Description of getRxnsFromKEGG

0001 function [model,isSpontaneous,isUndefinedStoich,isIncomplete,...
0002     isGeneral]=getRxnsFromKEGG(keggPath)
0003 % getRxnsFromKEGG
0004 %   Retrieves information on all reactions stored in KEGG database
0005 %
0006 %   Input:
0007 %   keggPath            if keggRxns.mat is not in the RAVEN\external\kegg
0008 %                       directory, this function will attempt to read data
0009 %                       from a local FTP dump of the KEGG database.
0010 %                       keggPath is the path to the root of this database
0011 %
0012 %   Output:
0013 %   model               a model structure generated from the database. The
0014 %                       following fields are filled
0015 %       id                  'KEGG'
0016 %       name         'Automatically generated from KEGG database'
0017 %       rxns                KEGG reaction ids
0018 %       rxnNames            Name for each reaction entry
0019 %       mets                KEGG compound ids. If the equations use
0020 %                           stoichiometry such as ID(n+1) then the whole
0021 %                           expression is saved as the id
0022 %       eccodes             Corresponding ec-number if available
0023 %       rxnMiriams          Contains reaction specific information such as
0024 %                           KO id and pathways that the reaction is
0025 %                           associated to
0026 %       S                   Stoichiometric matrix
0027 %       lb                  -1000 for all reactions
0028 %       ub                  1000 for all reactions
0029 %       rev                 1 for reversible and 0 for irreversible. For
0030 %                           reactions present in pathway maps the
0031 %                           reversibility is taken from there
0032 %       b                   0 for all metabolites
0033 %   isSpontaneous       a cell array with the reactions labelled as
0034 %                       "spontaneous"
0035 %   isUndefinedStoich    a cell array with the reactions labelled as with
0036 %                       undefined stoichiometry
0037 %   isIncomplete        a cell array with the reactions labelled as
0038 %                       "incomplete", "erroneous" or "unclear"
0039 %   isGeneral           a cell array with the reactions labelled as
0040 %                       "general reaction"
0041 %
0042 %   NOTE: Reactions on the form A <=> A + B will not be loaded. If the file
0043 %   keggRxns.mat is in the RAVEN/external/kegg directory it will be loaded
0044 %   instead of parsing of the KEGG files. If it does not exist it will be
0045 %   saved after parsing of the KEGG files. In general, you should remove
0046 %   the keggRxns.mat file along with other KEGG mat files if you want to
0047 %   rebuild the model structure from a newer version of KEGG.
0048 %
0049 %   Usage: [model,isSpontaneous,isUndefinedStoich,isIncomplete,...
0050 %    isGeneral]=getRxnsFromKEGG(keggPath)
0051 %
0052 % NOTE: This is how one entry looks in the file
0053 %
0054 % ENTRY       R00010                      Reaction
0055 % NAME        alpha,alpha-trehalose glucohydrolase
0056 % DEFINITION  alpha,alpha-Trehalose + H2O <=> 2 D-Glucose
0057 % EQUATION    C01083 + C00001 <=> 2 C00031
0058 % REMARK      Same as: R06103
0059 % RCLASS      RC00049  C00031_C01083
0060 % ENZYME      3.2.1.28
0061 % PATHWAY     rn00500  Starch and sucrose metabolism
0062 %             rn01100  Metabolic pathways
0063 % ORTHOLOGY   K01194  alpha,alpha-trehalase [EC:3.2.1.28]
0064 % DBLINKS     RHEA: 32678
0065 % ///
0066 %
0067 % The file is not tab-delimited. Instead each label is 12 characters
0068 % (except for '///')
0069 %
0070 
0071 if nargin<1
0072     keggPath='RAVEN/external/kegg';
0073 else
0074     keggPath=char(keggPath);
0075 end
0076 
0077 %Check if the reactions have been parsed before and saved. If so, load the
0078 %model
0079 ravenPath=findRAVENroot();
0080 rxnsFile=fullfile(ravenPath,'external','kegg','keggRxns.mat');
0081 if exist(rxnsFile, 'file')
