Fragpipe params

proteobench/io/params/fragger.py

@@ -89,9 +89,15 @@ def extract_params(file: BytesIO) -> ProteoBenchParameters:
         msfragger_executable = ""
 
     params.search_engine_version = msfragger_executable
-    params.enzyme = fragpipe_params.loc["msfragger.search_enzyme_name_1"]
+    enzyme = fragpipe_params.loc["msfragger.search_enzyme_name_1"]
+    if fragpipe_params.loc["msfragger.search_enzyme_name_2"] != "null":
+        enzyme += f"|{fragpipe_params.loc['msfragger.search_enzyme_name_2']}"
+    params.enzyme = enzyme
     params.allowed_miscleavages = fragpipe_params.loc["msfragger.allowed_missed_cleavage_1"]
+    # TODO: Fragpipe reports "0.0" mass shift for each unmodified AA here, which is not useful and could be removed
     params.fixed_mods = fragpipe_params.loc["msfragger.table.fix-mods"]
+    # TODO: Fragpipe reports a lot of default suggestions for variable mods and assigns them as 'false' here,
+    # which is not useful and could be removed, i.e. only retain true variable mods
     params.variable_mods = fragpipe_params.loc["msfragger.table.var-mods"]
     params.max_mods = fragpipe_params.loc["msfragger.max_variable_mods_per_peptide"]
     params.min_peptide_length = fragpipe_params.loc["msfragger.digest_min_length"]
@@ -100,30 +106,54 @@ def extract_params(file: BytesIO) -> ProteoBenchParameters:
     precursor_mass_units = "Da"
     if int(fragpipe_params.loc["msfragger.precursor_mass_units"]):
         precursor_mass_units = "ppm"
-    params.precursor_mass_tolerance = (
-        f'{fragpipe_params.loc["msfragger.precursor_true_tolerance"]} {precursor_mass_units}'
-    )
+    params.precursor_mass_tolerance = f'{fragpipe_params.loc["msfragger.precursor_mass_lower"]} {precursor_mass_units}|{fragpipe_params.loc["msfragger.precursor_mass_upper"]} {precursor_mass_units}'
 
     fragment_mass_units = "Da"
     if int(fragpipe_params.loc["msfragger.fragment_mass_units"]):
         fragment_mass_units = "ppm"
     params.fragment_mass_tolerance = f'{fragpipe_params.loc["msfragger.fragment_mass_tolerance"]} {fragment_mass_units}'
     # ! ionquant is not necessarily fixed?
-    params.ident_fdr_protein = fragpipe_params.loc["ionquant.proteinfdr"]
-    params.ident_fdr_peptide = fragpipe_params.loc["ionquant.peptidefdr"]
-    params.ident_fdr_psm = fragpipe_params.loc["ionquant.ionfdr"]
-
-    for key in ["ident_fdr_protein", "ident_fdr_peptide", "ident_fdr_psm"]:
-        value = getattr(params, key)
-        try:
-            value = int(value) / 100
-            setattr(params, key, value)
-        except ValueError:
-            logging.warning(f"Could not convert {value} to int.")
-
-    params.min_precursor_charge = int(fragpipe_params.loc["msfragger.misc.fragger.precursor-charge-lo"])
-    params.max_precursor_charge = int(fragpipe_params.loc["msfragger.misc.fragger.precursor-charge-hi"])
-    params.enable_match_between_runs = bool(fragpipe_params.loc["ionquant.mbr"])
+    if fragpipe_params.loc["quantitation.run-label-free-quant"] == "true":
+        params.ident_fdr_protein = fragpipe_params.loc["ionquant.proteinfdr"]
+        params.ident_fdr_peptide = fragpipe_params.loc["ionquant.peptidefdr"]
+        params.ident_fdr_psm = fragpipe_params.loc["ionquant.ionfdr"]
+    elif fragpipe_params.loc["diann.run-dia-nn"] == "true":
+        params.ident_fdr_protein = fragpipe_params.loc["diann.q-value"]
+        params.ident_fdr_peptide = fragpipe_params.loc["diann.q-value"]
+        params.ident_fdr_psm = fragpipe_params.loc["diann.q-value"]
+
+    # I think this is incorrect? The values are stored as proportions in the fragpipe.workflow file? Commenting out for now
+    # for key in ["ident_fdr_protein", "ident_fdr_peptide", "ident_fdr_psm"]:
+    #     value = getattr(params, key)
+    #     try:
+    #         value = int(value) / 100
+    #         setattr(params, key, value)
+    #     except ValueError:
+    #         logging.warning(f"Could not convert {value} to int.")
+
+    if fragpipe_params.loc["msfragger.override_charge"] == "true":
+        params.min_precursor_charge = int(fragpipe_params.loc["msfragger.misc.fragger.precursor-charge-lo"])
+        params.max_precursor_charge = int(fragpipe_params.loc["msfragger.misc.fragger.precursor-charge-hi"])
+    else:  # Fragpipe takes charge info from data, this is the default
+        params.min_precursor_charge = 1
+        params.max_precursor_charge = None
+    if fragpipe_params.loc["quantitation.run-label-free-quant"] == "true":
+        params.enable_match_between_runs = bool(fragpipe_params.loc["ionquant.mbr"])
+    elif fragpipe_params.loc["diann.run-dia-nn"] == "true":
+        diann_quant_dict = {
+            1: "Any LC (high accuracy)",
+            2: "Any LC (high precision)",
+            3: "Robust LC (high accuracy)",
+            4: "Robust LC (high precision)",
+        }
+        if "--reanalyse" in fragpipe_params.loc["diann.fragpipe.cmd-opts"]:
+            params.enable_match_between_runs = True
+        else:
+            params.enable_match_between_runs = False
+        params.quantification_method_DIANN = diann_quant_dict[int(fragpipe_params.loc["diann.quantification-strategy"])]
+    if fragpipe_params.loc["protein-prophet.run-protein-prophet"] == "true":
+        params.protein_inference = "ProteinProphet: {}".format(fragpipe_params.loc["protein-prophet.cmd-opts"])
+
     return params
 
