Merge branch 'main' into parse_settings

proteobench/modules/dda_quant/datapoint.py

@@ -1,12 +1,13 @@
 import json
+import numpy as np
 from dataclasses import asdict, dataclass
 from datetime import datetime
 
 
 @dataclass
 class Datapoint:
     """Data used to stored the"""
-
+    # TODO add threshold value used for presence ion/peptidoform
     id: str = None
     search_engine: str = None
     software_version: int = 0
@@ -44,13 +45,14 @@ def calculate_plot_data(self, df):
         nr_missing_0 = 0
         for spec in species:
             f = len(df[df[spec] == True])
-            sum_s = (df[df[spec] == True]["1|2_expected_ratio_diff"]).sum()
+            sum_s = np.nan_to_num(df[df[spec] == True]["1|2_expected_ratio_diff"], nan=0, neginf=-1000, posinf=1000).sum()
             ratio = sum_s / f
             prop_ratio = (f / len(df)) * ratio
             prop_ratios.append(prop_ratio)
             sum_ratios += prop_ratio
             nr_missing_0 += f
-
+
+        # TODO rename/document code
         self.weighted_sum = round(sum_ratios, ndigits=3)
         self.nr_prec = len(df)
 
@@ -64,6 +66,7 @@ def generate_id(self):
         )
         print(self.id)
 
+    # TODO, not used? Can be removed?
     def dump_json_object(self, file_name):
         f = open(file_name, "a")
         f.write(json.dumps(asdict(self)))

proteobench/modules/dda_quant/io_parse_settings/module_settings.toml

@@ -0,0 +1,12 @@
+[species_expected_ratio]
+[species_expected_ratio.YEAST]
+"1|2" = 2.0
+
+[species_expected_ratio.ECOLI]
+"1|2" = 0.25
+
+[species_expected_ratio.HUMAN]
+"1|2" = 1.0
+
+[general]
+min_count_multispec = 1

proteobench/modules/dda_quant/io_parse_settings/parse_settings_alphapept.toml

@@ -16,29 +16,18 @@ LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02 = 2
 LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03 = 2
 
 [run_mapper]
-LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01"
-LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02"
-LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03"
-LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01"
-LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02"
-LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03"
+LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01 = "Condition_A_Sample_Alpha_01"
+LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02 = "Condition_A_Sample_Alpha_02"
+LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03 = "Condition_A_Sample_Alpha_03"
+LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01 = "Condition_B_Sample_Alpha_01"
+LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02 = "Condition_B_Sample_Alpha_02"
+LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03 = "Condition_B_Sample_Alpha_03"
 
-[species_dict]
-"YEAST" = "_YEAST"
-"ECOLI" = "_ECOLI"
-"HUMAN" = "_HUMAN"
-
-[species_expected_ratio]
-[species_expected_ratio.YEAST]
-"1|2" = 0.5
-
-[species_expected_ratio.ECOLI]
-"1|2" = 1.5
-
-[species_expected_ratio.HUMAN]
-"1|2" = 1.0
+[species_mapper]
+"_YEAST" = "YEAST"
+"_ECOLI" = "ECOLI"
+"_HUMAN" = "HUMAN"
 
 [general]
 contaminant_flag = "Cont_"
 decoy_flag = true
-min_count_multispec = 1

proteobench/modules/dda_quant/io_parse_settings/parse_settings_custom.toml

@@ -13,29 +13,18 @@ LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02 = 2
 LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03 = 2
 
 [run_mapper]
-LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01"
-LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02"
-LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03"
-LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01"
-LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02"
-LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03"
+LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01 = "Condition_A_Sample_Alpha_01"
+LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02 = "Condition_A_Sample_Alpha_02"
+LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03 = "Condition_A_Sample_Alpha_03"
+LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01 = "Condition_B_Sample_Alpha_01"
+LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02 = "Condition_B_Sample_Alpha_02"
+LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03 = "Condition_B_Sample_Alpha_03"
 
-[species_dict]
-"YEAST" = "_YEAST"
-"ECOLI" = "_ECOLI"
-"HUMAN" = "_HUMAN"
-
-[species_expected_ratio]
-[species_expected_ratio.YEAST]
-"1|2" = 0.5
-
-[species_expected_ratio.ECOLI]
-"1|2" = 1.5
-
-[species_expected_ratio.HUMAN]
-"1|2" = 1.0
+[species_mapper]
+"_YEAST" = "YEAST"
+"_ECOLI" = "ECOLI"
+"_HUMAN" = "HUMAN"
 
