From 560347a47326fc3f73f800faf1098c8e25f0a9bd Mon Sep 17 00:00:00 2001
From: Yasset Perez-Riverol <ypriverol@gmail.com>
Date: Mon, 21 Oct 2024 14:08:55 +0200
Subject: [PATCH] peptide and features finished.

---
 quantmsio/commands/diann_command.py | 11 +++-------
 quantmsio/core/diann.py             | 33 ++++++++++++-----------------
 2 files changed, 17 insertions(+), 27 deletions(-)

diff --git a/quantmsio/commands/diann_command.py b/quantmsio/commands/diann_command.py
index a3847e2..cff64b6 100644
--- a/quantmsio/commands/diann_command.py
+++ b/quantmsio/commands/diann_command.py
@@ -71,12 +71,7 @@ def diann_convert_to_parquet(
     duckdb_threads: The number of threads for the DuckDB engine (e.g 4)
     file_num: The number of files being processed at the same time
     """
-    if (
-        report_path is None
-        or mzml_info_folder is None
-        or output_folder is None
-        or sdrf_path is None
-    ):
+    if report_path is None or mzml_info_folder is None or output_folder is None or sdrf_path is None:
         raise click.UsageError("Please provide all the required parameters")
 
     if not os.path.exists(output_folder):
@@ -97,7 +92,7 @@ def diann_convert_to_parquet(
     dia_nn.write_feature_to_file(
         qvalue_threshold=qvalue_threshold,
         mzml_info_folder=mzml_info_folder,
-        output_path = feature_output_path,
-        file_num = file_num,
+        output_path=feature_output_path,
+        file_num=file_num,
         protein_file=protein_file,
     )
diff --git a/quantmsio/core/diann.py b/quantmsio/core/diann.py
index 513a0bf..abb39d6 100644
--- a/quantmsio/core/diann.py
+++ b/quantmsio/core/diann.py
@@ -19,6 +19,7 @@
 
 MODIFICATION_PATTERN = re.compile(r"\((.*?)\)")
 
+
 def find_modification(peptide):
     """
     Identify the modification site based on the peptide containing modifications.
@@ -143,13 +144,7 @@ def get_peptide_map_from_database(self):
         logging.info("Time to load peptide map {} seconds".format(et))
         return best_ref_map
 
-    def main_report_df(
-        self,
-        qvalue_threshold: float,
-        mzml_info_folder: str,
-        file_num: int,
-        protein_str: str = None
-    ):
+    def main_report_df(self, qvalue_threshold: float, mzml_info_folder: str, file_num: int, protein_str: str = None):
         def intergrate_msg(n):
             nonlocal report
             nonlocal mzml_info_folder
@@ -209,7 +204,7 @@ def intergrate_msg(n):
             report["Modified.Sequence"] = report["Modified.Sequence"].map(modifications_map)
             # pep
             report["scan_reference_file_name"] = report["Precursor.Id"].map(best_ref_map)
-            #report["scan"] = None
+            # report["scan"] = None
             report.rename(columns=DIANN_MAP, inplace=True)
             # add extra msg
             report = self.add_additional_msg(report)
@@ -235,21 +230,24 @@ def add_additional_msg(self, report: pd.DataFrame) -> pd.DataFrame:
         )
         report["scan"] = report["scan"].apply(generate_scan_number)
         report.loc[:, "gg_names"] = report["gg_names"].str.split(",")
-        report.loc[:, "additional_intensities"] = report["Precursor.Normalised"].apply(lambda v: [{"name": "normalized intensity", "value": np.float32(v)}])
-        report.loc[:, "additional_scores"] = report[["Q.Value","PG.Q.Value"]].apply(lambda row: [{"name": "qvalue", "value": row["Q.Value"]}, {"name": "pg_qvalue", "value": row["PG.Q.Value"]}],axis=1)
+        report.loc[:, "additional_intensities"] = report["Precursor.Normalised"].apply(
+            lambda v: [{"name": "normalized intensity", "value": np.float32(v)}]
+        )
+        report.loc[:, "additional_scores"] = report[["Q.Value", "PG.Q.Value"]].apply(
+            lambda row: [
+                {"name": "qvalue", "value": row["Q.Value"]},
+                {"name": "pg_qvalue", "value": row["PG.Q.Value"]},
+            ],
+            axis=1,
+        )
         report.loc[:, "modification_details"] = None
         report.loc[:, "cv_params"] = None
         report.loc[:, "gg_accessions"] = None
         report.loc[:, "best_id_score"] = None
         return report
 
-
     def generate_feature(
-        self,
-        qvalue_threshold: float,
-        mzml_info_folder: str,
-        file_num: int = 50,
-        protein_str: str = None
+        self, qvalue_threshold: float, mzml_info_folder: str, file_num: int = 50, protein_str: str = None
     ):
         for report in self.main_report_df(qvalue_threshold, mzml_info_folder, file_num, protein_str):
             s = time.time()
@@ -260,7 +258,6 @@ def generate_feature(
             logging.info("Time to generate psm and feature file {} seconds".format(et))
             yield feature
 
-
     def write_feature_to_file(
         self,
         qvalue_threshold: float,
@@ -308,5 +305,3 @@ def merge_sdrf_to_feature(self, report):
             inplace=True,
         )
         return report
-
-