0082     fprintf(['Importing KEGG reactions from ' strrep(rxnsFile,'\','/') '... ']);
0083     load(rxnsFile);
0084 else
0085     fprintf(['NOTE: Cannot locate ' strrep(rxnsFile,'\','/') ', it will therefore be generated from the local KEGG database\n']);
0086     if ~isfile(fullfile(keggPath,'reaction')) || ~isfile(fullfile(keggPath,'reaction.lst')) || ~isfile(fullfile(keggPath,'reaction_mapformula.lst'))
0087         EM=fprintf(['The files ''reaction'', ''reaction.lst'' and ''reaction_mapformula.lst'' cannot be located at ' strrep(keggPath,'\','/') '/ and should be downloaded from the KEGG FTP\n']);
0088         dispEM(EM);
0089     else
0090         fprintf('Generating keggRxns.mat file... ');
0091         %Add new functionality in the order specified in models
0092         model.id='KEGG';
0093         model.name='Automatically generated from KEGG database';
0094         
0095         %Preallocate memory for 15000 reactions
0096         model.rxns=cell(15000,1);
0097         model.rxnNames=cell(15000,1);
0098         model.eccodes=cell(15000,1);
0099         model.subSystems=cell(15000,1);
0100         model.rxnMiriams=cell(15000,1);
0101         model.rxnNotes=cell(15000,1);
0102         equations=cell(15000,1);
0103         %Temporarily store the equations
0104         
0105         isSpontaneous=false(15000,1);
0106         isIncomplete=false(15000,1);
0107         isGeneral=false(15000,1);
0108 
0109         %First load information on reaction ID, reaction name, KO, pathway,
0110         %and ec-number
0111         fid = fopen(fullfile(keggPath,'reaction'), 'r');
0112         
0113         %Keep track of how many reactions have been added
0114         rxnCounter=0;
0115         
0116         %Loop through the file
0117         orthology=false;
0118         pathway=false;
0119         module=false;
0120         while 1
0121             %Get the next line
0122             tline = fgetl(fid);
0123             
0124             %Abort at end of file
0125             if ~ischar(tline)
0126                 break;
0127             end
0128             
0129             %Skip '///'
0130             if numel(tline)<12
0131                 continue;
0132             end
0133             
0134             %Check if it's a new reaction
0135             if strcmp(tline(1:12),'ENTRY       ')
0136                 rxnCounter=rxnCounter+1;
0137                 
0138                 %Add empty strings where there should be such
0139                 model.rxnNames{rxnCounter}='';
0140                 model.eccodes{rxnCounter}='';
0141                 %model.subSystems{rxnCounter}=''; %remain empty cell
0142                 model.rxnNotes{rxnCounter}='';
0143                 equations{rxnCounter}='';
0144                 
0145                 %Add reaction ID (always 6 characters)
0146                 model.rxns{rxnCounter}=tline(13:18);
0147                 orthology=false;
0148                 pathway=false;
0149                 module=false;
0150                 
0151                 %Add KEGG reaction ID miriam
0152                 tempStruct=model.rxnMiriams{rxnCounter};
0153                 tempStruct.name{1,1}='kegg.reaction';
0154                 tempStruct.value{1,1}=tline(13:18);
0155                 model.rxnMiriams{rxnCounter}=tempStruct;
0156             end
0157             
0158             %Add name
0159             if strcmp(tline(1:12),'NAME        ')
0160                 model.rxnNames{rxnCounter}=tline(13:end);
0161             end
0162             
0163             %Add whether the comment includes "incomplete", "erroneous" or
0164             %"unclear"
0165             if strcmp(tline(1:12),'COMMENT     ')
0166                 %Read all text until '///', 'RPAIR', 'ENZYME', 'PATHWAY' or 'RCLASS'
0167                 commentText=tline(13:end);
0168                 while 1
0169                     tline = fgetl(fid);
0170                     if ~strcmp(tline(1:3),'///') && ~strcmp(tline(1:3),'RPA') && ~strcmp(tline(1:3),'ENZ') && ~strcmp(tline(1:3),'PAT') && ~strcmp(tline(1:3),'RCL')
0171                         commentText=[commentText ' ' strtrim(tline)];
