Merge pull request #53 from MannLabs/development

preparation for v1.4.0

.github/workflows/pip_installation.yml

@@ -1,6 +1,4 @@
 on:
-  push:
-    branches: [ main, development ]
   pull_request:
     branches: [ main, development ]
   workflow_dispatch:

.gitignore

@@ -137,6 +137,9 @@ dmypy.json
 # Pyre type checker
 .pyre/
 
+# Data
+testdata/
+
 ######################
 # OS generated files #
 ######################
@@ -156,4 +159,4 @@ gui/out/
 ###################
 # mono crash logs #
 ###################
-mono_crash*
+mono_crash*

alphadia/__init__.py

@@ -2,7 +2,7 @@
 
 
 __project__ = "alphadia"
-__version__ = "1.3.2"
+__version__ = "1.4.0"
 __license__ = "Apache"
 __description__ = "An open-source Python package of the AlphaPept ecosystem"
 __author__ = "Mann Labs"
@@ -13,7 +13,7 @@
     "software",
     "AlphaPept ecosystem",
 ]
-__python_version__ = ">=3.8,<3.10"
+__python_version__ = ">=3.8,<=3.12"
 __classifiers__ = [
     "Development Status :: 1 - Planning",
     # "Development Status :: 2 - Pre-Alpha",

alphadia/cli.py

@@ -4,10 +4,12 @@
 import logging
 import time
 import yaml
+import os
 
 # alphadia imports
 import alphadia
 from alphadia.workflow import reporting
+from alphadia import utils
 
 # alpha family imports
 
@@ -22,9 +24,8 @@
 )
 
 @click.pass_context
-@click.version_option(alphadia.__version__, "-v", "--version")
+@click.version_option(alphadia.__version__, "-v", "--version", message="%(version)s")
 def run(ctx, **kwargs):
-
     if ctx.invoked_subcommand is None:
         click.echo(run.get_help(ctx))
 
@@ -38,7 +39,7 @@ def gui():
     help="Extract DIA precursors from a list of raw files using a spectral library."
 )
 @click.argument(
-    "output-location",
+    "output-directory",
     type=click.Path(exists=True, file_okay=False, dir_okay=True),
     required=False,
 )
@@ -49,12 +50,26 @@ def gui():
     multiple=True,
     type=click.Path(exists=True, file_okay=True, dir_okay=True),
 )
+@click.option(
+    '--directory',
+    '-d',
+    help="Directory containing raw data input files.",
+    type=click.Path(exists=True, file_okay=False, dir_okay=True),
+)
 @click.option(
     '--library',
     '-l',
     help="Spectral library in AlphaBase hdf5 format.",
     type=click.Path(exists=True, file_okay=True, dir_okay=False),
 )
+@click.option(
+    '--wsl',
+    '-w',
+    help="Run alphadia using WSL. Windows paths will be converted to WSL paths.",
+    type=bool,
+    default=False,
+    is_flag=True,
+)
 @click.option(
     "--fdr",
     help='False discovery rate for the final output.',
@@ -112,38 +127,59 @@ def extract(**kwargs):
         with open(kwargs['config'], 'r') as f:
             config_update = yaml.safe_load(f)
 
-    output_location = None
-    if kwargs['output_location'] is not None:
-        output_location = kwargs['output_location']
+    # update output directory based on config file
+    output_directory = None
+    if kwargs['output_directory'] is not None:
+        if kwargs['wsl']:
+            kwargs['output_directory'] = utils.windows_to_wsl(kwargs['output_directory'])
+        output_directory = kwargs['output_directory']
 
-    if "output" in config_update:
-        output_location = config_update['output']
+    if "output_directory" in config_update:
+        if kwargs['wsl']:
+            config_update['output_directory'] = utils.windows_to_wsl(config_update['output_directory'])
+        output_directory = config_update['output_directory']
 
-    if output_location is None:
-        logging.error("No output location specified.")
+
+    if output_directory is None:
+        logging.error("No output directory specified.")
         return
 
-    reporting.init_logging(kwargs['output_location'])
+    reporting.init_logging(output_directory)
     logger = logging.getLogger()
-
+    
     # assert input files have been specified
-    files = None
+    files = []
     if kwargs['file'] is not None:
         files = list(kwargs['file'])
+        if kwargs['wsl']:
+            files = [utils.windows_to_wsl(f) for f in files]
+
+    # load whole directory if specified
+    if kwargs['directory'] is not None:
+        if kwargs['wsl']:
+            kwargs['directory'] = utils.windows_to_wsl(kwargs['directory'])
+        files += [os.path.join(kwargs['directory'], f) for f in os.listdir(kwargs['directory'])]
 
