Merge branch 'develop' into fix/mito_nuc_trans

code/test/sanityCheck.py

@@ -4,6 +4,7 @@
 
 import pandas as pd
 import cobra
+from collections import Counter
 
 
 def get_column_from_tsv(tsv_file, column_id, to_list=True):
@@ -106,11 +107,28 @@ def checkUnusedEntities(model, entity_type):
     assert len(unused_entities) == 0, f"Found unused {entity_type}: {unused_entities}"
 
 
+def checkDupRxn(model):
+    """
+    Check for duplicate reactions in the model
+    """
+
+    reaction_equations = [rxn.build_reaction_string(use_metabolite_names=False) for rxn in model.reactions]
+    duplicate_reactions = [reaction for reaction, count in Counter(reaction_equations).items() if count > 1]
+    dup_list = [model.reactions[idx].id for idx, val in enumerate(reaction_equations) if val in duplicate_reactions]
+
+    if duplicate_reactions:
+        output = f"The following {len(dup_list)} reactions are duplicates, please check: " + ';'.join(dup_list)
+        print(output)
+
+    assert len(duplicate_reactions) == 0, "Found duplicated reactions!"
+
+
 if __name__ == "__main__":
     rxns, mets, genes, model = load_yml("model/Human-GEM.yml")
     checkRxnAnnotation(rxns)
     checkMetAnnotation(mets)
     checkGeneAnnotation(genes)
     checkUnusedEntities(model, "metabolites")
     checkUnusedEntities(model, "genes")
+    checkDupRxn(model)
     print("All checks have passed.")

data/deprecatedIdentifiers/deprecatedMetabolites.tsv

@@ -2237,6 +2237,7 @@ mets	metsNoComp	metBiGGID	metKEGGID	metHMDBID	metChEBIID	metPubChemID	metLipidMa
 "MAM01445m"	"MAM01445"	"cholate"	"C00695"	"HMDB0000619"	"CHEBI:16359"	"221493"	"LMST04010001"	""	"HC00502"	"cholate"	"MNXM450"	"m01445m"	"m01445m"
 "MAM01514m"	"MAM01514"	"cholcoa"	"C01794"	"HMDB0001374"	"CHEBI:15519"	"439573"	""	""	"HC00844"	"cholcoa"	"MNXM162468;MNXM431"	"m01514m"	"m01514m"
 "MAM00759c"	"MAM00759"	""	"C13712"	""	""	""	"LMST02030130"	""	""	"M00759"	"MNXM3494"	"m00759c"	"m00759c"
+"MAM02397c"	"MAM02397"	""	"C16239"	""	""	""	""	""	""	"M02397"	"MNXM19093"	"m02397c"	"m02397c"
 "MAM01268c"	"MAM01268"	""	""	""	""	""	"PROTEIN"	""	""	"M01268"	""	"m01268c"	"m01268c"
 "MAM01268m"	"MAM01268"	""	""	""	""	""	"PROTEIN"	""	""	"M01268"	""	"m01268m"	"m01268m"
 "MAM01268n"	"MAM01268"	""	""	""	""	""	"PROTEIN"	""	""	"M01268"	""	"m01268n"	"m01268n"

