update: ae&de

quantmsio/commands/absolute_expression_command.py

@@ -1,7 +1,7 @@
 import click
 
 from quantmsio.core.ae import AbsoluteExpressionHander
-
+from quantmsio.utils.file_utils import extract_protein_list
 
 @click.command(
     "convert-ae",
@@ -17,6 +17,11 @@
     help="the sdrf_file path",
     required=True,
 )
+@click.option(
+    "--protein_file",
+    help="Protein file that meets specific requirements",
+    required=False,
+)
 @click.option(
     "--project_file",
     help="quantms.io project file",
@@ -29,6 +34,7 @@ def convert_ibaq_absolute(
     ibaq_file: str,
     sdrf_file: str,
     project_file: str,
+    protein_file:str,
     output_folder: str,
     output_prefix_file: str,
     delete_existing: bool = True,
@@ -40,14 +46,17 @@ def convert_ibaq_absolute(
     :param sdrf_file: sdrf file
     :param project_file: quantms.io project file
     :param output_folder: Folder to generate the df expression file.
+    :protein_file: Filtered protein file.
     :param output_prefix_file: Prefix of the df expression file
     :param delete_existing: Delete existing files in the output folder
     :return: none
     """
+    protein_list = extract_protein_list(protein_file) if protein_file else None
+    protein_str = "|".join(protein_list) if protein_list else None
     ae_handler = AbsoluteExpressionHander()
     if project_file:
         ae_handler.load_project_file(project_file)
-    ae_handler.load_ibaq_file(ibaq_file)
+    ae_handler.load_ibaq_file(ibaq_file,protein_str)
     ae_handler.load_sdrf_file(sdrf_file)
     ae_handler.convert_ibaq_to_quantms(
         output_folder=output_folder,

quantmsio/commands/differential_expression_command.py

@@ -1,7 +1,7 @@
 import click
 
 from quantmsio.core.de import DifferentialExpressionHandler
-
+from quantmsio.utils.file_utils import extract_protein_list
 
 @click.command(
     "convert-de",
@@ -24,13 +24,19 @@
     required=False,
     default="0.05",
 )
+@click.option(
+    "--protein_file",
+    help="Protein file that meets specific requirements",
+    required=False,
+)
 @click.option("--output_folder", help="Folder to generate the df expression file.", required=True)
 @click.option("--output_prefix_file", help="Prefix of the df expression file", required=False)
 @click.option("--delete_existing", help="Delete existing files in the output folder", is_flag=True)
 def convert_msstats_differential(
     msstats_file: str,
     sdrf_file: str,
     project_file: str,
+    protein_file:str,
     fdr_threshold: float,
     output_folder: str,
     output_prefix_file: str,
@@ -51,11 +57,12 @@ def convert_msstats_differential(
 
     if msstats_file is None or sdrf_file is None or output_folder is None:
         raise click.UsageError("Please provide all the required parameters")
-
+    protein_list = extract_protein_list(protein_file) if protein_file else None
+    protein_str = "|".join(protein_list) if protein_list else None
     de_handler = DifferentialExpressionHandler()
     if project_file:
         de_handler.load_project_file(project_file)
-    de_handler.load_msstats_file(msstats_file)
+    de_handler.load_msstats_file(msstats_file,protein_str)
     de_handler.load_sdrf_file(sdrf_file)
     de_handler.set_fdr_threshold(fdr_threshold=fdr_threshold)
     de_handler.convert_msstats_to_quantms(

quantmsio/commands/feature_command.py

@@ -33,7 +33,7 @@
 )
 @click.option(
     "--protein_file",
-    help="the mzTab file, this will be used to extract the protein information",
+    help="Protein file that meets specific requirements",
     required=False,
 )
 @click.option(

quantmsio/commands/psm_command.py

@@ -28,7 +28,7 @@
 )
 @click.option(
     "--protein_file",
-    help="the mzTab file, this will be used to extract the protein information",
+    help="Protein file that meets specific requirements",
     required=False,
 )
 @click.option(

quantmsio/core/ae.py

@@ -57,14 +57,16 @@ def load_project_file(self, project_file: str):
         self.project_manager = ProjectHandler()
         self.project_manager.load_project_info(project_file)
 
-    def load_ibaq_file(self, path):
+    def load_ibaq_file(self, path,protein_str=None):
         usecols = ["ProteinName", "SampleID", "Condition", "Ibaq", "IbaqLog"]
         ibaq_columns = get_ibaq_columns(path)
         for col in usecols:
             if col not in ibaq_columns:
                 raise Exception(f"Not found {col} in ibaq file")
         ibaqs = pd.read_csv(path, usecols=usecols)
         ibaqs.rename(columns=AbsoluteExpressionHander.LABEL_MAP, inplace=True)
+        if protein_str:
+            ibaqs = ibaqs[ibaqs["protein"].str.contains(f"{protein_str}",na=False)]
         self.ae_file_path = path
         self.ibaq_df = ibaqs
 

quantmsio/core/de.py

@@ -76,7 +76,7 @@ def __init__(self):
         self.msstats_df = None
         self.de_file_path = None
 
-    def load_msstats_file(self, msstats_file_path: str):
+    def load_msstats_file(self, msstats_file_path: str,protein_str:str=None):
         """
         Load a MSstats differential file
         :param msstats_file_path: MSstats differential file path
@@ -90,6 +90,8 @@ def load_msstats_file(self, msstats_file_path: str):
         self.msstats_df = pd.read_csv(msstats_file_path, sep="\t")
         # Rename columns to a lower case
         self.msstats_df.columns = self.msstats_df.columns.str.lower()
+        if protein_str:
+            self.msstats_df = self.msstats_df[self.msstats_df["protein"].str.contains(f"{protein_str}",na=False)]
 
     def load_project_file(self, project_file: str):
         """