CoreMS 2.0.9__tar.gz → 2.2.1__tar.gz

{CoreMS-2.0.9 → CoreMS-2.2.1}/CoreMS.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: CoreMS
-Version: 2.0.9
+Version: 2.2.1
 Summary: Mass Spectrometry Framework for Small Molecules Analysis
 Home-page: https://github.com/EMSL-Computing/CoreMS
 Author: Corilo, Yuri
@@ -66,7 +66,7 @@ CoreMS aims to provide
 
 ## Current Version
 
- `2.0.9`
+ `2.2.1`
 
 ***
 
@@ -339,7 +339,7 @@ UML (unified modeling language) diagrams for Direct Infusion FT-MS and GC-MS cla
 
 If you use CoreMS in your work, please use the following citation:
 
-Version [2.0.9 Release on GitHub](https://github.com/EMSL-Computing/CoreMS/releases/tag/v2.0.9), archived on Zenodo:  
+Version [2.2.1 Release on GitHub](https://github.com/EMSL-Computing/CoreMS/releases/tag/v2.2.1), archived on Zenodo:  
 
 [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.4641552.svg)](https://doi.org/10.5281/zenodo.4641552)
 

{CoreMS-2.0.9 → CoreMS-2.2.1}/CoreMS.egg-info/SOURCES.txt

@@ -50,8 +50,6 @@ corems/mass_spectra/input/brukerSolarix.py
 corems/mass_spectra/input/coremsHDF5.py
 corems/mass_spectra/input/massList.py
 corems/mass_spectra/input/rawFileReader.py
-corems/mass_spectra/input/win_only/BrukerCompassXtract.py
-corems/mass_spectra/input/win_only/ThermoMSFileReader.py
 corems/mass_spectra/input/win_only/__init__.py
 corems/mass_spectra/output/__init__.py
 corems/mass_spectra/output/export.py
@@ -125,9 +123,11 @@ ext_lib/ThermoFisher.CommonCore.BackgroundSubtraction.xml
 ext_lib/ThermoFisher.CommonCore.Data.dll
 ext_lib/ThermoFisher.CommonCore.Data.xml
 ext_lib/ThermoFisher.CommonCore.MassPrecisionEstimator.dll
+ext_lib/ThermoFisher.CommonCore.MassPrecisionEstimator.xml
 ext_lib/ThermoFisher.CommonCore.RawFileReader.dll
 ext_lib/ThermoFisher.CommonCore.RawFileReader.xml
 ext_lib/__init__.py
+ext_lib/version
 ext_lib/dotnet/OpenMcdf.Extensions.dll
 ext_lib/dotnet/OpenMcdf.dll
 ext_lib/dotnet/OpenMcdf.xml
@@ -145,8 +145,6 @@ ext_lib/dotnet/ThermoFisher.CommonCore.MassPrecisionEstimator.dll
 ext_lib/dotnet/ThermoFisher.CommonCore.MassPrecisionEstimator.xml
 ext_lib/dotnet/ThermoFisher.CommonCore.RawFileReader.dll
 ext_lib/dotnet/ThermoFisher.CommonCore.RawFileReader.xml
-ext_lib/dotnet/__init__.py
-tests/test.py
 tests/test_LossFinder.py
 tests/test_assembly_identification.py
 tests/test_calibration.py
@@ -155,11 +153,12 @@ tests/test_gcms.py
 tests/test_input.py
 tests/test_massErrorPrediction.py
 tests/test_mass_spectrum.py
-tests/test_molecularFormula.py
-tests/test_molecularFormulaDBFactory.py
-tests/test_molecularFormulaSearch.py
+tests/test_mass_spectrum_export_import.py
+tests/test_molecular_formula.py
+tests/test_molecular_formula_db_factory.py
+tests/test_molecular_formula_search.py
 tests/test_mspeak.py
 tests/test_mz_search.py
 tests/test_output.py
-tests/test_searchMassListRef.py
+tests/test_search_mass_list.py
 tests/test_setting_settings.py

