Merge pull request #15 from hgb-bin-proteomics/develop

v1.2.2

create_spectral_library.py

@@ -6,8 +6,8 @@
 # [email protected]
 
 # version tracking
-__version = "1.2.1"
-__date = "2024-07-18"
+__version = "1.2.2"
+__date = "2024-07-21"
 
 # REQUIREMENTS
 # pip install pandas
@@ -310,6 +310,7 @@ def get_positions_in_protein(row: pd.Series) -> Dict[str, int]:
     return {"A": pep_pos_A + xl_pos_A, "B": pep_pos_B + xl_pos_B}
 
 ##### DECOY GENERATION #####
+# decoy generation implemented as described by Zhang et al. here: https://doi.org/10.1021/acs.jproteome.7b00614
 
 def generate_decoy_csm(row: pd.Series, crosslinker: str = CROSSLINKER) -> pd.Series:
     """
@@ -722,7 +723,7 @@ def main(spectra_file: Union[List[str], List[BinaryIO]] = SPECTRA_FILE,
          iRT_m: float = iRT_PARAMS["iRT_m"],
          iRT_t: float = iRT_PARAMS["iRT_t"],
          is_streamlit: bool = False,
-         save_output: bool = True) -> pd.DataFrame:
+         save_output: bool = True) -> Dict[str, pd.DataFrame]:
 
     if is_streamlit:
         print("INFO: Creating spectral library with input files:\nSpectra: " +
@@ -991,7 +992,7 @@ def main(spectra_file: Union[List[str], List[BinaryIO]] = SPECTRA_FILE,
         print("SUCCESS: Decoy Spectral library created with filename:")
         print(".".join(csms_file.split(".")[:-1]) + "_spectralLibraryDECOY.csv")
 
-    return spectral_library
+    return {"TargetLib": spectral_library, "DecoyLib": spectral_library_decoy}
 
 ##### SCRIPT #####
 

tests/tests.py

@@ -5,77 +5,108 @@
 # https://github.com/michabirklbauer/
 # [email protected]
 
-# check output shape
+# check output shape target
 def test1_spectral_library_exporter():
 
     from create_spectral_library import main
 
     sl = main()
+    sl = sl["TargetLib"]
 
     assert sl.shape[0] == 12 and sl.shape[1] == 27
 
-# check values
+# check values target
 def test2_spectral_library_exporter():
 
     from create_spectral_library import main
 
     sl = main()
+    sl = sl["TargetLib"]
 
