Merge pull request #305 from SysBioChalmers/complex-annotation

Complex annotation

.gitignore

@@ -52,7 +52,9 @@ helpsearch*/
 *.pyc
 *.env
 
-# Non-complying tables #
-########################
-*.xls
+# Non-complying tables and files #
+##################################
+*.xls*
+*.doc*
+*.ppt*
 *.tab

README.md

@@ -33,7 +33,7 @@ This repository contains the current consensus genome-scale metabolic model of _
 
 | Taxonomy | Latest update | Version | Reactions | Metabolites | Genes |
 |:-------|:--------------|:------|:------|:----------|:-----|
-| _Saccharomyces cerevisiae_ | 28-May-2022 | develop | 4069 | 2749 | 1150 |
+| _Saccharomyces cerevisiae_ | 28-May-2022 | develop | 4069 | 2749 | 1161 |
 
 # Installation & usage
 

code/getEarlierModelVersion.m

@@ -11,7 +11,7 @@
 
 if any(regexp(version,'^\d+\.\d+\.\d+$')) % Tag is a version number
     tagpath=['git show refs/tags/v' version ':'];
-elseif any(contains(tag,{'main','develop'}))
+elseif any(contains(version,{'main','develop'}))
     tagpath=['git show ' version ':'];
 else
     error('version should be ''main'', ''develop'', or a specific release, e.g. ''8.1.0''')

code/loadYeastModel.m

@@ -36,4 +36,5 @@
         model = ravenCobraWrapper(model);
     end
 end
+disp('If there is only 1 warning and no errors, it can savely be ignored.')
 end

code/modelCuration/complexAnnotation.m

@@ -0,0 +1,28 @@
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+% complexAnnotation
+% Correct complex annotation
+% As for the reference of new GPRs, please find detailed information in:
+% data/modelCuration/complexAnnotation/complexAnnotation.tsv
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+% Load model
+cd ..
+model = loadYeastModel;
+
+% Add new genes
+newModel       = curateMetsRxnsGenes(model,'none','../data/modelCuration/complexAnnotation/complexAnnotationGenes.tsv');
+
+% Add gene standard name for new genes
+fid = fopen('../data/modelCuration/complexAnnotation/complexAnnotation.tsv');
+complexAnnot = textscan(fid,'%q %q %q %q %q %q %q','Delimiter','\t','HeaderLines',1);
+fclose(fid);
+newGPR.ID     = complexAnnot{1};
+newGPR.GPR    = complexAnnot{3};
+newModel=changeGrRules(newModel,newGPR.ID,newGPR.GPR);
+
+% Delete unused genes (if any)
+newModel = deleteUnusedGenes(newModel);
+
+% Save model:
+saveYeastModel(newModel)
+cd modelCuration

code/modelTests/essentialGenes.m

@@ -1,4 +1,4 @@
-function [accurancy,tp,tn,fn,fp] = essentialGenes
+function [accuracy,tp,tn,fn,fp] = essentialGenes(model)
 % essentialGenes
 %   Modify media + find essential genes in model. Adapted from:
 %   https://doi.org/10.1371/journal.pcbi.1004530
@@ -12,11 +12,11 @@
 %   Usage: [accurancy,tp,tn,fn,fp] = essentialGenes
 %
 
-initCobraToolbox
-cd ..
-model = loadYeastModel;
-model = ravenCobraWrapper(model);
-cd modelTests
+if nargin<1;
+    cd ..
+    model = loadYeastModel;
+    cd modelTests
+end
 ko_tol = 1e-6;
 
 %constraints from genotype: check the genotype of the strains used in deletion experiment
@@ -41,7 +41,10 @@
 exp_viable = intersect(exp_viable,verifiedORFs);
 
 %calculate the growth rate after the single gene deletion using the original model and update model
-grRatio = singleGeneDeletion(model);
+[genes, fluxes, originalGenes, details, grRatioMuts]=findGeneDeletions(model,'sgd','fba');
+grRatio=ones(1,numel(model.genes));
+grRatio(genes)=grRatioMuts;
+
 mod_viable  = model.genes(grRatio >= ko_tol);
 mod_viable = intersect(mod_viable,verifiedORFs);
 mod_inviable = model.genes(grRatio < ko_tol );
@@ -59,7 +62,7 @@
 %compare the prediction performances of two models
 %prediction accuracy was used to evaluate the quality of model update in
 %each step
-accurancy = (n_tp+n_tn)/(n_tp+n_tn+n_fn+n_fp);
+accuracy = (n_tp+n_tn)/(n_tp+n_tn+n_fn+n_fp);
 sensitivity = (100*n_tp/(n_tp+n_fn));
 specificity = (100*n_tn/(n_tn+n_fp));
 positivePredictive = (100*n_tp/(n_tp+n_fp));
@@ -75,7 +78,7 @@
 