{CoreMS-2.0.9 → CoreMS-2.2.1}/CoreMS.egg-info/requires.txt

@@ -1,6 +1,5 @@
 beautifulsoup4~=4.12.0
 chardet~=5.1.0
-comtypes~=1.1.14
 h5py~=3.8.0
 IsoSpecPy~=2.2.2
 lmfit~=1.1.0
@@ -10,8 +9,6 @@ netCDF4~=1.6.3
 numpy~=1.24.2
 openpyxl~=3.1.1
 pandas~=1.5.3
-psutil~=5.9.4
-PySide2~=5.15.2.1
 pyswarm~=0.6
 pythonnet~=3.0.1
 s3path~=0.4.1

{CoreMS-2.0.9 → CoreMS-2.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: CoreMS
-Version: 2.0.9
+Version: 2.2.1
 Summary: Mass Spectrometry Framework for Small Molecules Analysis
 Home-page: https://github.com/EMSL-Computing/CoreMS
 Author: Corilo, Yuri
@@ -66,7 +66,7 @@ CoreMS aims to provide
 
 ## Current Version
 
- `2.0.9`
+ `2.2.1`
 
 ***
 
@@ -339,7 +339,7 @@ UML (unified modeling language) diagrams for Direct Infusion FT-MS and GC-MS cla
 
 If you use CoreMS in your work, please use the following citation:
 
-Version [2.0.9 Release on GitHub](https://github.com/EMSL-Computing/CoreMS/releases/tag/v2.0.9), archived on Zenodo:  
+Version [2.2.1 Release on GitHub](https://github.com/EMSL-Computing/CoreMS/releases/tag/v2.2.1), archived on Zenodo:  
 
 [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.4641552.svg)](https://doi.org/10.5281/zenodo.4641552)
 

{CoreMS-2.0.9 → CoreMS-2.2.1}/README.md

@@ -50,7 +50,7 @@ CoreMS aims to provide
 
 ## Current Version
 
- `2.0.9`
+ `2.2.1`
 
 ***
 
@@ -323,7 +323,7 @@ UML (unified modeling language) diagrams for Direct Infusion FT-MS and GC-MS cla
 
 If you use CoreMS in your work, please use the following citation:
 
-Version [2.0.9 Release on GitHub](https://github.com/EMSL-Computing/CoreMS/releases/tag/v2.0.9), archived on Zenodo:  
+Version [2.2.1 Release on GitHub](https://github.com/EMSL-Computing/CoreMS/releases/tag/v2.2.1), archived on Zenodo:  
 
 [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.4641552.svg)](https://doi.org/10.5281/zenodo.4641552)
 

{CoreMS-2.0.9 → CoreMS-2.2.1}/corems/__init__.py

@@ -1,5 +1,5 @@
 __author__ = 'Yuri E. Corilo'
-__version__ = '2.0.9'
+__version__ = '2.2.1'
 __doc__ = '''
 <div align="left">
 
@@ -362,94 +362,6 @@ class SuppressPrints:
         sys.stdout.close()
         sys.stdout = self._original_stdout
 
