update: query

quantmsio/commands/absolute_expression_command.py

@@ -22,9 +22,15 @@
     help="quantms.io project file",
     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)
+@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_ibaq_absolute(
     ibaq_file: str,
     sdrf_file: str,

quantmsio/commands/attach_file_command.py

@@ -8,7 +8,9 @@
     short_help="Register the file to project.json.",
 )
 @click.option("--project_file", help="the project.json file", required=True)
-@click.option("--attach_file", help="The path of the file that will be registered", required=True)
+@click.option(
+    "--attach_file", help="The path of the file that will be registered", required=True
+)
 @click.option(
     "--category",
     type=click.Choice(

quantmsio/commands/diann_convert_command.py

@@ -20,7 +20,9 @@
     help="the design file path",
     required=True,
 )
-@click.option("--qvalue_threshold", help="qvalue_threshold", required=True, default=0.05)
+@click.option(
+    "--qvalue_threshold", help="qvalue_threshold", required=True, default=0.05
+)
 @click.option(
     "--mzml_info_folder",
     help="the foldef of mzml_info tsv file",
@@ -45,7 +47,9 @@
     "--duckdb_max_memory",
     help="The maximum amount of memory allocated by the DuckDB engine (e.g 4GB)",
 )
-@click.option("--duckdb_threads", help="The number of threads for the DuckDB engine (e.g 4)")
+@click.option(
+    "--duckdb_threads", help="The number of threads for the DuckDB engine (e.g 4)"
+)
 @click.option(
     "--file_num",
     help="The number of files being processed at the same time",
@@ -81,8 +85,14 @@ def diann_convert_to_parquet(
     if not output_prefix_file:
         output_prefix_file = ""
 
-    feature_output_path = output_folder + "/" + create_uuid_filename(output_prefix_file, ".feature.parquet")
-    psm_output_path = output_folder + "/" + create_uuid_filename(output_prefix_file, ".psm.parquet")
+    feature_output_path = (
+        output_folder
+        + "/"
+        + create_uuid_filename(output_prefix_file, ".feature.parquet")
+    )
+    psm_output_path = (
+        output_folder + "/" + create_uuid_filename(output_prefix_file, ".psm.parquet")
+    )
 
     dia_nn = DiaNNConvert()
 

quantmsio/commands/differential_expression_command.py

@@ -24,9 +24,15 @@
     required=False,
     default="0.05",
 )
-@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)
+@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,

quantmsio/commands/feature_command.py

@@ -62,15 +62,24 @@ def convert_feature_file(
     :return: none
     """
 
-    if sdrf_file is None or msstats_file is None or mztab_file is None or output_folder is None:
+    if (
+        sdrf_file is None
+        or msstats_file is None
+        or mztab_file is None
+        or output_folder is None
+    ):
         raise click.UsageError("Please provide all the required parameters")
     if use_cache is None:
         use_cache = False
 
     feature_manager = FeatureHandler()
     if not output_prefix_file:
         output_prefix_file = ""
-    feature_manager.parquet_path = output_folder + "/" + create_uuid_filename(output_prefix_file, ".feature.parquet")
+    feature_manager.parquet_path = (
+        output_folder
+        + "/"
+        + create_uuid_filename(output_prefix_file, ".feature.parquet")
+    )
     if consensusxml_file is not None:
         feature_manager.convert_mztab_msstats_to_feature(
             mztab_file=mztab_file,

quantmsio/commands/generate_gene_msg_command.py

@@ -14,7 +14,9 @@
 )
 @click.option(
     "--map_parameter",
-    type=click.Choice(["map_protein_name", "map_protein_accession"], case_sensitive=False),
+    type=click.Choice(
+        ["map_protein_name", "map_protein_accession"], case_sensitive=False
+    ),
     help="map type",
 )
 @click.option("--species", help="species", default="human")
@@ -38,4 +40,6 @@ def map_gene_msg_to_parquet(
     if not output_path.endswith("parquet"):
         raise click.UsageError("Please provide file extension(.parquet)")
 