     % start with a clean slate: set all exchange reactions to upper bound =
     % 1000 and lower bound = 0 (ie, unconstrained excretion, no uptake)
-    exchangeRxns = findExcRxns(model);
+    [~, exchangeRxns] = getExchangeRxns(model);
     model.lb(exchangeRxns) = 0;
     model.ub(exchangeRxns) = 1000;
 
@@ -106,12 +109,9 @@
                     'r_4600'; ... % Ca(2+) exchange
                     'r_2020' };
 
-    constrainedUptakeRxnIndexes = findRxnIDs(model,constrainedUptake);
-    glucoseExchangeIndex = findRxnIDs(model,glucoseExchange);
-    unconstrainedUptakeRxnIndexes = findRxnIDs(model,unconstrainedUptake);
-    model.lb(constrainedUptakeRxnIndexes) = -0.5;
-    model.lb(glucoseExchangeIndex) = -20;
-    model.lb(unconstrainedUptakeRxnIndexes) = -1000;
+    model=setParam(model,'lb',constrainedUptake,-0.5);
+    model=setParam(model,'lb',glucoseExchange,-20);
+    model=setParam(model,'lb',unconstrainedUptake,-1000);
 end
 
 function genes = inviableORFs

code/modelTests/findDuplicatedRxns.m

@@ -12,13 +12,12 @@ function findDuplicatedRxns(model)
 for i = 1:length(model.rxns)-1
     for j = i+1:length(model.rxns)
         if isequal(model.S(:,i),model.S(:,j)) || isequal(model.S(:,i),-model.S(:,j))
-            printRxnFormula(model,model.rxns(i),true,true,true);
+            constructEquations(model,model.rxns(i));
             disp(['Name: ' model.rxnNames{i} ' - GPR: ' model.grRules{i} ' - LB=' num2str(model.lb(i)) ' - UB=' num2str(model.ub(i))])
-            printRxnFormula(model,model.rxns(j),true,true,true);
+            constructEquations(model,model.rxns(j));
             disp(['Name: ' model.rxnNames{j} ' - GPR: ' model.grRules{j} ' - LB=' num2str(model.lb(j)) ' - UB=' num2str(model.ub(j))])
             disp(" ")
         end
     end
 end
-
 end