-def get_filenames(app=None):
-
-    from PySide2.QtCore import Qt, QCoreApplication
-    from PySide2.QtWidgets import QApplication, QFileDialog
-    from pathlib import Path
-
-    app = QApplication(sys.argv)
-    file_dialog = QFileDialog()
-    file_dialog.setWindowFlags(Qt.WindowStaysOnTopHint)
-    file_location, _ = file_dialog.getOpenFileNames()
-
-    if file_location:
-        QCoreApplication.processEvents()
-        return file_location
-
-    else:
-
-        return None
-
-def get_filename(app=None):
-
-    from PySide2.QtCore import Qt, QCoreApplication
-    from PySide2.QtWidgets import QApplication, QFileDialog
-    from pathlib import Path
-
-    app = QApplication(sys.argv)
-    file_dialog = QFileDialog()
-    file_dialog.setWindowFlags(Qt.WindowStaysOnTopHint)
-    file_location, _ = file_dialog.getOpenFileName()
-
-    if file_location:
-        QCoreApplication.processEvents()
-        return Path(file_location)
-
-    else:
-
-        return None
-
-def get_dirname(app=None):
-
-    from PySide2.QtCore import Qt, QCoreApplication
-    from PySide2.QtWidgets import QApplication, QFileDialog 
-    from pathlib import Path
-
-    app = QApplication(sys.argv)
-    file_dialog = QFileDialog()
-    file_dialog.setWindowFlags(Qt.WindowStaysOnTopHint)
-    file_location = file_dialog.getExistingDirectory()
-
-    if file_location:
-        QCoreApplication.processEvents()
-        return Path(file_location)
-
-    else:
-
-        return None
-
-def get_dirnames(app=None):
-
-    from PySide2.QtCore import Qt, QCoreApplication
-    from PySide2.QtWidgets import QApplication, QFileDialog, QTreeView, QListView, QAbstractItemView
-    from pathlib import Path
-
-    if not app:
-        app = QApplication(sys.argv)
-    # file_dialog = QFileDialog()
-    # file_dialog.setWindowFlags(Qt.WindowStaysOnTopHint)
-    # file_location = file_dialog.getOpenFileNames()
-
-    file_dialog = QFileDialog()
-    file_dialog.setFileMode(QFileDialog.DirectoryOnly)
-    file_dialog.setOption(QFileDialog.DontUseNativeDialog, True)
-    file_view = file_dialog.findChild(QListView, 'listView')
-
-    # to make it possible to select multiple directories:
-    if file_view:
-        file_view.setSelectionMode(QAbstractItemView.MultiSelection)
-    f_tree_view = file_dialog.findChild(QTreeView)
-    if f_tree_view:
-        f_tree_view.setSelectionMode(QAbstractItemView.MultiSelection)
-
-    if file_dialog.exec():
-        paths = file_dialog.selectedFiles()
-
-        QCoreApplication.processEvents()
-        for path in paths:
-            yield Path(path)
-
 def chunks(lst, n):
     """Yield successive n-sized chunks from lst."""
     for i in range(0, len(lst), n):
@@ -461,4 +373,4 @@ def corems_md5(fname):
 
     md5_returned = hashlib.sha256(bytes_io).hexdigest()
 
-    return "{}:{}".format("sha256", md5_returned)
+    return "{}:{}".format("sha256", md5_returned)

{CoreMS-2.0.9 → CoreMS-2.2.1}/corems/encapsulation/factory/parameters.py

@@ -1,12 +1,40 @@
+import dataclasses
+
 from corems.encapsulation.factory.processingSetting  import LiquidChromatographSetting, MolecularFormulaSearchSettings, TransientSetting, MassSpecPeakSetting, MassSpectrumSetting
 from corems.encapsulation.factory.processingSetting  import CompoundSearchSettings, GasChromatographSetting
 from corems.encapsulation.factory.processingSetting  import DataInputSetting
 
+def reset_ms_parameters():
+    """Reset the MSParameter class to the default values"""
+    MSParameters.molecular_search = MolecularFormulaSearchSettings()
+    MSParameters.transient = TransientSetting()
+    MSParameters.mass_spectrum = MassSpectrumSetting()
+    MSParameters.ms_peak = MassSpecPeakSetting()
+    MSParameters.data_input = DataInputSetting()
+
+def reset_gcms_parameters():
+    """Reset the GCMSParameters class to the default values"""
+    GCMSParameters.molecular_search = CompoundSearchSettings()
+    GCMSParameters.gc_ms = GasChromatographSetting()
+
+def reset_lcms_parameters():
+    """Reset the LCMSParameters class to the default values"""
+    LCMSParameters.lc_ms = LiquidChromatographSetting()
+    LCMSParameters.mass_spectrum = MassSpectrumSetting()
+    LCMSParameters.ms_peak = MassSpecPeakSetting()
+    LCMSParameters.ms1_molecular_search = MolecularFormulaSearchSettings()
+    LCMSParameters.ms2_molecular_search = MolecularFormulaSearchSettings()
+
 class MSParameters:
     """MSParameters class is used to store the parameters used for the processing of the mass spectrum
     
     Each attibute is a class that contains the parameters for the processing of the mass spectrum, see the corems.encapsulation.factory.processingSetting module for more details.
 