0172                     else
0173                         break;
0174                     end
0175                 end
0176                 if any(regexpi(commentText,'SPONTANEOUS'))
0177                     %It should start this way
0178                     isSpontaneous(rxnCounter)=true;
0179                 end
0180                 if any(regexpi(commentText,'INCOMPLETE')) || any(regexpi(commentText,'ERRONEOUS')) || any(regexpi(commentText,'UNCLEAR'))
0181                     isIncomplete(rxnCounter)=true;
0182                 end
0183                 if any(regexpi(commentText,'GENERAL REACTION'))
0184                     %It should start this way
0185                     isGeneral(rxnCounter)=true;
0186                 end
0187                 
0188                 %Go to next iteration if it is '///'
0189                 if numel(tline)<12
0190                     continue;
0191                 end
0192             end
0193             
0194             %Add ec-number
0195             if strcmp(tline(1:12),'ENZYME      ')
0196                 model.eccodes{rxnCounter}=tline(13:end);
0197                 model.eccodes{rxnCounter}=deblank(model.eccodes{rxnCounter});
0198                 model.eccodes{rxnCounter}=regexprep(model.eccodes{rxnCounter},'\s+',';');
0199             end
0200             if numel(tline)>8
0201                 if strcmp(tline(1:9),'REFERENCE')
0202                     pathway=false;
0203                     orthology=false;
0204                 end
0205             end
0206             
0207             %Add module ids
0208             if numel(tline)>18
0209                 if strcmp(tline(1:12),'MODULE      ') || module==true
0210                     pathway=false;
0211                     orthology=false;
0212                     if isstruct(model.rxnMiriams{rxnCounter})
0213                         addToIndex=numel(model.rxnMiriams{rxnCounter}.name)+1;
0214                     else
0215                         addToIndex=1;
0216                     end
0217                     tempStruct=model.rxnMiriams{rxnCounter};
0218                     tempStruct.name{addToIndex,1}='kegg.module';
0219                     tempStruct.value{addToIndex,1}=tline(13:18);
0220                     model.rxnMiriams{rxnCounter}=tempStruct;
0221                 end
0222             end
0223             
0224             %Add RHEA id
0225             if numel(tline)>18
0226                 if strcmp(tline(1:18),'DBLINKS     RHEA: ')
0227                     pathway=false;
0228                     orthology=false;
0229                     module=false;
0230                     if isstruct(model.rxnMiriams{rxnCounter})
0231                         addToIndex=numel(model.rxnMiriams{rxnCounter}.name)+1;
0232                     else
0233                         addToIndex=1;
0234                     end
0235                     tempStruct=model.rxnMiriams{rxnCounter};
0236                     tempStruct.name{addToIndex,1}='rhea';
0237                     tempStruct.value{addToIndex,1}=tline(19:end);
0238                     model.rxnMiriams{rxnCounter}=tempStruct;
0239                 end
0240             end
0241             
0242             %Add KO-ids
0243             if numel(tline)>16
0244                 if strcmp(tline(1:16),'ORTHOLOGY   KO: ') || strcmp(tline(1:16),'            KO: ') || strcmp(tline(1:12),'ORTHOLOGY   ') || orthology==true
0245                     pathway=false;
0246                     module=false;
0247                     %Check if KO has been added already (each reaction may
0248                     %belong to several)
0249                     if isstruct(model.rxnMiriams{rxnCounter})
0250                         addToIndex=numel(model.rxnMiriams{rxnCounter}.name)+1;
0251                     else
0252                         addToIndex=1;
0253                     end
0254                     
0255                     tempStruct=model.rxnMiriams{rxnCounter};
0256                     tempStruct.name{addToIndex,1}='kegg.orthology';
0257                     if strcmp(tline(13:16),'KO:')