-    value_tests = True
-
-    if str(sl.loc[0, "FragmentCharge"]) != "1":
-        value_tests = False
-    if str(sl.loc[0, "FragmentType"]) != "y":
-        value_tests = False
-    if str(sl.loc[0, "FragmentNumber"]) != "1":
-        value_tests = False
-    if str(sl.loc[0, "FragmentPepId"]) != "0":
-        value_tests = False
-    if str(sl.loc[0, "CLContainingFragment"]) != "False":
-        value_tests = False
-
-    if str(sl.loc[1, "FragmentCharge"]) != "1":
-        value_tests = False
-    if str(sl.loc[1, "FragmentType"]) != "b":
-        value_tests = False
-    if str(sl.loc[1, "FragmentNumber"]) != "2":
-        value_tests = False
-    if str(sl.loc[1, "FragmentPepId"]) != "0":
-        value_tests = False
-    if str(sl.loc[1, "CLContainingFragment"]) != "True":
-        value_tests = False
-
-    if str(sl.loc[2, "FragmentCharge"]) != "1":
-        value_tests = False
-    if str(sl.loc[2, "FragmentType"]) != "b":
-        value_tests = False
-    if str(sl.loc[2, "FragmentNumber"]) != "2":
-        value_tests = False
-    if str(sl.loc[2, "FragmentPepId"]) != "0":
-        value_tests = False
-    if str(sl.loc[2, "CLContainingFragment"]) != "True":
-        value_tests = False
-
-    if str(sl.loc[3, "FragmentCharge"]) != "1":
-        value_tests = False
-    if str(sl.loc[3, "FragmentType"]) != "y":
-        value_tests = False
-    if str(sl.loc[3, "FragmentNumber"]) != "3":
-        value_tests = False
-    if str(sl.loc[3, "FragmentPepId"]) != "0":
-        value_tests = False
-    if str(sl.loc[3, "CLContainingFragment"]) != "False":
-        value_tests = False
-
-    if str(sl.loc[11, "FragmentCharge"]) != "1":
-        value_tests = False
-    if str(sl.loc[11, "FragmentType"]) != "y":
-        value_tests = False
-    if str(sl.loc[11, "FragmentNumber"]) != "5":
-        value_tests = False
-    if str(sl.loc[11, "FragmentPepId"]) != "1":
-        value_tests = False
-    if str(sl.loc[11, "CLContainingFragment"]) != "False":
-        value_tests = False
-
-    assert value_tests
+    assert str(sl.loc[0, "ModifiedPeptide"]) == "KQQGHR_KQQGHR"
+    assert str(sl.loc[0, "FragmentCharge"]) == "1"
+    assert str(sl.loc[0, "FragmentType"]) == "y"
+    assert str(sl.loc[0, "FragmentNumber"]) == "1"
+    assert str(sl.loc[0, "FragmentPepId"]) == "0"
+    assert str(sl.loc[0, "CLContainingFragment"]) == "False"
+
+    assert str(sl.loc[1, "ModifiedPeptide"]) == "KQQGHR_KQQGHR"
+    assert str(sl.loc[1, "FragmentCharge"]) == "1"
+    assert str(sl.loc[1, "FragmentType"]) == "b"
+    assert str(sl.loc[1, "FragmentNumber"]) == "2"
+    assert str(sl.loc[1, "FragmentPepId"]) == "0"
+    assert str(sl.loc[1, "CLContainingFragment"]) == "True"
+
+    assert str(sl.loc[2, "ModifiedPeptide"]) == "KQQGHR_KQQGHR"
+    assert str(sl.loc[2, "FragmentCharge"]) == "1"
+    assert str(sl.loc[2, "FragmentType"]) == "b"
+    assert str(sl.loc[2, "FragmentNumber"]) == "2"
+    assert str(sl.loc[2, "FragmentPepId"]) == "0"
+    assert str(sl.loc[2, "CLContainingFragment"]) == "True"
+
+    assert str(sl.loc[3, "ModifiedPeptide"]) == "KQQGHR_KQQGHR"
+    assert str(sl.loc[3, "FragmentCharge"]) == "1"
+    assert str(sl.loc[3, "FragmentType"]) == "y"
+    assert str(sl.loc[3, "FragmentNumber"]) == "3"
+    assert str(sl.loc[3, "FragmentPepId"]) == "0"
+    assert str(sl.loc[3, "CLContainingFragment"]) == "False"
+
+    assert str(sl.loc[11, "ModifiedPeptide"]) == "KQQGHR_KQQGHR"
+    assert str(sl.loc[11, "FragmentCharge"]) == "1"
+    assert str(sl.loc[11, "FragmentType"]) == "y"
+    assert str(sl.loc[11, "FragmentNumber"]) == "5"
+    assert str(sl.loc[11, "FragmentPepId"]) == "1"
+    assert str(sl.loc[11, "CLContainingFragment"]) == "False"
+
+# check output shape decoy
+def test3_spectral_library_exporter():
+
+    from create_spectral_library import main
+
+    sl = main()
+    sl = sl["DecoyLib"]
+
+    assert sl.shape[0] == 22 and sl.shape[1] == 27
+
+# check values decoy
+def test4_spectral_library_exporter():
+
+    from create_spectral_library import main
+
+    sl = main()
+    sl = sl["DecoyLib"]
+
+    assert str(sl.loc[0, "ModifiedPeptide"]) == "HGQQKR_HGQQKR"
+    assert str(sl.loc[0, "FragmentCharge"]) == "1"
+    assert str(sl.loc[0, "FragmentType"]) == "y"
+    assert str(sl.loc[0, "FragmentNumber"]) == "1"
+    assert str(sl.loc[0, "FragmentPepId"]) == "0"
+    assert str(sl.loc[0, "CLContainingFragment"]) == "False"
+
+    assert str(sl.loc[1, "ModifiedPeptide"]) == "HGQQKR_HGQQKR"
+    assert str(sl.loc[1, "FragmentCharge"]) == "1"
+    assert str(sl.loc[1, "FragmentType"]) == "b"
+    assert str(sl.loc[1, "FragmentNumber"]) == "2"
+    assert str(sl.loc[1, "FragmentPepId"]) == "0"
+    assert str(sl.loc[1, "CLContainingFragment"]) == "False"
+
+    assert str(sl.loc[2, "ModifiedPeptide"]) == "HGQQKR_HGQQKR"
+    assert str(sl.loc[2, "FragmentCharge"]) == "1"
+    assert str(sl.loc[2, "FragmentType"]) == "y"
+    assert str(sl.loc[2, "FragmentNumber"]) == "3"
+    assert str(sl.loc[2, "FragmentPepId"]) == "0"
+    assert str(sl.loc[2, "CLContainingFragment"]) == "True"
+
+    assert str(sl.loc[3, "ModifiedPeptide"]) == "HGQQKR_HGQQKR"
+    assert str(sl.loc[3, "FragmentCharge"]) == "1"
+    assert str(sl.loc[3, "FragmentType"]) == "y"
+    assert str(sl.loc[3, "FragmentNumber"]) == "3"
+    assert str(sl.loc[3, "FragmentPepId"]) == "0"
+    assert str(sl.loc[3, "CLContainingFragment"]) == "True"
+
+    assert str(sl.loc[21, "ModifiedPeptide"]) == "HGQQKR_HGQQKR"
+    assert str(sl.loc[21, "FragmentCharge"]) == "1"
+    assert str(sl.loc[21, "FragmentType"]) == "y"
+    assert str(sl.loc[21, "FragmentNumber"]) == "5"
+    assert str(sl.loc[21, "FragmentPepId"]) == "1"
+    assert str(sl.loc[21, "CLContainingFragment"]) == "True"