ -6,6 +6,7 @@ from rcsb_embedding_model.dataset.esm_prot_from_chain import EsmProtFromChain
 from rcsb_embedding_model.modules.structure_module import StructureModule
 from rcsb_embedding_model.types.api_types import StructureFormat, Accelerator, Devices, OptionalPath, StructureLocation, \
     SrcProteinFrom, FileOrStreamTuple, SrcLocation
+from rcsb_embedding_model.utils.model import get_residue_model, get_aggregator_model
 from rcsb_embedding_model.writer.batch_writer import JsonStorage
 
 
@@ -45,13 +46,19 @@ def predict(
         collate_fn=lambda _: _
     )
 
-    module = StructureModule()
+    res_model = get_residue_model()
+    aggregator_model = get_aggregator_model()
+    module = StructureModule(
+        res_model=res_model,
+        aggregator_model=aggregator_model
+    )
     inference_writer = JsonStorage(out_path, out_name) if out_path is not None and out_name is not None else None
     trainer = Trainer(
         callbacks=[inference_writer] if inference_writer is not None else None,
         num_nodes=num_nodes,
         accelerator=accelerator,
         devices=devices,
+        strategy="ddp",
         logger=False
     )
 

rcsb_embedding_model-0.0.22/src/rcsb_embedding_model/modules/chain_module.py

@@ -0,0 +1,19 @@
+import logging
+
+from lightning import LightningModule
+
+logger = logging.getLogger(__name__)
+
+class ChainModule(LightningModule):
+
+    def __init__(
+            self,
+            model
+    ):
+        super().__init__()
+        logger.info(f"Using device: {self.device}")
+        self.aggregator = model
+
+    def predict_step(self, batch, batch_idx):
+        (x, x_mask), dom_id = batch
+        return self.aggregator(x, x_mask), dom_id

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/src/rcsb_embedding_model/modules/esm_module.py

@@ -1,16 +1,19 @@
+import logging
+
 from esm.sdk.api import SamplingConfig
 from lightning import LightningModule
 
-from rcsb_embedding_model.utils.model import get_residue_model
-
+logger = logging.getLogger(__name__)
 
 class EsmModule(LightningModule):
 
     def __init__(
-            self
+            self,
+            model
     ):
         super().__init__()
-        self.esm3 = get_residue_model(self.device)
+        logger.info(f"Using device: {self.device}")
+        self.esm3 = model
 
     def predict_step(self, prot_batch, batch_idx):
         return tuple([self.__compute_embeddings(esm_prot) for esm_prot, name in prot_batch]), tuple([name for esm_prot, name in prot_batch])

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/src/rcsb_embedding_model/modules/structure_module.py

@@ -1,18 +1,23 @@
+import logging
+
 from esm.sdk.api import SamplingConfig
 from lightning import LightningModule
 
 from rcsb_embedding_model.utils.data import collate_seq_embeddings
-from rcsb_embedding_model.utils.model import get_residue_model, get_aggregator_model
 
+logger = logging.getLogger(__name__)
 
 class StructureModule(LightningModule):
 
     def __init__(
-            self
+            self,
+            res_model,
+            aggregator_model
     ):
         super().__init__()
-        self.esm3 = get_residue_model(self.device)
-        self.aggregator = get_aggregator_model(device=self.device)
+        logger.info(f"Using device: {self.device}")
+        self.esm3 = res_model
+        self.aggregator =  aggregator_model
 
     def predict_step(self, prot_batch, batch_idx):
         prot_embeddings = []

{rcsb_embedding_model-0.0.20 → rcsb_embedding_model-0.0.22}/src/rcsb_embedding_model/utils/model.py

@@ -16,6 +16,8 @@ def get_aggregator_model(device=None):
         filename=FILE_NAME,
         revision=REVISION
     )
+    if device is None:
+        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
     weights = torch.load(model_path, weights_only=True, map_location=device)
     aggregator_model = ResidueEmbeddingAggregator()
     aggregator_model.load_state_dict(weights)

rcsb_embedding_model-0.0.20/src/rcsb_embedding_model/modules/chain_module.py

@@ -1,16 +0,0 @@
-from lightning import LightningModule
-
-from rcsb_embedding_model.utils.model import get_aggregator_model
-
-
-class ChainModule(LightningModule):
-
-    def __init__(
-            self
-    ):
-        super().__init__()
-        self.model = get_aggregator_model(device=self.device)
-
-    def predict_step(self, batch, batch_idx):
-        (x, x_mask), dom_id = batch
-        return self.model(x, x_mask), dom_id