Merge pull request #14 from SysBioChalmers/devel

Pre-release,  v0.1

.gitignore

@@ -1,3 +1,23 @@
-# OS generated files #
-######################
-.DS_Store
+# Windows default autosave extension
+*.asv
+Thumbs.db
+
+# OSX / *nix default autosave extension
+*.*~
+
+# Compiled MEX binaries (all platforms)
+*.mex*
+
+# OS hidden extensions
+.DS_Store
+
+# specific extensions
+*.pyc
+*.mps
+
+# Windows executable files
+*.exe
+*.exe.*
+
+# Windows dynamic library files
+*.dll

ComplementaryData/BiGG/getMetsFromBiGG.m

@@ -0,0 +1,58 @@
+%
+% FILE NAME: getMetsFromBiGG.m
+% 
+% DATE CREATED: 2018-04-24
+% 
+% 
+% PROGRAMMER: Hao Wang
+% Department of Biology and Biological Engineering
+% Chalmers University of Technology
+% 
+% 
+% PURPOSE: Generate data structure for BiGG metabolites
+% 
+
+% Move to the target folder
+cd('/Users/haowa/Box Sync/HMR3/BiGG');
+
+% Load the text format BiGG reactions, after fixing errors with M01870
+T=readtable('bigg_models_metabolites_20180424.txt','ReadVariableNames',1);
+BiGGMets=table2struct(T,'ToScalar',true);
+
+% Rename the fields according to RAVEN specification
+BiGGMets.mets=BiGGMets.bigg_id;
+BiGGMets.metNames=BiGGMets.name;
+BiGGMets.universalID=BiGGMets.universal_bigg_id;
+
+% Add oldids field as cell array based on info from old_bigg_ids
+num=numel(BiGGMets.bigg_id);
+BiGGMets.oldids=cell(num,1);
+BiGGMets.oldids(:)={''};
+% Add MNX field
+metMNXID=regexp(BiGGMets.database_links,'MNXM\d+','match');
+BiGGMets.metMNXID=cell(num,1);
+BiGGMets.metMNXID(:)={''};
+count=0;
+for i=1:num
+ if ~isempty(BiGGMets.old_bigg_ids{i})
+ % Remove existing BiGG ids and universal ids from oldids
+ BiGGMets.oldids{i}=setdiff(transpose(strsplit(BiGGMets.old_bigg_ids{i},'; ')),BiGGMets.mets{i});
+ BiGGMets.oldids{i}=setdiff(BiGGMets.oldids{i},BiGGMets.universalID{i});
+ end
+ if ~isempty(metMNXID{i})
+ if numel(metMNXID{i})==1
+ BiGGMets.metMNXID{i}=metMNXID{i}{1};
+ else
+ BiGGMets.metMNXID{i}=metMNXID{i};
+ count=count+1;
+ end
+ end
+end
+%count=0
+numel(find(~cellfun(@isempty,BiGGMets.metMNXID))) %ans = 9915
+
+% Remove some fields
+BiGGMets=rmfield(BiGGMets,{'bigg_id','name','universal_bigg_id','old_bigg_ids'});
+
+save('BiGGMets.mat','BiGGMets');
+

ComplementaryData/BiGG/getRxnsFromBiGG.m

@@ -0,0 +1,59 @@
+%
+% FILE NAME: getRxnsFromBiGG.m
+% 
+% DATE CREATED: 2018-01-17
+% MODIFIED: 2018-04-24
+% 
+% 
+% PROGRAMMER: Hao Wang
+% Department of Biology and Biological Engineering
+% Chalmers University of Technology
+% 
+% 
+% PURPOSE: Generate Matlab structure for BiGG reactions
+% 
+
+% Move to the target folder
+cd('/Users/haowa/Box Sync/HMR3/BiGG');
+
+% Load the text format BiGG reactions
+T=readtable('bigg_models_reactions_20180424.txt','ReadVariableNames',1);
+BiGGRxns=table2struct(T,'ToScalar',true);
+
+% Rename the fields according to RAVEN specification
+BiGGRxns.rxns=BiGGRxns.bigg_id;
+BiGGRxns.rxnNames=BiGGRxns.name;
+BiGGRxns.rxnEquations=BiGGRxns.reaction_string;
+
+% Add oldids field as cell array based on info from old_bigg_ids
+num=numel(BiGGRxns.bigg_id);
+BiGGRxns.oldids=cell(num,1);
+BiGGRxns.oldids(:)={''};
+% Add MNX field
+rxnMNXID=regexp(BiGGRxns.database_links,'MNXR\d+','match');
+BiGGRxns.rxnMNXID=cell(num,1);
+BiGGRxns.rxnMNXID(:)={''};
+count=0;
+for i=1:num
+ % Ignore old_bigg_ids identical to bigg_id
+ if ~isequal(BiGGRxns.rxns{i},BiGGRxns.old_bigg_ids{i})
+ % Remove existing BiGG ids from oldids
+ BiGGRxns.oldids{i}=setdiff(transpose(strsplit(BiGGRxns.old_bigg_ids{i},'; ')),BiGGRxns.rxns{i});
+ end
+ if ~isempty(rxnMNXID{i})
+ if numel(rxnMNXID{i})==1
+ BiGGRxns.rxnMNXID{i}=rxnMNXID{i}{1};
+ else
+ BiGGRxns.rxnMNXID{i}=rxnMNXID{i};
+ count=count+1;
+ end
+ end
+end
+%count=0
+numel(find(~cellfun(@isempty,BiGGRxns.rxnMNXID))) %ans = 15904
+
+% Remove some fields
+BiGGRxns=rmfield(BiGGRxns,{'bigg_id','name','reaction_string','old_bigg_ids'});
+
+save('BiGGRxns.mat','BiGGRxns');
+