-    if "files" in config_update:
-        files = config_update['files'] if type(config_update['files']) is list else [config_update['files']]
+    # load list of raw files from config file
+    if "raw_file_list" in config_update:
+        if kwargs['wsl']:
+            config_update['raw_file_list'] = [utils.windows_to_wsl(f) for f in config_update['raw_file_list']]
+        files += config_update['raw_file_list'] if type(config_update['raw_file_list']) is list else [config_update['raw_file_list']]
 
     if (files is None) or (len(files) == 0):
-        logging.error("No files specified.")
+        logging.error("No raw files specified.")
         return
 
     # assert library has been specified
     library = None
     if kwargs['library'] is not None:
+        if kwargs['wsl']:
+            kwargs['library'] = utils.windows_to_wsl(kwargs['library'])
         library = kwargs['library']
 
     if "library" in config_update:
+        if kwargs['wsl']:
+            config_update['library'] = utils.windows_to_wsl(config_update['library'])
         library = config_update['library']
 
     if library is None:
@@ -154,27 +190,25 @@ def extract(**kwargs):
     for f in files:
         logger.progress(f"  {f}")
     logger.progress(f"Using library {library}.")
-    logger.progress(f"Saving output to {output_location}.")
-
+
+
+    if kwargs['wsl']:
+        config_update['general']['wsl'] = True
+
+    logger.progress(f"Saving output to {output_directory}.")
+
     try:
 
         import matplotlib
         # important to supress matplotlib output
         matplotlib.use('Agg')
 
-        from alphabase.spectral_library.base import SpecLibBase
         from alphadia.planning import Plan
 
-        lib = SpecLibBase()
-        lib.load_hdf(library, load_mod_seq=True)
-        #lib._precursor_df['elution_group_idx'] = lib._precursor_df['precursor_idx']
-
-        #config_update = eval(kwargs['config_update']) if kwargs['config_update'] else None
-
         plan = Plan(
-            output_location,
+            output_directory,
             files,
-            lib,
+            library,
             config_update = config_update
             )
 
@@ -185,4 +219,4 @@ def extract(**kwargs):
         )
 
     except Exception as e:
-        logging.exception(e)
+        logger.error(e)

alphadia/data/thermo.py

@@ -1,6 +1,8 @@
 # native imports
 import math
 import os
+import logging
+logger = logging.getLogger()
 
 # alphadia imports
 from alphadia import utils
@@ -114,6 +116,7 @@ def __init__(
             self, 
             path, 
             astral_ms1=False,
+            cv=None,
             **kwargs
             ):
             super().__init__(**kwargs)
@@ -122,6 +125,7 @@ def __init__(
             self.sample_name = os.path.basename(self.raw_file_path)
 
             self.astral_ms1 = astral_ms1
+            self.cv = cv
             self.filter_spectra()
 
             self.cycle = calculate_cycle(self.spectrum_df)
@@ -145,16 +149,29 @@ def __init__(
             self.frame_max_index = len(self.rt_values)-1
 
         def filter_spectra(self):
+
+            print(self.cv, 'cv' in self.spectrum_df.columns)
+
+            # filter for astral MS1
             if self.astral_ms1:
                 self.spectrum_df = self.spectrum_df[self.spectrum_df['nce'] > 0.1]
                 self.spectrum_df.loc[self.spectrum_df['nce'] < 1.1, 'ms_level'] = 1
                 self.spectrum_df.loc[self.spectrum_df['nce'] < 1.1, 'precursor_mz'] = -1.0
                 self.spectrum_df.loc[self.spectrum_df['nce'] < 1.1, 'isolation_lower_mz'] = -1.0
                 self.spectrum_df.loc[self.spectrum_df['nce'] < 1.1, 'isolation_upper_mz'] = -1.0
-                self.spectrum_df['spec_idx'] = np.arange(len(self.spectrum_df))
             else:
                 self.spectrum_df = self.spectrum_df[(self.spectrum_df['nce'] < 0.1) | (self.spectrum_df['nce'] > 1.1)]
-                self.spectrum_df['spec_idx'] = np.arange(len(self.spectrum_df))
+
+            # filter for cv
+            if self.cv is not None:
+                if 'cv' in self.spectrum_df.columns:
+                    # use np.isclose to account for floating point errors
+                    logger.info(f"Filtering for CV {self.cv}")
+                    logger.info(f"Before: {len(self.spectrum_df)}")
+                    self.spectrum_df = self.spectrum_df[np.isclose(self.spectrum_df['cv'], self.cv, atol=0.1)]
+                    logger.info(f"After: {len(self.spectrum_df)}")
+
+            self.spectrum_df['spec_idx'] = np.arange(len(self.spectrum_df))
 
         def jitclass(self):
             return ThermoJIT(