data/modelCuration/complexAnnotation/complexAnnotation.tsv

@@ -0,0 +1,45 @@
+rxnID_yeast_model	genes_yeast_model	final_GPR	complex portal	SGD	uniprot	argument
+r_0005	YGR032W or YMR306W or (YLR342W and YCR034W) 	(YPR165W and YLR342W) or (YPR165W and YGR032W) or (YPR165W and YMR306W)	https://www.ebi.ac.uk/complexportal/complex/CPX-1812; https://www.ebi.ac.uk/complexportal/complex/CPX-1813; https://www.ebi.ac.uk/complexportal/complex/CPX-3028	https://www.yeastgenome.org/locus/S000003264; https://www.yeastgenome.org/locus/S000004923	https://www.uniprot.org/uniprot/P40989; https://www.uniprot.org/uniprot/Q04952	YGR032W or YMR306W are subunit.
+r_0021	YDR204W and YGL119W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	YDR204W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	https://www.ebi.ac.uk/complexportal/complex/CPX-1155			YGL119W is not a subunit of the complex
+r_0022	YDR204W and YGL119W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	YDR204W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	https://www.ebi.ac.uk/complexportal/complex/CPX-1155			YGL119W is not a subunit of the complex
+r_0263	(YHL003C and YMR298W) or (YKL008C and YMR298W) 	YKL008C and YHL003C and YMR298W	https://www.ebi.ac.uk/complexportal/complex/CPX-1706	https://www.yeastgenome.org/locus/S000001491; https://www.yeastgenome.org/locus/S000004913 https://www.yeastgenome.org/locus/S000000995	https://www.uniprot.org/uniprot/P28496; https://www.uniprot.org/uniprot/Q03579 https://www.uniprot.org/uniprot/P38703	These three subunits work as a whole
+r_0264	(YHL003C and YMR298W) or (YKL008C and YMR298W) 	YKL008C and YHL003C and YMR298W	https://www.ebi.ac.uk/complexportal/complex/CPX-1706	https://www.yeastgenome.org/locus/S000001491; https://www.yeastgenome.org/locus/S000004913 https://www.yeastgenome.org/locus/S000000995	https://www.uniprot.org/uniprot/P28496; https://www.uniprot.org/uniprot/Q03579 https://www.uniprot.org/uniprot/P38703	These three subunits work as a whole
+r_0265	(YHL003C and YMR298W) or (YKL008C and YMR298W) 	YKL008C and YHL003C and YMR298W	https://www.ebi.ac.uk/complexportal/complex/CPX-1706	https://www.yeastgenome.org/locus/S000001491; https://www.yeastgenome.org/locus/S000004913 https://www.yeastgenome.org/locus/S000000995	https://www.uniprot.org/uniprot/P28496; https://www.uniprot.org/uniprot/Q03579 https://www.uniprot.org/uniprot/P38703	These three subunits work as a whole
+r_0266	(YHL003C and YMR298W) or (YKL008C and YMR298W) 	YKL008C and YHL003C and YMR298W	https://www.ebi.ac.uk/complexportal/complex/CPX-1706	https://www.yeastgenome.org/locus/S000001491; https://www.yeastgenome.org/locus/S000004913 https://www.yeastgenome.org/locus/S000000995	https://www.uniprot.org/uniprot/P28496; https://www.uniprot.org/uniprot/Q03579 https://www.uniprot.org/uniprot/P38703	These three subunits work as a whole
+r_0532	YDR204W and YGL119W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	YDR204W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	https://www.ebi.ac.uk/complexportal/complex/CPX-1155			YGL119W is not a subunit of the complex
+r_0831	YDR148C and YFL018C and YIL125W	(YDR148C and YFL018C and YIL125W and YFR049W) or (YDR148C and YFL018C and YIL125W)	https://www.ebi.ac.uk/complexportal/complex/CPX-1293			"YDR148C, YFL018C, YIL125W and YFR049W could also catalyze the reaction"
+r_0832	YDR148C and YFL018C and YIL125W	(YDR148C and YFL018C and YIL125W and YFR049W) or (YDR148C and YFL018C and YIL125W)	https://www.ebi.ac.uk/complexportal/complex/CPX-1293			"YDR148C, YFL018C, YIL125W and YFR049W could also catalyze the reaction"
+r_0906	(YKL088W and YKR072C and YOR054C) or (YKL088W and YKR072C) or (YKL088W and YOR054C) or YKL088W	YKL088W and YKR072C and YOR054C	https://www.ebi.ac.uk/complexportal/complex/CPX-393	https://www.yeastgenome.org/locus/S000001571; https://www.yeastgenome.org/locus/S000001780	https://www.uniprot.org/uniprot/P36076; https://www.uniprot.org/uniprot/P36024	These three subunits work as a whole
+r_0943	YDR400W and YLR017W and YLR209C	YLR209C		https://www.yeastgenome.org/locus/S000004199	https://www.uniprot.org/uniprot/Q05788	YLR209C catalyze the reaction independently
+r_0963	YDR204W and YGL119W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	YDR204W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	https://www.ebi.ac.uk/complexportal/complex/CPX-1155			YGL119W is not a subunit of the complex
+r_0985	YDR204W and YGL119W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	YDR204W and YGR255C and YLR201C and YML110C and YOL096C and YOR125C	https://www.ebi.ac.uk/complexportal/complex/CPX-1155			YGL119W is not a subunit of the complex
+r_0992	(YDL040C and YGR147C and YHR013C) or (YDL040C and YGR147C)	YHR013C and YDL040C and YOR253W	https://www.ebi.ac.uk/complexportal/complex/CPX-783	https://www.yeastgenome.org/locus/S000003379		YGR147C is not a subunit of the complex
+r_0993	(YBR058C-A and YDR062W and YMR296C) or (YDR062W and YMR296C)	YMR296C and YKL212W and YDR062W and YGR038W and YLR350W and YBR058C-A	https://www.ebi.ac.uk/complexportal/complex/CPX-3158			
+r_2852	YBR097W and YLR240W	(YBR097W and YLR240W) or (YBR097W and YLR240W and YPL120W and YLR360W) or (YBR097W and YLR240W and YBR128C and YPL120W and YLR211C)	https://www.ebi.ac.uk/complexportal/complex/CPX-1881; https://www.ebi.ac.uk/complexportal/complex/CPX-1677			Another two complex could also catalyze the reaction
+r_2853	YBR097W and YLR240W	(YBR097W and YLR240W) or (YBR097W and YLR240W and YPL120W and YLR360W) or (YBR097W and YLR240W and YBR128C and YPL120W and YLR211C)	https://www.ebi.ac.uk/complexportal/complex/CPX-1881; https://www.ebi.ac.uk/complexportal/complex/CPX-1677			Another two complex could also catalyze the reaction