 [general]
 contaminant_flag = "Cont_"
 decoy_flag = true
-min_count_multispec = 1

proteobench/modules/dda_quant/io_parse_settings/parse_settings_maxquant.toml

@@ -14,38 +14,18 @@ LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02 = 2
 LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03 = 2
 
 [run_mapper]
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01"
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02"
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03"
-"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01"
-"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02"
-"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03"
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01" = "Condition_A_Sample_Alpha_01"
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02" = "Condition_A_Sample_Alpha_02"
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03" = "Condition_A_Sample_Alpha_03"
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01" = "Condition_B_Sample_Alpha_01"
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02" = "Condition_B_Sample_Alpha_02"
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03" = "Condition_B_Sample_Alpha_03"
 
-[species_dict]
-"YEAST" = "_YEAST"
-"ECOLI" = "_ECOLI"
-"HUMAN" = "_HUMAN"
-
-[species_expected_ratio]
-[species_expected_ratio.YEAST]
-"1|2" = 0.5
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01" = 0.5
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02" = 0.5
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03" = 0.5
-
-[species_expected_ratio.ECOLI]
-"1|2" = 1.5
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01" = 1.5
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02" = 1.5
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03" = 1.5
-
-[species_expected_ratio.HUMAN]
-"1|2" = 1.0
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01" = 1.0
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02" = 1.0
-"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03|LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03" = 1.0
+[species_mapper]
+"_YEAST" = "YEAST"
+"_ECOLI" = "ECOLI"
+"_HUMAN" = "HUMAN"
 
 [general]
 contaminant_flag = "Cont_"
 decoy_flag = true
-min_count_multispec = 1

proteobench/modules/dda_quant/io_parse_settings/parse_settings_msfragger.toml

@@ -13,30 +13,19 @@ Charge = "Charge"
 "B_3 Intensity" = 2
 
 [run_mapper]
-"A_1 Intensity" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01"
-"A_2 Intensity" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02"
-"A_3 Intensity" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03"
-"B_1 Intensity" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01"
-"B_2 Intensity" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02"
-"B_3 Intensity" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03"
+"A_1 Intensity" = "Condition_A_Sample_Alpha_01"
+"A_2 Intensity" = "Condition_A_Sample_Alpha_02"
+"A_3 Intensity" = "Condition_A_Sample_Alpha_03"
+"B_1 Intensity" = "Condition_B_Sample_Alpha_01"
+"B_2 Intensity" = "Condition_B_Sample_Alpha_02"
+"B_3 Intensity" = "Condition_B_Sample_Alpha_03"
 
 
-[species_dict]
-"YEAST" = "_YEAST"
-"ECOLI" = "_ECOLI"
-"HUMAN" = "_HUMAN"
-
-[species_expected_ratio]
-[species_expected_ratio.YEAST]
-"1|2" = 0.5
-
-[species_expected_ratio.ECOLI]
-"1|2" = 1.5
-
-[species_expected_ratio.HUMAN]
-"1|2" = 1.0
+[species_mapper]
+"_YEAST" = "YEAST"
+"_ECOLI" = "ECOLI"
+"_HUMAN" = "HUMAN"
 
 [general]
 contaminant_flag = "Cont_"
 decoy_flag = false
-min_count_multispec = 1

proteobench/modules/dda_quant/io_parse_settings/parse_settings_proline.toml

@@ -14,29 +14,18 @@ abundance_DDA_Condition_B_Sample_Alpha_02 = 2
 abundance_DDA_Condition_B_Sample_Alpha_03 = 2
 
 [run_mapper]
-"abundance_DDA_Condition_A_Sample_Alpha_01" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01"
-"abundance_DDA_Condition_A_Sample_Alpha_02" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02"
-"abundance_DDA_Condition_A_Sample_Alpha_03" = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03"
-"abundance_DDA_Condition_B_Sample_Alpha_01" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01"
-"abundance_DDA_Condition_B_Sample_Alpha_02" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02"
-"abundance_DDA_Condition_B_Sample_Alpha_03" = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03"
+"abundance_DDA_Condition_A_Sample_Alpha_01" = "Condition_A_Sample_Alpha_01"
+"abundance_DDA_Condition_A_Sample_Alpha_02" = "Condition_A_Sample_Alpha_02"
+"abundance_DDA_Condition_A_Sample_Alpha_03" = "Condition_A_Sample_Alpha_03"
+"abundance_DDA_Condition_B_Sample_Alpha_01" = "Condition_B_Sample_Alpha_01"
+"abundance_DDA_Condition_B_Sample_Alpha_02" = "Condition_B_Sample_Alpha_02"
+"abundance_DDA_Condition_B_Sample_Alpha_03" = "Condition_B_Sample_Alpha_03"
 