+    Parameters
+    ----------
+    use_defaults: bool, optional
+        if True, the class will be instantiated with the default values, otherwise the current values will be used. Default is False.
+
     Attributes
     -----------
     molecular_search: MolecularFormulaSearchSettings
@@ -19,6 +47,11 @@ class MSParameters:
         MassSpecPeakSetting object
     data_input: DataInputSetting
         DataInputSetting object
+
+    Notes
+    -----
+    One can use the use_defaults parameter to reset the parameters to the default values.
+    Alternatively, to use the current values - modify the class's contents before instantiating the class.
     """
 
     molecular_search = MolecularFormulaSearchSettings()
@@ -27,27 +60,64 @@ class MSParameters:
     ms_peak = MassSpecPeakSetting()
     data_input = DataInputSetting()
 
+    def __init__(self, use_defaults = False) -> None:
+        if not use_defaults:
+            self.molecular_search = dataclasses.replace(MSParameters.molecular_search)
+            self.transient = dataclasses.replace(MSParameters.transient)
+            self.mass_spectrum = dataclasses.replace(MSParameters.mass_spectrum)
+            self.ms_peak = dataclasses.replace(MSParameters.ms_peak)
+            self.data_input = dataclasses.replace(MSParameters.data_input)
+        else:
+            self.molecular_search = MolecularFormulaSearchSettings()
+            self.transient = TransientSetting()
+            self.mass_spectrum = MassSpectrumSetting()
+            self.ms_peak = MassSpecPeakSetting()
+            self.data_input = DataInputSetting()
+
 class GCMSParameters:
     """GCMSParameters class is used to store the parameters used for the processing of the gas chromatograph mass spectrum
 
     Each attibute is a class that contains the parameters for the processing of the data, see the corems.encapsulation.factory.processingSetting module for more details.
 
+    Parameters
+    ----------
+    use_defaults: bool, optional
+        if True, the class will be instantiated with the default values, otherwise the current values will be used. Default is False.
+
     Attributes
     -----------
     molecular_search: MolecularFormulaSearchSettings
         MolecularFormulaSearchSettings object
     gc_ms: GasChromatographSetting
         GasChromatographSetting object
+
+    Notes
+    -----
+    One can use the use_defaults parameter to reset the parameters to the default values.
+    Alternatively, to use the current values - modify the class's contents before instantiating the class.
     """
 
     molecular_search = CompoundSearchSettings()
     gc_ms = GasChromatographSetting()
 
+    def __init__(self, use_defaults = False) -> None:
+        if not use_defaults:
+            self.molecular_search = dataclasses.replace(GCMSParameters.molecular_search)
+            self.gc_ms = dataclasses.replace(GCMSParameters.gc_ms)
+        else:
+            self.molecular_search = CompoundSearchSettings()
+            self.gc_ms = GasChromatographSetting()
+
 class LCMSParameters:
     """LCMSParameters class is used to store the parameters used for the processing of the liquid chromatograph mass spectrum
 
     Each attibute is a class that contains the parameters for the processing of the data, see the corems.encapsulation.factory.processingSetting module for more details.
 
+    Parameters
+    ----------
+    use_defaults: bool, optional
+        if True, the class will be instantiated with the default values, otherwise the current values will be used. Default is False.
+
     Attributes
     -----------
     lc_ms: LiquidChromatographSetting
@@ -60,17 +130,32 @@ class LCMSParameters:
         MolecularFormulaSearchSettings object
     ms2_molecular_search: MolecularFormulaSearchSettings
         MolecularFormulaSearchSettings object
+
+    Notes
+    -----
+    One can use the use_defaults parameter to reset the parameters to the default values.
+    Alternatively, to use the current values - modify the class's contents before instantiating the class.
     """
     lc_ms = LiquidChromatographSetting()
-    
     mass_spectrum = MassSpectrumSetting()
-
     ms_peak = MassSpecPeakSetting()
-
     ms1_molecular_search = MolecularFormulaSearchSettings()
-    
     ms2_molecular_search = MolecularFormulaSearchSettings()
 
+    def __init__(self, use_defaults = False) -> None:
+        if not use_defaults:
+            self.lc_ms = dataclasses.replace(LCMSParameters.lc_ms)
+            self.mass_spectrum = dataclasses.replace(LCMSParameters.mass_spectrum)
+            self.ms_peak = dataclasses.replace(LCMSParameters.ms_peak)
+            self.ms1_molecular_search = dataclasses.replace(LCMSParameters.ms1_molecular_search)
+            self.ms2_molecular_search = dataclasses.replace(LCMSParameters.ms2_molecular_search)
+        else:
+            self.lc_ms = LiquidChromatographSetting()
+            self.mass_spectrum = MassSpectrumSetting()
+            self.ms_peak = MassSpecPeakSetting()
+            self.ms1_molecular_search = MolecularFormulaSearchSettings()
+            self.ms2_molecular_search = MolecularFormulaSearchSettings()
+
 def default_parameters(file_location):  # pragma: no cover
     """Generate parameters dictionary with the default parameters for data processing
        To gather parameters from instrument data during the data parsing step, a parameters dictionary with the default parameters needs to be generated.