0258                         %This is in the old version
0259                         tempStruct.value{addToIndex,1}=tline(17:22);
0260                     else
0261                         %This means that it found one KO in the new format
0262                         %and that subsequent lines might be other KOs
0263                         orthology=true;
0264                         tempStruct.value{addToIndex,1}=tline(13:18);
0265                     end
0266                     model.rxnMiriams{rxnCounter}=tempStruct;
0267                 end
0268             end
0269             
0270             %Add pathways
0271             if numel(tline)>18
0272                 if strcmp(tline(1:18),'PATHWAY     PATH: ') || strcmp(tline(1:18),'            PATH: ') || strcmp(tline(1:12),'PATHWAY     ') || pathway==true
0273                     orthology=false;
0274                     module=false;
0275                     %Check if annotation has been added already
0276                     if isstruct(model.rxnMiriams{rxnCounter})
0277                         addToIndex=numel(model.rxnMiriams{rxnCounter}.name)+1;
0278                     else
0279                         addToIndex=1;
0280                     end
0281                     
0282                     tempStruct=model.rxnMiriams{rxnCounter};
0283                     tempStruct.name{addToIndex,1}='kegg.pathway';
0284                     %If it is the old version
0285                     if strcmp(tline(14:17),'PATH:')
0286                         tempStruct.value{addToIndex,1}=tline(19:25);
0287                     else
0288                         %If it is the new version
0289                         tempStruct.value{addToIndex,1}=tline(13:19);
0290                         pathway=true;
0291                     end
0292                     
0293                     %Do not save global or overview pathways. The ids for
0294                     %such pathways begin with rn011 or rn012
0295                     if ~strcmp('rn011',tempStruct.value{addToIndex,1}(1:5)) && ~strcmp('rn012',tempStruct.value{addToIndex,1}(1:5))
0296                         model.rxnMiriams{rxnCounter}=tempStruct;
0297                         
0298                         %Also save the subSystems names. For the old KEGG
0299                         %format, only the first mentioned subsystem is
0300                         %picked. Use the newer KEGG format to fetch all the
0301                         %subsystems
0302                         if strcmp(tline(14:17),'PATH:')
0303                             %The old format
0304                             model.subSystems{rxnCounter}=tline(28:end);
0305                         else
0306                             %The new format
0307                             model.subSystems{rxnCounter,1}{1,numel(model.subSystems{rxnCounter,1})+1}=tline(22:end);
0308                         end
0309                     end
0310                 end
0311             end
0312         end
0313         
0314         %Close the file
0315         fclose(fid);
0316         
0317         %This is done here since the the indexes won't match since some
0318         %reactions are removed along the way
0319         isIncomplete=model.rxns(isIncomplete);
0320         isGeneral=model.rxns(isGeneral);
0321         isSpontaneous=model.rxns(isSpontaneous);
0322 
0323         %If too much space was allocated, shrink the model
0324         model.rxns=model.rxns(1:rxnCounter);
0325         model.rxnNames=model.rxnNames(1:rxnCounter);
0326         model.eccodes=model.eccodes(1:rxnCounter);
0327         equations=equations(1:rxnCounter);
0328         model.rxnMiriams=model.rxnMiriams(1:rxnCounter);
0329         model.rxnNotes=model.rxnNotes(1:rxnCounter);
0330         model.subSystems=model.subSystems(1:rxnCounter);
0331         
0332         %Then load the equations from another file. This is because the
0333         %equations are easier to retrieve from there
0334         
0335         %The format is rxnID: equation The reactions should have been