 

proteobench/modules/quant_base/quant_base_module.py

@@ -71,6 +71,7 @@ def __init__(self, token: str = None, proteobench_repo_name: str = "", proteobot
         "i2MassChroQ": extract_params_i2masschroq,
         "DIA-NN": extract_params_diann,
         "AlphaDIA": extract_params_alphadia,
+        "FragPipe (DIA-NN quant)": extract_params_fragger,
         # "Spectronaut": extract_params_spectronaut
     }
 

test/params/fragpipe_extracted_params.csv

@@ -4,10 +4,10 @@ software_version,21.0
 search_engine,MSFragger
 search_engine_version,4.0
 ident_fdr_psm,0.01
-ident_fdr_peptide,0.01
-ident_fdr_protein,0.01
+ident_fdr_peptide,1
+ident_fdr_protein,1
 enable_match_between_runs,True
-precursor_mass_tolerance,20 ppm
+precursor_mass_tolerance,-20 ppm|20 ppm
 fragment_mass_tolerance,20 ppm
 enzyme,stricttrypsin
 allowed_miscleavages,2
@@ -17,9 +17,9 @@ fixed_mods,"0.0,C-Term Peptide,true,-1; 0.0,N-Term Peptide,true,-1; 0.0,C-Term P
 variable_mods,"15.9949,M,true,3; 42.0106,[^,true,1; 79.96633,STY,false,3; -17.0265,nQnC,false,1; -18.0106,nE,false,1; 4.025107,K,false,2; 6.020129,R,false,2; 8.014199,K,false,2; 10.008269,R,false,2; 0.0,site_10,false,1; 0.0,site_11,false,1; 0.0,site_12,false,1; 0.0,site_13,false,1; 0.0,site_14,false,1; 0.0,site_15,false,1; 0.0,site_16,false,1"
 max_mods,3
 min_precursor_charge,1
-max_precursor_charge,4
+max_precursor_charge,
 scan_window,
 quantification_method_DIANN,
 second_pass,
-protein_inference,
+protein_inference,ProteinProphet: --maxppmdiff 2000000
 predictors_library,

test/params/fragpipe_v22_extracted_params.csv

@@ -4,10 +4,10 @@ software_version,22.0
 search_engine,MSFragger
 search_engine_version,
 ident_fdr_psm,0.01
-ident_fdr_peptide,0.01
-ident_fdr_protein,0.01
+ident_fdr_peptide,1
+ident_fdr_protein,1
 enable_match_between_runs,True
-precursor_mass_tolerance,10 ppm
+precursor_mass_tolerance,-20 ppm|20 ppm
 fragment_mass_tolerance,0.02 Da
 enzyme,stricttrypsin
 allowed_miscleavages,2
@@ -17,9 +17,9 @@ fixed_mods,"0.0,C-Term Peptide,true,-1; 0.0,N-Term Peptide,true,-1; 0.0,C-Term P
 variable_mods,"15.9949,M,true,3; 42.0106,[^,true,1; 79.96633,STY,false,3; -17.0265,nQnC,false,1; -18.0106,nE,false,1; 4.025107,K,false,2; 6.020129,R,false,2; 8.014199,K,false,2; 10.008269,R,false,2; 0.0,site_10,false,1; 0.0,site_11,false,1; 0.0,site_12,false,1; 0.0,site_13,false,1; 0.0,site_14,false,1; 0.0,site_15,false,1; 0.0,site_16,false,1"
 max_mods,3
 min_precursor_charge,1
-max_precursor_charge,4
+max_precursor_charge,
 scan_window,
 quantification_method_DIANN,
 second_pass,
-protein_inference,
-predictors_library,
+protein_inference,ProteinProphet: --maxppmdiff 2000000
+predictors_library,

test/params/fragpipe_win_paths_extracted_params.csv

@@ -4,10 +4,10 @@ software_version,21.1
 search_engine,MSFragger
 search_engine_version,4.0
 ident_fdr_psm,0.01
-ident_fdr_peptide,0.01
-ident_fdr_protein,0.01
+ident_fdr_peptide,1
+ident_fdr_protein,1
 enable_match_between_runs,True
-precursor_mass_tolerance,10 ppm
+precursor_mass_tolerance,-20 ppm|20 ppm
 fragment_mass_tolerance,0.02 Da
 enzyme,stricttrypsin
 allowed_miscleavages,2
@@ -17,9 +17,9 @@ fixed_mods,"0.0,C-Term Peptide,true,-1; 0.0,N-Term Peptide,true,-1; 0.0,C-Term P
 variable_mods,"15.9949,M,true,3; 42.0106,[^,true,1; 79.96633,STY,false,3; -17.0265,nQnC,false,1; -18.0106,nE,false,1; 4.025107,K,false,2; 6.020129,R,false,2; 8.014199,K,false,2; 10.008269,R,false,2; 0.0,site_10,false,1; 0.0,site_11,false,1; 0.0,site_12,false,1; 0.0,site_13,false,1; 0.0,site_14,false,1; 0.0,site_15,false,1; 0.0,site_16,false,1"
 max_mods,3
 min_precursor_charge,1
-max_precursor_charge,4
+max_precursor_charge,
 scan_window,
 quantification_method_DIANN,
 second_pass,
-protein_inference,
+protein_inference,ProteinProphet: --maxppmdiff 2000000
 predictors_library,