-    map_gene_msgs_to_parquet(parquet_path, fasta_path, map_parameter, output_path, label, species)
+    map_gene_msgs_to_parquet(
+        parquet_path, fasta_path, map_parameter, output_path, label, species
+    )

quantmsio/commands/generate_project_report_command.py

@@ -7,7 +7,9 @@
     "generate-project-report",
     short_help="generate report of project " "format",
 )
-@click.option("--project_folder", help="Folder to generate the df expression file.", required=True)
+@click.option(
+    "--project_folder", help="Folder to generate the df expression file.", required=True
+)
 def generate_report_about_project(project_folder):
     """
     project_folder: The folder path for the full project.

quantmsio/commands/generate_spectra_message_command.py

@@ -46,4 +46,6 @@ def map_spectrum_message_to_parquet(
     """
     if not output_path.endswith("parquet"):
         raise click.UsageError("Please provide file extension(.parquet)")
-    generate_features_of_spectrum(parquet_path, mzml_directory, output_path, label, file_num, partition)
+    generate_features_of_spectrum(
+        parquet_path, mzml_directory, output_path, label, file_num, partition
+    )

quantmsio/commands/generate_start_and_end_command.py

@@ -15,7 +15,9 @@
     help="parquet type",
 )
 @click.option("--output_path", help="save path", required=True)
-def inject_start_and_end_from_fasta(parquet_path: str, fasta_path: str, label: str, output_path: str):
+def inject_start_and_end_from_fasta(
+    parquet_path: str, fasta_path: str, label: str, output_path: str
+):
     """
     Register the file with project.json
     :param parquet_path: psm or feature parquet file path

quantmsio/commands/get_unanimous_command.py

@@ -22,7 +22,9 @@ def labels():
 @click.option("--output_path", help="output file path")
 @click.option(
     "--map_parameter",
-    type=click.Choice(["map_protein_name", "map_protein_accession"], case_sensitive=False),
+    type=click.Choice(
+        ["map_protein_name", "map_protein_accession"], case_sensitive=False
+    ),
     help="map type",
 )
 @click.option(
@@ -53,7 +55,9 @@ def get_unanimous_for_parquet(parquet_path, fasta, output_path, map_parameter, l
 @click.option("--output_path", help="output file path")
 @click.option(
     "--map_parameter",
-    type=click.Choice(["map_protein_name", "map_protein_accession"], case_sensitive=False),
+    type=click.Choice(
+        ["map_protein_name", "map_protein_accession"], case_sensitive=False
+    ),
     help="map type",
 )
 def get_unanimous_for_tsv(path, fasta, output_path, map_parameter):

quantmsio/commands/load_best_scan_number_command.py

@@ -8,9 +8,13 @@
     short_help="inject bset_psm_scan_number to feature",
 )
 @click.option("--diann_psm_path", help="diann psm parquet file path", required=True)
-@click.option("--diann_feature_path", help="diann feature parquet file path", required=True)
+@click.option(
+    "--diann_feature_path", help="diann feature parquet file path", required=True
+)
 @click.option("--output_path", help="save path", required=True)
-def inject_bset_psm_scan_number(diann_psm_path: str, diann_feature_path: str, output_path: str):
+def inject_bset_psm_scan_number(
+    diann_psm_path: str, diann_feature_path: str, output_path: str
+):
     """
     Register the file with project.json
     :param diann_psm_path: diann psm parquet file path

quantmsio/commands/plot.py

@@ -32,15 +32,21 @@ def plot_peptides(ctx, psm_parquet_path: str, sdrf_path: str, save_path: str):
     :param save_path: img save path [xxx.png]
     :return: none
     """
-    plot_peptides_of_lfq_condition(psm_parquet_path=psm_parquet_path, sdrf_path=sdrf_path, save_path=save_path)
+    plot_peptides_of_lfq_condition(
+        psm_parquet_path=psm_parquet_path, sdrf_path=sdrf_path, save_path=save_path
+    )
 