data/deprecatedIdentifiers/deprecatedReactions.tsv

@@ -230,6 +230,8 @@ rxns	rxnKEGGID	rxnBiGGID	rxnEHMNID	rxnHepatoNET1ID	rxnREACTOMEID	rxnRecon3DID	rx
 "MAR00096"	""	"BUP2"	""	""	""	"BUP2"	"MNXR96346"	""	""	""	0	""	""	"BUP2"
 "MAR02027"	""	"r0193"	""	""	""	"r0193"	"MNXR105313"	""	""	""	0	""	""	"r0193"
 "MAR02006"	""	"CPS_m"	""	""	""	"r0034"	""	""	""	""	0	""	""	"r0034"
+"MAR06400"	"R07768"	""	"R07768C"	""	""	"HMR_6400"	"MNXR111356"	"HMR_6400"	"RCR11944"	""	0	""	""	"HMR_6400"
+"MAR06401"	"R07769"	""	"R07769C"	""	""	"HMR_6401"	"MNXR111357"	"HMR_6401"	"RCR11945"	""	0	""	""	"HMR_6401"
 "MAR03940"	"R01795"	"PHETHPTOX"	"R01795C"	"r0399"	""	"r0399"	"MNXR102632;MNXR107174"	"HMR_3940"	"RCR11388"	""	0	""	"RHEA:30787"	"HMR_3940"
 "MAR07785"	""	"ATPtx"	""	""	""	"ATPtx"	"MNXR96140"	"HMR_7785"	"RCR20040"	""	0	""	""	"HMR_7785"
 "MAR07775"	""	"NADHtpu"	""	""	""	"NADHtpu;NADtpu"	"MNXR101881"	"HMR_7775"	"RCR20024"	""	0	""	""	"HMR_7775"
@@ -247,3 +249,7 @@ rxns	rxnKEGGID	rxnBiGGID	rxnEHMNID	rxnHepatoNET1ID	rxnREACTOMEID	rxnRecon3DID	rx
 "MAR07926"	""	"DGTPtm"	""	""	""	"DGTPtm"	"MNXR97327"	""	""	""	0	""	""	"DGTPtm"
 "MAR01483"	""	"LIPOti"	""	""	""	"LIPOti"	"MNXR101093"	""	""	""	0	""	""	"LIPOti"
 "MAR06406"	""	"LIPOti"	"RT1089"	""	""	""	"MNXR101093"	"HMR_6406"	"RCR41534"	""	0	""	""	"HMR_6406"
+"MAR00478"	"R00844"	""	""	""	"R-HSA-75889"	"HMR_0478"	"MNXR99876"	"HMR_0478"	"RCR14575"	""	0	"RHEA:11098"	"RHEA:11096"	"HMR_0478"
+"MAR00482"	"R00848;R00849"	"G3PD2m"	"R00848C"	"r0205"	""	"HMR_0482;r0205"	"MNXR99875;MNXR106713"	"HMR_0482;HMR_0483"	"RCR14578;RCR21050"	""	0	""	"RHEA:31283;RHEA:18977"	"HMR_0482;HMR_0483;MAR00483"
+"MAR20012"	"R02123"	"HMR_6647;RADH;R_HMR_6647;R_RADH"	""	""	""	""	"MNXR146215"	""	""	""	0	"RHEA:42080"	""	""
+"MAR20089"	"R02701"	""	""	""	""	""	"MNXR151808"	""	""	""	0	"RHEA:18533"	""	""

data/metabolicTasks/metabolicTasks_CellfieConsensus.txt

@@ -220,12 +220,12 @@
 	36	Synthesis of lactose		glucose[g]	1	1	lactose[g]	1	1								
 				UDP-galactose[g]	1	1	UDP[g]	1	1								
 							H+[g]	1	1								
-	37	Glycogen biosynthesis		NA[c]	1	1	glycogenin G4G7[c]	1	1								
-				UDP-glucose[c]	11		H+[c]	11									
-							UDP[c]	11									
-	38	Glycogen degradation		glycogenin G4G7[c]	1	1	glucose[c]	8									
-				Pi[c]	3		NA[c]	1	1								
-				H2O[c]	8		glucose-1-phosphate[c]	3									
+#	37	Glycogen biosynthesis		NA[c]	1	1	glycogenin G4G7[c]	1	1								The metabolite NA[c] in this task is conflicting with HumanGEM
+#				UDP-glucose[c]	11		H+[c]	11									
+#							UDP[c]	11									
+#	38	Glycogen degradation		glycogenin G4G7[c]	1	1	glucose[c]	8									The metabolite NA[c] in this task is conflicting with HumanGEM
+#				Pi[c]	3		NA[c]	1	1								
+#				H2O[c]	8		glucose-1-phosphate[c]	3									
 	39	Fructose degradation (to glucose-3-phosphate)		fructose[c]	1	1	D-glyceraldehyde 3-phosphate[c]	2	2								
 				ATP[c]	2	2	ADP[c]	2	2								
 							H+[c]	2	2								
@@ -1130,14 +1130,14 @@
 	180	Pyridoxal-phosphate synthesis		pyridoxal[c]	1	1	pyridoxal-phosphate[c]	1	1								
 				ATP[c]	1	1	ADP[c]	1	1								
 							H+[c]	1	1								
-	181	Synthesis of bilirubin		heme[c]	1	1	bilirubin[c]	1	1								
-				NADPH[c]	0	4	H2O[c]	0	3								
-				H+[c]	0	6	Fe2+[c]	0	1								
-				O2[c]	0	3	Fe3+[c]	0	1								
-				FADH2[c]	0	1	NADP+[c]	0	4								
-							FAD[c]	0	1								
-							CO[c]	1	1								
-							NA[c]	0	1								
+#	181	Synthesis of bilirubin		heme[c]	1	1	bilirubin[c]	1	1								The metabolite NA[c] in this task is conflicting with HumanGEM
+#				NADPH[c]	0	4	H2O[c]	0	3								
+#				H+[c]	0	6	Fe2+[c]	0	1								
+#				O2[c]	0	3	Fe3+[c]	0	1								
+#				FADH2[c]	0	1	NADP+[c]	0	4								
+#							FAD[c]	0	1								
+#							CO[c]	1	1								
+#							NA[c]	0	1								
 	182	Heme synthesis		glycine[m]	16	16	heme[m]	2	2								
 				succinyl-CoA[m]	16	18	H2O[m]	26	26								
 				Fe2+[m]	2	2	CoA[m]	16	18								