0336         %loaded in the exact same order
0337         fid = fopen(fullfile(keggPath,'reaction.lst'), 'r');
0338         
0339         %Loop through the file
0340         for i=1:rxnCounter
0341             %Get the next line
0342             tline = fgetl(fid);
0343             
0344             equations{i}=tline(9:end);
0345         end
0346         
0347         %Close the file
0348         fclose(fid);
0349         
0350         %Several equations may have two whitespaces between the last
0351         %reactant and the reversible arrow sign. The number of whitespaces
0352         %is thus reduced to one
0353         equations = regexprep(equations,'  <=>', ' <=>');
0354         
0355         %Construct the S matrix and list of metabolites
0356         [S, mets, badRxns]=constructS(equations);
0357         model.S=S;
0358         model.mets=mets;
0359         
0360         %There is some limited evidence for directionality in
0361         %reaction_mapformula.lst. The information there concerns how the
0362         %reactions are drawn in the KEGG maps. If a reaction is
0363         %irreversible in the same direction for all maps, then I consider
0364         %is irreversible, otherwise reversible. Also, not all reactions are
0365         %present in the maps, so not all will have directionality. They
0366         %will be considered to be reversible
0367         
0368         %The format is R00005: 00330: C01010 => C00011 Generate a
0369         %reversibility structure with the fields: *rxns: reaction ids
0370         %*product: one met id that is a product. This is because the
0371         %*reactions might be written in another direction compared to in
0372         % the reactions.lst file
0373         %*rev: 1 if reversible, otherwise 0
0374         reversibility.rxns={};
0375         reversibility.product={};
0376         reversibility.rev=[];
0377         
0378         fid = fopen(fullfile(keggPath,'reaction_mapformula.lst'), 'r');
0379         while 1
0380             %Get the next line
0381             tline = fgetl(fid);
0382             
0383             %Abort at end of file
0384             if ~ischar(tline)
0385                 break;
0386             end
0387             
0388             rxn=tline(1:6);
0389             prod=tline(end-5:end);
0390             rev=any(strfind(tline,'<=>'));
0391             if isempty(reversibility.rxns)
0392                 reversibility.rxns{1}=rxn;
0393                 reversibility.product{1}=prod;
0394                 reversibility.rev(1)=rev;
0395             elseif strcmp(reversibility.rxns(end),rxn)
0396                 %Check if the reaction was added before. It's an ordered
0397                 %list, so only check the last element If it's reversible in
0398                 %the new reaction or reversible in the old reaction then
0399                 %set (keep) to be reversible
0400                 if rev==1 || reversibility.rev(end)==1
0401                     reversibility.rev(end)=1;
0402                 else
0403                     %This means that the reaction was already loaded, that
0404                     %it was irreversible before and irreversible in the new
0405                     %reaction. However, it could be that they are written
0406                     %in diferent directions. If the product differ, then
0407                     %set to be reversible. This assumes that the reactions
0408                     %are written with the same metabolite as the last one
0409                     %if they are in the same direction
0410                     if ~strcmp(prod,reversibility.product(end))
0411                         reversibility.rev(end)=1;
0412                     end
0413                 end
0414             else
0415                 reversibility.rxns=[reversibility.rxns;rxn];
0416                 reversibility.product=[reversibility.product;prod];
0417                 reversibility.rev=[reversibility.rev;rev];
0418             end
0419         end
0420         fclose(fid);
0421         
0422         %Update the reversibility