 
-@plot.command("plot-ibaq-distribution", short_help="plot ibaq distribution of expression")
+@plot.command(
+    "plot-ibaq-distribution", short_help="plot ibaq distribution of expression"
+)
 @click.option("--ibaq_path", help="ibaq file path", required=True)
 @click.option("--save_path", help="img save path [xxx.svg]", required=True)
 @click.option("--select_column", help="Selected column in Ibaq File", required=False)
 @click.pass_context
-def plot_ibaq_distribution(ctx, ibaq_path: str, save_path: str, select_column: str) -> None:
+def plot_ibaq_distribution(
+    ctx, ibaq_path: str, save_path: str, select_column: str
+) -> None:
     """
     plot ibaq distribution of expression
     :param ibaq_path: ibaq file path
@@ -59,7 +65,9 @@ def plot_ibaq_distribution(ctx, ibaq_path: str, save_path: str, select_column: s
 @click.option("--save_path", help="img save path [xxx.svg]", required=True)
 @click.option("--num_samples", help="The number of samples plotted", default=10)
 @click.pass_context
-def plot_kde_intensity_distribution(feature_path: str, save_path: str, num_samples: int):
+def plot_kde_intensity_distribution(
+    feature_path: str, save_path: str, num_samples: int
+):
     """
     plot ibaq distribution of expression
     :param feature_path: feature file path
@@ -99,7 +107,9 @@ def plot_bar_peptide_distribution(feature_path: str, save_path: str, num_samples
 @click.option("--save_path", help="img save path [xxx.svg]", required=True)
 @click.option("--num_samples", help="The number of samples plotted", default=10)
 @click.pass_context
-def plot_box_intensity_distribution(feature_path: str, save_path: str, num_samples: int):
+def plot_box_intensity_distribution(
+    feature_path: str, save_path: str, num_samples: int
+):
     """
     plot ibaq distribution of expression
     :param feature_path: feature file path

quantmsio/commands/psm_command.py

@@ -31,7 +31,9 @@
     help="Prefix of the parquet file needed to generate the file name",
     required=False,
 )
-@click.option("--verbose", help="Output debug information.", default=False, is_flag=True)
+@click.option(
+    "--verbose", help="Output debug information.", default=False, is_flag=True
+)
 def convert_psm_file(
     mztab_file: str,
     output_folder: str,
@@ -55,7 +57,9 @@ def convert_psm_file(
         output_prefix_file = ""
 
     psm_manager = PSMHandler()
-    psm_manager.parquet_path = output_folder + "/" + create_uuid_filename(output_prefix_file, ".psm.parquet")
+    psm_manager.parquet_path = (
+        output_folder + "/" + create_uuid_filename(output_prefix_file, ".psm.parquet")
+    )
     psm_manager.convert_mztab_to_psm(
         mztab_path=mztab_file,
         parquet_path=psm_manager.parquet_path,
@@ -65,7 +69,9 @@ def convert_psm_file(
 
 
 @click.command("compare-set-psms", short_help="plot venn for a set of Psms parquet")
-@click.option("-p", "--parquets", type=str, help="List of psm parquet path", multiple=True)
+@click.option(
+    "-p", "--parquets", type=str, help="List of psm parquet path", multiple=True
+)
 @click.option("-t", "--tags", type=str, help="List of parquet label", multiple=True)
 def compare_set_of_psms(parquets, tags):
     """
@@ -74,7 +80,9 @@ def compare_set_of_psms(parquets, tags):
     :param tags: a set of psm label
     """
     if len(parquets) != len(tags):
-        raise click.UsageError("Please provide same length of parquet_list and label_list")
+        raise click.UsageError(
+            "Please provide same length of parquet_list and label_list"
+        )
 
     plot_peptidoform_charge_venn(parquets, tags)
     plot_sequence_venn(parquets, tags)

quantmsio/commands/statistics.py

@@ -23,7 +23,8 @@ def statistics():
 @click.option("--parquet_path", help="psm parquet path in lfq", required=True)
 @click.option(
     "--save_path",
-    help="file with the statistics (e.g. statistics.csv), if not provided," " will print to stdout",
+    help="file with the statistics (e.g. statistics.csv), if not provided,"
+    " will print to stdout",
 )
 @click.pass_context
 def feature_file_statistics(ctx, absolute_path: str, parquet_path: str, save_path: str):
@@ -40,11 +41,15 @@ def write_stats(file, stats: ParquetStatistics):
         file.write("Number of proteins: {}\n".format(stats.get_number_of_proteins()))
         file.write("Number of peptides: {}\n".format(stats.get_number_of_peptides()))
         file.write("Number of samples: {}\n".format(stats.get_number_of_samples()))
-        file.write("Number of peptidoforms: {}\n".format(stats.get_number_of_peptidoforms()))
+        file.write(
+            "Number of peptidoforms: {}\n".format(stats.get_number_of_peptidoforms())
+        )
         file.write("Number of msruns: {}\n".format(stats.get_number_msruns()))
 