model/Human-GEM.yml

@@ -32657,14 +32657,6 @@
       - charge: -1
       - inchis: ""
       - metFrom: "HMRdatabase"
-    - !!omap
-      - id: "MAM02397c"
-      - name: "lipoyl-[ACP]"
-      - compartment: "c"
-      - formula: "C19H33N2O8PRS3"
-      - charge: -1
-      - inchis: ""
-      - metFrom: "HMRdatabase"
     - !!omap
       - id: "MAM02398c"
       - name: "lipoyl-AMP"
@@ -100713,49 +100705,6 @@
       - subsystem:
           - "Lipoic acid metabolism"
       - confidence_score: 0
-    - !!omap
-      - id: "MAR06400"
-      - name: ""
-      - metabolites: !!omap
-          - MAM01098c: 2
-          - MAM02039c: -5
-          - MAM02040c: 6
-          - MAM02397c: 1
-          - MAM02471c: 2
-          - MAM02554c: 7
-          - MAM02555c: -7
-          - MAM02643c: -1
-          - MAM02681c: 2
-          - MAM02682c: -2
-          - MAM02877c: -2
-      - lower_bound: 0
-      - upper_bound: 1000
-      - gene_reaction_rule: "ENSG00000121897 or ENSG00000175536"
-      - rxnNotes: ""
-      - rxnFrom: "HMRdatabase"
-      - eccodes: "2.8.1.8;2.3.1.181"
-      - references: ""
-      - subsystem:
-          - "Lipoic acid metabolism"
-      - confidence_score: 0
-    - !!omap
-      - id: "MAR06401"
-      - name: "lipoyl-[acp]:protein N6-lipoyltransferase"
-      - metabolites: !!omap
-          - MAM00184c: 1
-          - MAM00209c: 1
-          - MAM01350c: -1
-          - MAM02397c: -1
-      - lower_bound: 0
-      - upper_bound: 1000
-      - gene_reaction_rule: "ENSG00000175536"
-      - rxnNotes: ""
-      - rxnFrom: "HMRdatabase"
-      - eccodes: "2.3.1.181"
-      - references: ""
-      - subsystem:
-          - "Lipoic acid metabolism"
-      - confidence_score: 0
     - !!omap
       - id: "MAR06402"
       - name: "octanoyl-[acp]:protein N6-octanoyltransferase"
@@ -133497,25 +133446,6 @@
       - subsystem:
           - "Glycerophospholipid metabolism"
       - confidence_score: 0
-    - !!omap
-      - id: "MAR00478"
-      - name: "sn-Glycerol-3-phosphate:NADP+ 2-oxidoreductase"
-      - metabolites: !!omap
-          - MAM01690c: -1
-          - MAM02039c: -1
-          - MAM02554c: 1
-          - MAM02555c: -1
-          - MAM02914c: 1
-      - lower_bound: 0
-      - upper_bound: 1000
-      - gene_reaction_rule: "ENSG00000167588"
-      - rxnNotes: ""
-      - rxnFrom: "HMRdatabase"
-      - eccodes: "1.1.1.94"
-      - references: "PMID:16460752"
-      - subsystem:
-          - "Glycerophospholipid metabolism"
-      - confidence_score: 0
     - !!omap
       - id: "MAR00479"
       - name: "sn-Glycerol-3-phosphate:NAD+ 2-oxidoreductase"
@@ -133531,7 +133461,7 @@
       - rxnNotes: ""
       - rxnFrom: "HMRdatabase"
       - eccodes: "1.1.1.8"
-      - references: "PMID:17045662;PMID:7772607"
+      - references: "PMID:16460752;PMID:17045662;PMID:7772607"
       - subsystem:
           - "Glycerophospholipid metabolism"
       - confidence_score: 0
@@ -133554,24 +133484,6 @@
       - subsystem:
           - "Glycerophospholipid metabolism"
       - confidence_score: 0
-    - !!omap
-      - id: "MAR00482"
-      - name: "sn-Glycerol-3-phosphate:(acceptor) 2-oxidoreductase"
-      - metabolites: !!omap
-          - MAM01690c: 1
-          - MAM01802c: -1
-          - MAM01803c: 1
-          - MAM02914c: -1
-      - lower_bound: 0
-      - upper_bound: 1000
-      - gene_reaction_rule: "ENSG00000115159"
-      - rxnNotes: ""
-      - rxnFrom: "HMRdatabase"
-      - eccodes: "1.1.5.3"
-      - references: "PMID:11955283;PMID:2115809;PMID:2923620;PMID:3338458;PMID:340460;PMID:8401296;PMID:8549872;PMID:8579375;PMID:9171333;PMID:9244403;PMID:9559543"
-      - subsystem:
-          - "Glycerophospholipid metabolism"
-      - confidence_score: 0
     - !!omap
       - id: "MAR00484"
       - name: "CDPglycerol phosphoglycerohydrolase"
@@ -155269,7 +155181,7 @@
           - MAM02833c: 1
       - lower_bound: 0
       - upper_bound: 1000
-      - gene_reaction_rule: "ENSG00000128918 or ENSG00000165092"
+      - gene_reaction_rule: "ENSG00000128918 or ENSG00000165092 or ENSG00000184254"
       - rxnNotes: ""
       - rxnFrom: "HMRdatabase"
       - eccodes: "1.2.1.36"
@@ -302678,26 +302590,6 @@
       - subsystem:
           - "Retinol metabolism"
       - confidence_score: 0
-    - !!omap
-      - id: "MAR20012"
-      - name: "MAR20012"
-      - metabolites: !!omap
-          - MAM02039c: 2
-          - MAM02040c: -1
-          - MAM02552c: -1
-          - MAM02553c: 1
-          - MAM02832c: -1
-          - MAM02833c: 1
-      - lower_bound: 0
-      - upper_bound: 1000
-      - gene_reaction_rule: "ENSG00000184254"
-      - rxnNotes: ""
-      - rxnFrom: ""
-      - eccodes: "1.2.1.36"
-      - references: ""
-      - subsystem:
-          - "Retinol metabolism"
-      - confidence_score: 0
     - !!omap
       - id: "MAR20013"
       - name: "MAR20013"
@@ -304045,24 +303937,6 @@
       - subsystem:
           - "Fatty acid metabolism"
       - confidence_score: 0
-    - !!omap
-      - id: "MAR20089"
-      - name: "MAR20089"
-      - metabolites: !!omap
-          - MAM01107c: -1
-          - MAM01596c: 1
-          - MAM02039c: -1
-          - MAM02897c: 1
-      - lower_bound: 0
-      - upper_bound: 1000
-      - gene_reaction_rule: "ENSG00000132437"
-      - rxnNotes: ""
-      - rxnFrom: ""
-      - eccodes: "4.1.1.28"
-      - references: ""
-      - subsystem:
-          - "Tryptophan metabolism"
-      - confidence_score: 0
     - !!omap
       - id: "MAR20090"
       - name: "MAR20090"