+r_2854	YBR097W and YLR240W	(YBR097W and YLR240W) or (YBR097W and YLR240W and YPL120W and YLR360W) or (YBR097W and YLR240W and YBR128C and YPL120W and YLR211C)	https://www.ebi.ac.uk/complexportal/complex/CPX-1881; https://www.ebi.ac.uk/complexportal/complex/CPX-1677			Another two complex could also catalyze the reaction
+r_2855	YBR097W and YLR240W	(YBR097W and YLR240W) or (YBR097W and YLR240W and YPL120W and YLR360W) or (YBR097W and YLR240W and YBR128C and YPL120W and YLR211C)	https://www.ebi.ac.uk/complexportal/complex/CPX-1881; https://www.ebi.ac.uk/complexportal/complex/CPX-1677			Another two complex could also catalyze the reaction
+r_2856	YBR097W and YLR240W	(YBR097W and YLR240W) or (YBR097W and YLR240W and YPL120W and YLR360W) or (YBR097W and YLR240W and YBR128C and YPL120W and YLR211C)	https://www.ebi.ac.uk/complexportal/complex/CPX-1881; https://www.ebi.ac.uk/complexportal/complex/CPX-1677			Another two complex could also catalyze the reaction
+r_2857	YBR097W and YLR240W	(YBR097W and YLR240W) or (YBR097W and YLR240W and YPL120W and YLR360W) or (YBR097W and YLR240W and YBR128C and YPL120W and YLR211C)	https://www.ebi.ac.uk/complexportal/complex/CPX-1881; https://www.ebi.ac.uk/complexportal/complex/CPX-1677			Another two complex could also catalyze the reaction
+r_2858	YBR097W and YLR240W	(YBR097W and YLR240W) or (YBR097W and YLR240W and YPL120W and YLR360W) or (YBR097W and YLR240W and YBR128C and YPL120W and YLR211C)	https://www.ebi.ac.uk/complexportal/complex/CPX-1881; https://www.ebi.ac.uk/complexportal/complex/CPX-1677			Another two complex could also catalyze the reaction
+r_2859	YBR097W and YLR240W	(YBR097W and YLR240W) or (YBR097W and YLR240W and YPL120W and YLR360W) or (YBR097W and YLR240W and YBR128C and YPL120W and YLR211C)	https://www.ebi.ac.uk/complexportal/complex/CPX-1881; https://www.ebi.ac.uk/complexportal/complex/CPX-1677			Another two complex could also catalyze the reaction
+r_3216	YLR386W and YNL325C	YFR019W and YNL325C and YFR021W and YNL054W and YLR386W	https://www.ebi.ac.uk/complexportal/complex/CPX-3088			"YFR019W, YFR021W and YNL054W are also necessary"
+r_3217	YLR386W and YNL325C	YFR019W and YNL325C and YFR021W and YNL054W and YLR386W	https://www.ebi.ac.uk/complexportal/complex/CPX-3088			"YFR019W, YFR021W and YNL054W are also necessary"
+r_3218	YLR386W and YNL325C	YFR019W and YNL325C and YFR021W and YNL054W and YLR386W	https://www.ebi.ac.uk/complexportal/complex/CPX-3088			"YFR019W, YFR021W and YNL054W are also necessary"
+r_3219	YLR386W and YNL325C	YFR019W and YNL325C and YFR021W and YNL054W and YLR386W	https://www.ebi.ac.uk/complexportal/complex/CPX-3088			"YFR019W, YFR021W and YNL054W are also necessary"
+r_3220	YLR386W and YNL325C	YFR019W and YNL325C and YFR021W and YNL054W and YLR386W	https://www.ebi.ac.uk/complexportal/complex/CPX-3088			"YFR019W, YFR021W and YNL054W are also necessary"
+r_3221	YLR386W and YNL325C	YFR019W and YNL325C and YFR021W and YNL054W and YLR386W	https://www.ebi.ac.uk/complexportal/complex/CPX-3088			"YFR019W, YFR021W and YNL054W are also necessary"
+r_3222	YLR386W and YNL325C	YFR019W and YNL325C and YFR021W and YNL054W and YLR386W	https://www.ebi.ac.uk/complexportal/complex/CPX-3088			"YFR019W, YFR021W and YNL054W are also necessary"
+r_3223	YLR386W and YNL325C	YFR019W and YNL325C and YFR021W and YNL054W and YLR386W	https://www.ebi.ac.uk/complexportal/complex/CPX-3088			"YFR019W, YFR021W and YNL054W are also necessary"
+r_1085	YBR127C and YDL185W and YEL027W and YEL051W and YGR020C and YHR026W and YHR039C-A and YKL080W and YLR447C and YMR054W and YOR332W and YPL234C and YPR036W	(YPR170W-B and YBR127C and YDL185W and YOR332W and YHR026W and YEL027W and YKL080W and YLR447C and YEL051W and YPL234C and YMR054W and YGR020C and YPR036W and YHR039C-A and YGR106C and YCL005W-A) or (YBR127C and YDL185W and YOR332W and YHR026W and YEL027W and YKL080W and YLR447C and YEL051W and YPL234C and YMR054W and YGR020C and YPR036W and YHR039C-A and YGR106C and YCL005W-A) 	https://www.ebi.ac.uk/complexportal/complex/CPX-1192			
+r_1086	YBR127C and YDL185W and YEL027W and YEL051W and YGR020C and YHR026W and YHR039C-A and YKL080W and YLR447C and YOR270C and YOR332W and YPL234C and YPR036W	(YBR127C and YDL185W and YEL027W and YEL051W and YGR020C and YHR026W and YHR039C-A and YKL080W and YLR447C and YOR270C and YOR332W and YPL234C and YPR036W and YPR170W-B and YGR106C and YCL005W-A) or (YBR127C and YDL185W and YEL027W and YEL051W and YGR020C and YHR026W and YHR039C-A and YKL080W and YLR447C and YOR270C and YOR332W and YPL234C and YPR036W  and YGR106C and YCL005W-A)	https://www.ebi.ac.uk/complexportal/complex/CPX-1193			
+r_0886	YMR205C or (YGR240C and YMR205C)	YGR240C and YMR205C	https://www.ebi.ac.uk/complexportal/complex/CPX-554	https://www.yeastgenome.org/locus/S000004818	https://www.uniprot.org/uniprot/P16862	YMR205C is a subunit
+r_0887	YMR205C or (YGR240C and YMR205C)	YGR240C and YMR205C	https://www.ebi.ac.uk/complexportal/complex/CPX-554	https://www.yeastgenome.org/locus/S000004818	https://www.uniprot.org/uniprot/P16862	YMR205C is a subunit
+r_2836	YDR373W and YNL267W	YDR373W and YNL267W				keep the original grRules
+r_2837	YDR373W and YNL267W	YDR373W and YNL267W				keep the original grRules
+r_2838	YDR373W and YNL267W	YDR373W and YNL267W				keep the original grRules
+r_2839	YDR373W and YNL267W	YDR373W and YNL267W				keep the original grRules
+r_2840	YDR373W and YNL267W	YDR373W and YNL267W				keep the original grRules
+r_2841	YDR373W and YNL267W	YDR373W and YNL267W				keep the original grRules
+r_2842	YDR373W and YNL267W	YDR373W and YNL267W				keep the original grRules
+r_2843	YDR373W and YNL267W	YDR373W and YNL267W				keep the original grRules