ComplementaryData/Ensembl/assocEnsemblGenes.m

@@ -0,0 +1,46 @@
+%
+% FILE NAME: assocEnsemblGenes.m
+% 
+% DATE CREATED: 2018-05-03
+% UPDATED: 2018-05-31
+% 
+% PROGRAMMER: Hao Wang
+% Department of Biology and Biological Engineering
+% Chalmers University of Technology
+% 
+% 
+% PURPOSE: Get exteranl associations of Ensembl genes
+% 
+
+% Move to Ensembl folder
+cd('/Users/haowa/Box Sync/HMR3/Ensembl');
+
+T=readtable('stableGeneID2NCBIgeneID.txt','ReadVariableNames',1);
+T=table2struct(T,'ToScalar',true);
+
+% Format NCBI Gene IDs
+NCBIGeneID=num2cell(T.NCBIGeneID); % convert vector to cell array
+NCBIGeneID=cellfun(@num2str,NCBIGeneID,'UniformOutput',false); % convert each element from double to string
+% Clear elements without NCBI Gene ID associations
+NaNidx=find(strcmp('NaN',NCBIGeneID));
+NCBIGeneID(NaNidx)={''};
+
+% Generate data structure for NCBIGeneID 
+Ensembl2NCBI.genes=T.GeneStableID;
+Ensembl2NCBI.NCBIGeneID=NCBIGeneID;
+
+% Manual curation 2018-06-13
+% remove repetitive associaiton "ENSG00000174876-278" based on NCBI
+Ensembl2NCBI.genes(34825)=[];
+Ensembl2NCBI.NCBIGeneID(34825)=[];
+save('Ensembl2NCBI.mat','Ensembl2NCBI');
+
+T=readtable('stableGeneID2UniprotID.txt','ReadVariableNames',1);
+T=table2struct(T,'ToScalar',true);
+
+% Generate data structure for UniprotID 
+Ensembl2Uniprot.genes=T.GeneStableID;
+Ensembl2Uniprot.ID=;
+Ensembl2Uniprot.ID=;
+save('Ensembl2Uniprot.mat','Ensembl2Uniprot');
+

ComplementaryData/HMR2/HMR2Curation.m

@@ -0,0 +1,53 @@
+%
+% FILE NAME: HMR2Curation.m
+% 
+% DATE CREATED: 2018-01-16
+% 
+% 
+% PROGRAMMER: Hao Wang
+% Department of Biology and Biological Engineering
+% Chalmers University of Technology
+% 
+% 
+% PURPOSE: Curating HMR2 database model toward HMR3
+%
+
+% Load the original Matlab file of HMR2 database
+load('HMRdatabase2_00.mat');
+
+% Import HMR2 reactions from the Excel file 'RXNS' sheet that leave out
+% the following five rows that represent reaction classes
+% 7709: 'Exchange reactions';
+% 8171: 'Fake reactions';
+% 8176: 'Biomass reactions';
+% 8177: 'HMR_biomass_Renalcancer';
+% 8183: 'Included for connectivity for INIT'
+T=readtable('HMRdatabase2_00.xlsx','Sheet','RXNS','ReadVariableNames',1);
+HMR2=table2struct(T,'ToScalar',true);
+
+% Adding reaction identifiers from external databases
+% additional empty spaces were also removed druing this process
+if isequal(ihuman.rxns,HMR2.RXNID) 
+ ihuman.rxnKEGGID=HMR2.KEGGID; %KEGG
+ ihuman.rxnEHMNID=HMR2.EHMNID; %EHMN
+ ihuman.rxnBiGGID=HMR2.BIGGDATABASEID; %Recon
+ ihuman.rxnHepatoNET1ID=HMR2.HEPATONET1ID; %Hepatonet1
+ ihuman.rxnREACTOMEID=HMR2.REACTOMEID; %Reactome
+ ihuman.rxnReferences=HMR2.REFERENCES; %References
+end
+
+% Some errors were spotted and fixed:
+% The reaction HMR_2190 was assocated to four REACTOM reactions.
+% The first one (REACT_22270) is correect, the other three
+% (REACT_22097; REACT_22133; REACT_22219) should be wrong because
+% they are empty ids by searching reactom.org, and thus removed
+index=find(strcmp('HMR_2190',ihuman.rxns));
+ihuman.rxnREACTOMEID{index}='REACT_22270';
+
+% The KEGG id of HMR_7709 was associated to 'R0302' that should be typo
+% It was manually checked and corrected to R03027
+index=find(strcmp('HMR_7709',ihuman.rxns));
+ihuman.rxnKEGGID{index}='R03027';
+
+% Save as version 2.0.1
+save('HMRdatabase2_01.mat','ihuman'); %===2018-01-16