-[species_dict]
-"YEAST" = "_YEAST"
-"ECOLI" = "_ECOLI"
-"HUMAN" = "_HUMAN"
-
-[species_expected_ratio]
-[species_expected_ratio.YEAST]
-"1|2" = 0.5
-
-[species_expected_ratio.ECOLI]
-"1|2" = 1.5
-
-[species_expected_ratio.HUMAN]
-"1|2" = 1.0
+[species_mapper]
+"_YEAST" = "YEAST"
+"_ECOLI" = "ECOLI"
+"_HUMAN" = "HUMAN"
 
 [general]
 contaminant_flag = "Cont_"
 decoy_flag = true
-min_count_multispec = 1

proteobench/modules/dda_quant/io_parse_settings/parse_settings_sage.toml

@@ -0,0 +1,29 @@
+[mapper]
+"proteins" = "Proteins"
+"peptide" = "Sequence"
+"charge" = "Charge"
+
+[replicate_mapper]
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01.mzML.gz" = 1
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02.mzML.gz" = 1
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03.mzML.gz" = 1
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01.mzML.gz" = 2
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02.mzML.gz" = 2
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03.mzML.gz" = 2
+
+[run_mapper]
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01.mzML" = "Condition_A_Sample_Alpha_01"
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02.mzML" = "Condition_A_Sample_Alpha_02"
+"LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03.mzML" = "Condition_A_Sample_Alpha_03"
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01.mzML" = "Condition_B_Sample_Alpha_01"
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02.mzML" = "Condition_B_Sample_Alpha_02"
+"LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03.mzML" = "Condition_B_Sample_Alpha_03"
+
+[species_mapper]
+"_YEAST" = "YEAST"
+"_ECOLI" = "ECOLI"
+"_HUMAN" = "HUMAN"
+
+[general]
+contaminant_flag = "Cont_"
+decoy_flag = false

proteobench/modules/dda_quant/io_parse_settings/parse_settings_wombat.toml

@@ -2,7 +2,6 @@
 protein_group = "Proteins"
 "modified_peptide" = "Modified sequence"
 
-
 [replicate_mapper]
 abundance_A_1 = 1
 abundance_A_2 = 1
@@ -12,29 +11,18 @@ abundance_B_2 = 2
 abundance_B_3 = 2
 
 [run_mapper]
-abundance_A_1 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_01"
-abundance_A_2 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_02"
-abundance_A_3 = "LFQ_Orbitrap_DDA_Condition_A_Sample_Alpha_03"
-abundance_B_1 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_01"
-abundance_B_2 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_02"
-abundance_B_3 = "LFQ_Orbitrap_DDA_Condition_B_Sample_Alpha_03"
-
-[species_dict]
-"YEAST" = "_YEAST"
-"ECOLI" = "_ECOLI"
-"HUMAN" = "_HUMAN"
-
-[species_expected_ratio]
-[species_expected_ratio.YEAST]
-"1|2" = 0.5
+abundance_A_1 = "Condition_A_Sample_Alpha_01"
+abundance_A_2 = "Condition_A_Sample_Alpha_02"
+abundance_A_3 = "Condition_A_Sample_Alpha_03"
+abundance_B_1 = "Condition_B_Sample_Alpha_01"
+abundance_B_2 = "Condition_B_Sample_Alpha_02"
+abundance_B_3 = "Condition_B_Sample_Alpha_03"
 
-[species_expected_ratio.ECOLI]
-"1|2" = 1.5
-
-[species_expected_ratio.HUMAN]
-"1|2" = 1.0
+[species_mapper]
+"_YEAST" = "YEAST"
+"_ECOLI" = "ECOLI"
+"_HUMAN" = "HUMAN"
 
 [general]
 contaminant_flag = "Cont_"
-decoy_flag = true
-min_count_multispec = 1
+decoy_flag = true