     def write_absolute_stats(file, stats: IbaqStatistics):
-        file.write("Ibaq Number of proteins: {}\n".format(stats.get_number_of_proteins()))
+        file.write(
+            "Ibaq Number of proteins: {}\n".format(stats.get_number_of_proteins())
+        )
         file.write("Ibaq Number of samples: {}\n".format(stats.get_number_of_samples()))
 
     if save_path:
@@ -65,7 +70,8 @@ def write_absolute_stats(file, stats: IbaqStatistics):
 @click.option("--parquet_path", help="psm parquet path in lfq", required=True)
 @click.option(
     "--save_path",
-    help="file with the statistics (e.g. statistics.csv), if not provided," " will print to stdout",
+    help="file with the statistics (e.g. statistics.csv), if not provided,"
+    " will print to stdout",
 )
 @click.pass_context
 def parquet_psm_statistics(ctx, parquet_path: str, save_path: str):
@@ -79,7 +85,9 @@ def parquet_psm_statistics(ctx, parquet_path: str, save_path: str):
     def write_stats(file, stats: ParquetStatistics):
         file.write("Number of proteins: {}\n".format(stats.get_number_of_proteins()))
         file.write("Number of peptides: {}\n".format(stats.get_number_of_peptides()))
-        file.write("Number of peptidoforms: {}\n".format(stats.get_number_of_peptidoforms()))
+        file.write(
+            "Number of peptidoforms: {}\n".format(stats.get_number_of_peptidoforms())
+        )
         file.write("Number of psms: {}\n".format(stats.get_number_of_psms()))
         file.write("Number of msruns: {}\n".format(stats.get_number_msruns()))
 

quantmsio/core/ae.py

@@ -85,13 +85,25 @@ def convert_ibaq_to_quantms(
         output_lines = ""
         if self.project_manager:
             output_lines += (
-                "#project_accession: " + self.project_manager.project.project_info["project_accession"] + "\n"
+                "#project_accession: "
+                + self.project_manager.project.project_info["project_accession"]
+                + "\n"
             )
-            output_lines += "#project_title: " + self.project_manager.project.project_info["project_title"] + "\n"
             output_lines += (
-                "#project_description: " + self.project_manager.project.project_info["project_description"] + "\n"
+                "#project_title: "
+                + self.project_manager.project.project_info["project_title"]
+                + "\n"
+            )
+            output_lines += (
+                "#project_description: "
+                + self.project_manager.project.project_info["project_description"]
+                + "\n"
+            )
+            output_lines += (
+                "#quantms_version: "
+                + self.project_manager.project.project_info["quantms_version"]
+                + "\n"
             )
-            output_lines += "#quantms_version: " + self.project_manager.project.project_info["quantms_version"] + "\n"
         factor_value = self.get_factor_value()
         if factor_value is not None:
             output_lines += "#factor_value: " + factor_value + "\n"
@@ -127,8 +139,12 @@ def convert_ibaq_to_quantms(
             f.write(output_lines)
 
         if self.project_manager:
-            self.project_manager.add_quantms_file(file_category="absolute_file", file_name=output_filename)
-        logger.info(f"Absolute expression file copied to {output_filename} and added to the project information")
+            self.project_manager.add_quantms_file(
+                file_category="absolute_file", file_name=output_filename
+            )
+        logger.info(
+            f"Absolute expression file copied to {output_filename} and added to the project information"
+        )
 
     def get_factor_value(self):
         """

quantmsio/core/core.py

@@ -18,7 +18,9 @@ def __init__(self, name_prefix: str):
         # Create a cache name using a hash and uuid
         if name_prefix is None:
             name_prefix = "generic"
-        self._cache_name = str("_cache_name_{}_{}".format(name_prefix, uuid.uuid4().hex))
+        self._cache_name = str(
+            "_cache_name_{}_{}".format(name_prefix, uuid.uuid4().hex)
+        )
         self.cache = diskcache.Cache(self._cache_name, statistics=True)
         self.cache.create_tag_index()