{CoreMS-2.0.9 → CoreMS-2.2.1}/corems/encapsulation/factory/processingSetting.py

@@ -67,6 +67,7 @@ class DataInputSetting:
                                  'Res.': Labels.rp,
                                  'resolution': Labels.rp,
                                  'Intensity': Labels.abundance,
+                                 'Peak Height': Labels.abundance,
                                  'I': Labels.abundance,
                                  'Abundance': Labels.abundance,
                                  'abs_abu': Labels.abundance,
@@ -135,7 +136,7 @@ class LiquidChromatographSetting:
         0-100 % used for extracted ion chromatogram peak detection. Default is 0.01.
     """
    
-    scans: list | tuple = (0, 1)
+    scans: list | tuple = (-1,-1)
     
     eic_tolerance_ppm: float = 5
     
@@ -214,6 +215,7 @@ class MassSpectrumSetting:
         Maximum m/z to use for peak picking. Default is 1200.0.
     picking_point_extrapolate : int, optional
         How many data points (in each direction) to extrapolate the mz axis and 0 pad the abundance axis. Default is 3.
+        Recommend 3 for reduced profile data or if peak picking faults
     calib_minimize_method : str, optional
         Minimization method to use for calibration. Default is 'Powell'.
     calib_pol_order : int, optional
@@ -224,6 +226,10 @@ class MassSpectrumSetting:
         Minimum ppm error to use for calibration. Default is -1.0.
     calib_sn_threshold : float, optional
         Signal to noise threshold to use for calibration. Default is 2.0.
+    calibration_ref_match_method: string, optional
+        Method for matching reference masses with measured masses for recalibration. Default is 'legacy'. 
+    calibration_ref_match_tolerance: float, optional
+        If using the new method for calibration reference mass matching, this tolerance is the initial matching tolerance. Default is 0.003
     do_calibration : bool, optional
         If True, perform calibration. Default is True.    
     """
@@ -252,13 +258,17 @@ class MassSpectrumSetting:
     # How many data points (in each direction) to extrapolate the mz axis and 0 pad the abundance axis
     # This will fix peak picking at spectrum limit issues
     #  0 to keep normal behaviour, typical value 3 to fix
-    picking_point_extrapolate: int = 0 
+    picking_point_extrapolate: int = 3 
 
     calib_minimize_method: str = 'Powell'
     calib_pol_order: int = 2
     max_calib_ppm_error: float = 1.0
     min_calib_ppm_error: float = -1.0
     calib_sn_threshold: float = 2.0
+    calibration_ref_match_method: str = 'legacy'
+    calibration_ref_match_method_implemented: tuple = ('legacy', 'merged')
+    calibration_ref_match_tolerance: float = 0.003
+    calibration_ref_match_std_raw_error_limit: float = 1.5
     #calib_ref_mzs: list = [0]
 
     do_calibration: bool = True
@@ -305,6 +315,9 @@ class MassSpecPeakSetting:
     legacy_resolving_power : bool, optional
         Flag indicating whether to use the legacy (CoreMS v1) resolving power calculation.
         Defaults to True.
+    centroid_legacy_polyfit : bool, optional
+        Use legacy (numpy polyfit) to fit centroid
+        Default false.
     """
 
     kendrick_base: Dict = dataclasses.field(default_factory=dict)
@@ -324,7 +337,10 @@ class MassSpecPeakSetting:
     peak_height_max_percent: float = 10  # 1-100 % used for baseline detection
 
     legacy_resolving_power: bool = True # Use the legacy (CoreMS v1) resolving power calculation (True)
-    
+    # TODO revisit this default
+
+    centroid_legacy_polyfit: bool = False
+
     def __post_init__(self):
 
         # default to CH2