0423         model.rev=ones(rxnCounter,1);
0424         %Match the reaction ids
0425         irrevIDs=find(reversibility.rev==0);
0426         [~, I]=ismember(reversibility.rxns(irrevIDs),model.rxns);
0427         [~, prodMetIDs]=ismember(reversibility.product(irrevIDs),model.mets);
0428         model.rev(I)=0;
0429         
0430         %See if the reactions are written in the same order in model.S
0431         linearInd=sub2ind(size(model.S), prodMetIDs, I);
0432         changeOrder=I(model.S(linearInd)<0);
0433         %Change the order of these reactions
0434         model.S(:,changeOrder)=model.S(:,changeOrder).*-1;
0435         
0436         %Add some stuff to get a correct model structure
0437         model.ub=ones(rxnCounter,1)*1000;
0438         model.lb=model.rev*-1000;
0439         model.c=zeros(rxnCounter,1);
0440         model.b=zeros(numel(model.mets),1);
0441         model=removeReactions(model,badRxns,true,true);
0442         
0443         %Identify reactions with undefined stoichiometry. Such
0444         %reactions involve metabolites with an ID containing the letter "n"
0445         %or "m"
0446         I=cellfun(@any,strfind(model.mets,'n')) | cellfun(@any,strfind(model.mets,'m'));
0447         [~, J]=find(model.S(I,:));
0448         isUndefinedStoich=model.rxns(unique(J));   
0449         %Sort model that metabolites with undefined stoichiometry would
0450         %appear in the end of metabolites list
0451         metList=[model.mets(~I);model.mets(I)];
0452         [~,metIndexes]=ismember(metList,model.mets);
0453         model=permuteModel(model,metIndexes,'mets');
0454         
0455         %Sort model that i) spontaneous, ii) with undefined
0456         %stoichiometry, iii) incomplete and iv) general reactions would bve
0457         %ranked in the end of the model
0458         endRxnList=unique([model.rxns(ismember(model.rxns,isSpontaneous));model.rxns(ismember(model.rxns,isUndefinedStoich));model.rxns(ismember(model.rxns,isIncomplete));model.rxns(ismember(model.rxns,isGeneral))],'stable');
0459         rxnList=[model.rxns(~ismember(model.rxns,endRxnList));endRxnList];
0460         [~,rxnIndexes]=ismember(rxnList,model.rxns);
0461         model=permuteModel(model,rxnIndexes,'rxns');
0462         
0463         %Add information in rxnNotes, whether reaction belongs to any of
0464         %type i-iv mentioned a few lines above
0465         for i=(numel(rxnList)-numel(endRxnList)+1):numel(model.rxns)
0466             if ismember(model.rxns(i),isSpontaneous)
0467                 model.rxnNotes(i)=strcat(model.rxnNotes(i),'Spontaneous');
0468             end
0469             if ismember(model.rxns(i),isUndefinedStoich)
0470                 if isempty(model.rxnNotes{i})
0471                     model.rxnNotes(i)=strcat(model.rxnNotes(i),'With undefined stoichiometry');
0472                 else
0473                     model.rxnNotes(i)=strcat(model.rxnNotes(i),', with undefined stoichiometry');
0474                 end
0475             end
0476             if ismember(model.rxns(i),isIncomplete)
0477                 if isempty(model.rxnNotes{i})
0478                     model.rxnNotes(i)=strcat(model.rxnNotes(i),'Incomplete');
0479                 else
0480                     model.rxnNotes(i)=strcat(model.rxnNotes(i),', incomplete');
0481                 end
0482             end
0483             if ismember(model.rxns(i),isGeneral)
0484                 if isempty(model.rxnNotes{i})
0485                     model.rxnNotes(i)=strcat(model.rxnNotes(i),'General');
0486                 else
0487                     model.rxnNotes(i)=strcat(model.rxnNotes(i),', general');
0488                 end
0489             end
0490             model.rxnNotes(i)=strcat(model.rxnNotes(i),' reaction');
0491         end
0492         
0493         %Save the model structure
0494         save(rxnsFile,'model','isGeneral','isIncomplete','isUndefinedStoich','isSpontaneous');
0495     end
0496 end
0497 fprintf('COMPLETE\n');
0498 
0499 end
getRxnsFromKEGG

PURPOSE

SYNOPSIS

DESCRIPTION

CROSS-REFERENCE INFORMATION

SOURCE CODE
