ms2pip - npm Package Compare versions

Comparing version 4.0.0.dev13 to 4.0.0.dev14

ms2pip/_utils/ion_mobility.py

+"""Module for ion mobility prediction with IM²Deep."""
+
+import logging
+
+import pandas as pd
+from psm_utils import PSMList
+
+logger = logging.getLogger(__name__)
+
+
+class IonMobility:
+    """Predict ion mobility using IM²Deep."""
+
+    def __init__(self, processes=1) -> None:
+        # Lazy import to avoid loading loading heavy dependencies when not needed
+        try:
+            from im2deep.im2deep import predict_ccs  # noqa: F401
+
+            self.predict_fn = predict_ccs
+            self.processes = processes
+        except ImportError as e:
+            raise ImportError(
+                "The 'im2deep' package is required for ion mobility prediction."
+            ) from e
+
+    def add_im_predictions(self, psm_list: PSMList) -> None:
+        """Add ion mobility predictions to the PSMList."""
+        logger.info("Predicting ion mobility...")
+        predictions: pd.Series = self.predict_fn(
+            psm_list, write_output=False, n_jobs=self.processes
+        )
+        psm_list["ion_mobility"] = predictions.values

ms2pip.egg-info/PKG-INFO

 Metadata-Version: 2.1
 Name: ms2pip
-Version: 4.0.0.dev13
+Version: 4.0.0.dev14
 Summary: MS2PIP: Accurate and versatile peptide fragmentation spectrum prediction.
@@ -5,0 +5,0 @@ Author: Ana Sílvia C. Silva

ms2pip.egg-info/SOURCES.txt

@@ -41,2 +41,3 @@ LICENSE
 ms2pip/_utils/feature_names.py
+ms2pip/_utils/ion_mobility.py
 ms2pip/_utils/psm_input.py
@@ -43,0 +44,0 @@ ms2pip/_utils/retention_time.py

ms2pip/__init__.py

 # isort: skip_file
 """MS2PIP: Accurate and versatile peptide fragmentation spectrum prediction."""
 
-__version__ = "4.0.0-dev13"
+__version__ = "4.0.0-dev14"
 
@@ -6,0 +6,0 @@ from warnings import filterwarnings

ms2pip/__main__.py

@@ -89,5 +89,7 @@ import logging
 @click.argument("psms", required=True)
[e-mail]("--psm-filetype", "-t", type=click.Choice(PSM_FILETYPES), default=None)
 @click.option("--output-name", "-o", type=str)
 @click.option("--output-format", "-f", type=click.Choice(SUPPORTED_FORMATS), default="tsv")
 @click.option("--add-retention-time", "-r", is_flag=True)
[e-mail]("--add-ion-mobility", "-i", is_flag=True)
 @click.option("--model", type=click.Choice(MODELS), default="HCD")
@@ -114,2 +116,3 @@ @click.option("--model-dir")
 @click.option("--add-retention-time", "-r", is_flag=True)
[e-mail]("--add-ion-mobility", "-i", is_flag=True)
 @click.option("--model", type=click.Choice(MODELS), default="HCD")
@@ -141,2 +144,3 @@ @click.option("--model-dir")
 @click.option("--add-retention-time", "-r", is_flag=True)
[e-mail]("--add-ion-mobility", "-i", is_flag=True)
 @click.option("--model", type=click.Choice(MODELS), default="HCD")
@@ -143,0 +147,0 @@ @click.option("--model-dir")

ms2pip/core.py

@@ -26,2 +26,3 @@ #!/usr/bin/env python
 from ms2pip._utils.retention_time import RetentionTime
+from ms2pip._utils.ion_mobility import IonMobility
 from ms2pip._utils.xgb_models import get_predictions_xgb, validate_requested_xgb_model
@@ -78,2 +79,3 @@ from ms2pip.constants import MODELS
     add_retention_time: bool = False,
+    add_ion_mobility: bool = False,
     psm_filetype: Optional[str] = None,
@@ -96,2 +98,4 @@ model: Optional[str] = "HCD",
         Add retention time predictions with DeepLC (Requires optional DeepLC dependency).
+    add_ion_mobility
+        Add ion mobility predictions with IM2Deep (Requires optional IM2Deep dependency).
     model
@@ -119,2 +123,7 @@ Model to use for prediction. Default: "HCD".
 
+    if add_ion_mobility:
+        logger.info("Adding ion mobility predictions")
+        im_predictor = IonMobility(processes=processes)
+        im_predictor.add_im_predictions(psm_list)
+
     with Encoder.from_psm_list(psm_list) as encoder:
@@ -137,2 +146,3 @@ ms2pip_parallelized = _Parallelized(
     add_retention_time: bool = False,
+    add_ion_mobility: bool = False,
     model: Optional[str] = "HCD",
@@ -156,2 +166,4 @@ model_dir: Optional[Union[str, Path]] = None,
         Add retention time predictions with DeepLC (Requires optional DeepLC dependency).
+    add_ion_mobility
+        Add ion mobility predictions with IM2Deep (Requires optional IM2Deep dependency).
     model
@@ -166,2 +178,7 @@ Model to use for prediction. Default: "HCD".
 