data/modelCuration/complexAnnotation/complexAnnotationGenes.tsv

@@ -0,0 +1,14 @@
+ 	geneShortName	ncbigene	refseq	uniprot	kegg.genes	ecogene	hprd	asap	ccds	ncbiprotein
+YPR165W	RHO1	856294	NM_001184262	P06780	sce:YPR165W					NP_015491
+YFR049W	YMR31	850610	NM_001180014	P19955	sce:YFR049W					NP_116707
+YOR253W	NAT5	854427	NM_001183672	Q08689	sce:YOR253W	2.3.1.258				NP_014896 
+YGR038W	ORM1	852926	NM_001181167	P53224	sce:YGR038W					NP_011552 
+YLR350W	ORM2	851064	NM_001182239	Q06144	sce:YLR350W					NP_013454 
+YBR128C	ATG14	852425	NM_001178476	P38270	sce:YBR128C					NP_009686 
+YLR211C	ATG38	850908	NM_001182098	Q05789	sce:YLR211C					NP_013312   
+YLR360W	VPS38	851074	NM_001182249	Q05919	sce:YLR360W					NP_013464
+YPL120W	VPS30	855983	NM_001183934	Q02948	sce:YPL120W					NP_015205
+YFR021W	ATG18	850577	NM_001179986	P43601	sce:YFR021W					NP_444297
+YNL054W	VAC7	855673	NM_001182893	P53950	sce:YNL054W					NP_014344
+YGR106C	VOA1	852998	NM_001181235	P53262	sce:YGR106C					NP_011620
+YPR170W-B		5848745	NM_001184525	P0C5R9	sce:YPR170W-B					NP_001106949