Added minimum ion score feature

pycamv/fragment/validate.py

@@ -287,6 +287,7 @@ def validate_spectra(
     raw_paths=None,
     scans_path=None,
     scan_list=None,
+    score=None,
     mat_sessions=None,
     out_path=None,
     cpu_count=None,
@@ -302,6 +303,7 @@ def validate_spectra(
     raw_paths : list of str
     scans_path : str, optional
     scan_list : list of int, optional
+    score : int, optional
     mat_sessions : list of str, optional
     out_path : str, optional
     cpu_count : int, optional
@@ -332,6 +334,7 @@ def validate_spectra(
     fixed_mods, var_mods, pep_queries = search.read_search_file(
         search_path,
         scan_list=scan_list,
+        score=score,
     )
 
     LOGGER.info("Found info for {} peptide queries".format(len(pep_queries)))

pycamv/main.py

@@ -81,6 +81,11 @@ def _parse_args(args):
         nargs="*", type=int,
         help="Individual scans to select for validation.",
     )
+    parser.add_argument(
+        "--score",
+        type=int,
+        help="Minimum Ion Score to include for validation.",
+    )
     parser.add_argument(
         "--mat-sessions",
         nargs="+",
@@ -189,6 +194,7 @@ def main(args):
         raw_paths=args.raw_paths,
         scans_path=args.scans_path,
         scan_list=args.scans,
+        score=args.score,
         mat_sessions=args.mat_sessions,
         out_path=args.out_path,
         reprocess=args.reprocess,

pycamv/search/discoverer.py

@@ -547,6 +547,7 @@ def _get_peptide_queries(
     var_mods,
     pd_version,
     scan_list=None,
+    score=None,
 ):
     out = []
     # index = 0
@@ -574,6 +575,9 @@ def _get_peptide_queries(
         if scan_list and scan not in scan_list:
             continue
 
+        if score and pep_score < score:
+            continue
+
         if exp_z < 1:
             LOGGER.warning(
                 "Charge: {}, {} for {} (Scan {})"
@@ -774,14 +778,16 @@ def _get_pd_version(conn):
     return tuple([int(i) for i in next(query)[0].split('.')])
 
 
-def read_discoverer_msf(msf_path, scan_list=None):
+def read_discoverer_msf(msf_path, scan_list=None, score=None):
     """
     Parse a ProteomeDiscoverer MSF file.
 
     Parameters
     ----------
     msf_path : str
         Path to MSF file.
+    scan_list : list of int, optional
+    score : int, optional
 
     Returns
     -------
@@ -794,6 +800,10 @@ def read_discoverer_msf(msf_path, scan_list=None):
             os.path.basename(msf_path),
         )
     )
+
+    if score:
+        LOGGER.info("Using minimum ion score >= {}".format(score))
+
     start = datetime.now()
 
     with sqlite3.connect(msf_path) as conn:
@@ -808,6 +818,7 @@ def read_discoverer_msf(msf_path, scan_list=None):
             var_mods,
             pd_version=pd_version,
             scan_list=scan_list,
+            score=score,
         )
 
     LOGGER.info(

pycamv/search/search.py

@@ -16,14 +16,16 @@
 }
 
 
-def read_search_file(path, scan_list=None):
+def read_search_file(path, scan_list=None, score=None):
     """
     Parse a search input file.
 
     Parameters
     ----------
     path : str
         Path to search input file.
+    scan_list : list of int, optional
+    score : int, optional
 
     Returns
     -------
@@ -39,4 +41,4 @@ def read_search_file(path, scan_list=None):
 
     LOGGER.debug("Using {} backend for {}".format(backend.__name__, ext))
 
-    return backend(path, scan_list=scan_list)
+    return backend(path, scan_list=scan_list, score=score)