+    Yields
+    ------
+    predictions: List[ProcessingResult]
+        Predicted spectra with theoretical m/z and predicted intensity values.
+
     """
@@ -193,2 +210,3 @@ if fasta_file and config:
             add_retention_time=add_retention_time,
+            add_ion_mobility=add_ion_mobility,
             model=model,
@@ -208,2 +226,3 @@ model_dir=model_dir,
     add_retention_time: bool = False,
+    add_ion_mobility: bool = False,
     model: Optional[str] = "HCD",
@@ -233,2 +252,4 @@ model_dir: Optional[Union[str, Path]] = None,
         Add retention time predictions with DeepLC (Requires optional DeepLC dependency).
+    add_ion_mobility
+        Add ion mobility predictions with IM2Deep (Requires optional IM2Deep dependency).
     model
@@ -258,2 +279,7 @@ Model to use for prediction. Default: "HCD".
 
+    if add_ion_mobility:
+        logger.info("Adding ion mobility predictions")
+        im_predictor = IonMobility(processes=processes)
+        im_predictor.add_im_predictions(psm_list)
+
     with Encoder.from_psm_list(psm_list) as encoder:
@@ -860,4 +886,4 @@ ms2pip_parallelized = _Parallelized(
             raise exceptions.TitlePatternError(
-                "Spectrum title pattern could not be matched to spectrum IDs "
-                f"`{spectrum.identifier}`. "
+                f"Spectrum title pattern `{spectrum_id_pattern}` could not be matched to "
+                f"spectrum ID `{spectrum.identifier}`. "
                 " Are you sure that the regex contains a capturing group?"
@@ -864,0 +890,0 @@ )

ms2pip/spectrum_input.py

@@ -37,3 +37,3 @@ """Read MS2 spectra."""
     for spectrum in get_ms2_spectra(str(spectrum_file)):
-        yield ObservedSpectrum(
+        obs_spectrum = ObservedSpectrum(
             mz=np.array(spectrum.mz, dtype=np.float32),
@@ -46,2 +46,10 @@ intensity=np.array(spectrum.intensity, dtype=np.float32),
         )
+        # Workaround for mobiusklein/mzdata#3
+        if (
+            obs_spectrum.identifier == ""
+            or obs_spectrum.mz.shape[0] == 0
+            or obs_spectrum.intensity.shape[0] == 0
+        ):
+            continue
+        yield obs_spectrum
 
@@ -48,0 +56,0 @@

ms2pip/spectrum_output.py

@@ -160,2 +160,3 @@ """
         "rt",
+        "im",
     ]
@@ -198,2 +199,3 @@
             "rt": result.psm.retention_time if result.psm.retention_time else None,
+            "im": result.psm.ion_mobility if result.psm.ion_mobility else None,
         }
@@ -248,5 +250,8 @@
             if len(modifications) > 1:
-                raise ValueError("Multiple modifications per amino acid not supported.")
+                raise ValueError("Multiple modifications per amino acid not supported in MSP.")
             modification = modifications[0]
-            return f"{position},{amino_acid},{modification.name}"
+            try:
+                return f"{position},{amino_acid},{modification.name}"
+            except AttributeError:  # MassModification has no attribute `name`
+                return f"{position},{amino_acid},{modification.value}"
 
@@ -293,2 +298,10 @@ sequence_mods = [
     @staticmethod
+    def _format_ion_mobility(psm: PSM) -> Union[str, None]:
+        """Format ion mobility as string."""
+        if psm.ion_mobility:
+            return f"IonMobility={psm.ion_mobility}"
+        else:
+            return None
+
+    @staticmethod
     def _format_identifier(psm: PSM) -> str:
@@ -309,2 +322,3 @@ """Format MS2PIP ID as string."""
                     MSP._format_retention_time(psm),
+                    MSP._format_ion_mobility(psm),
                     MSP._format_identifier(psm),
@@ -318,4 +332,8 @@ ],
 class MGF(_Writer):
-    """Write MGF files from MS2PIP processing results."""
+    """
+    Write MGF files from MS2PIP processing results.
 
+    See http://www.matrixscience.com/help/data_file_help.html for documentation on the MGF format.
+    """
+
     suffix = ".mgf"
@@ -342,2 +360,3 @@
             f"RTINSECONDS={result.psm.retention_time}" if result.psm.retention_time else None,
+            f"ION_MOBILITY={result.psm.ion_mobility}" if result.psm.ion_mobility else None,
         ]
@@ -438,3 +457,5 @@
 
-    Bibliospec SSL and MS2 files are also compatible with Skyline.
+    Bibliospec SSL and MS2 files are also compatible with Skyline. See
+    https://skyline.ms/wiki/home/software/BiblioSpec/page.view?name=BiblioSpec%20input%20and%20output%20file%20formats
+    for documentation on the Bibliospec file formats.
 
@@ -453,2 +474,3 @@ """
         "retention-time",
+        "ion-mobility",
     ]
@@ -563,2 +585,3 @@
                 "retention-time": result.psm.retention_time if result.psm.retention_time else None,
+                "ion-mobility": result.psm.ion_mobility if result.psm.ion_mobility else None,
             }
@@ -565,0 +588,0 @@ )

PKG-INFO

 Metadata-Version: 2.1
 Name: ms2pip
-Version: 4.0.0.dev13
+Version: 4.0.0.dev14
 Summary: MS2PIP: Accurate and versatile peptide fragmentation spectrum prediction.
@@ -5,0 +5,0 @@ Author: Ana Sílvia C. Silva

Alert delta unavailable

Currently unable to show alert delta for PyPI packages.

Improved metrics

Total package byte prevSize

35502734

0.01%

Number of package files

50

2.04%

Lines of code

3960

2.06%