model/metabolites.tsv

@@ -4080,7 +4080,6 @@ mets	metsNoComp	metBiGGID	metKEGGID	metHMDBID	metChEBIID	metPubChemID	metLipidMa
 "MAM02396n"	"MAM02396"	""	"C06315"	"HMDB0005082"	"CHEBI:6499"	"5280915"	"LMFA03040002"	""	""	"C06315"	"MNXM12130"	"m02396n"	"m02396n"
 "MAM02396x"	"MAM02396"	""	"C06315"	"HMDB0005082"	"CHEBI:6499"	"5280915"	"LMFA03040002"	""	""	"C06315"	"MNXM12130"	"m02396p"	"m02396p"
 "MAM02396r"	"MAM02396"	""	"C06315"	"HMDB0005082"	"CHEBI:6499"	"5280915"	"LMFA03040002"	""	""	"C06315"	"MNXM12130"	"m02396r"	"m02396r"
-"MAM02397c"	"MAM02397"	""	"C16239"	""	""	""	""	""	""	"M02397"	"MNXM19093"	"m02397c"	"m02397c"
 "MAM02398c"	"MAM02398"	""	"C16238"	""	""	"23724672"	""	"CE5656"	""	"CE5656"	"MNXM2392"	"m02398c"	"m02398c"
 "MAM02399c"	"MAM02399"	""	""	""	""	"53481572"	""	"CE2102"	""	"CE2102"	"MNXM60228"	"m02399c"	"m02399c"
 "MAM02401c"	"MAM02401"	""	"C03127"	""	"CHEBI:29160"	""	""	""	""	"M02401"	"MNXM89832"	"m02401c"	"m02401c"