parseHPA
Parses a database dump of the Human Protein Atlas (HPA)
Input:
fileName comma- or tab-separated database dump of HPA. For details
regarding the format, see
http://www.proteinatlas.org/about/download.
version version of HPA [optional, default=19]
Output:
hpaData
genes cell array with the unique gene names. In
version >=18 this is the ensemble name, see
geneNames below for the names in ver >=18
geneNames cell array with the gene names, indexed the
same way as genes.
tissues cell array with the tissue names. The list may not be
unique, as there can be multiple cell types per tissue
celltypes cell array with the cell type names for each tissue
levels cell array with the unique expression levels
types cell array with the unique evidence types
reliabilities cell array with the unique reliability levels
gene2Level gene-to-expression level mapping in sparse matrix form.
The value for element i,j is the index in
hpaData.levels of gene i in cell type j
gene2Type gene-to-evidence type mapping in sparse matrix form.
The value for element i,j is the index in
hpaData.types of gene i in cell type j. Doesn't
exist in version >=18.
gene2Reliability gene-to-reliability level mapping in sparse matrix form.
The value for element i,j is the index in
hpaData.reliabilities of gene i in cell type j
Usage: hpaData=parseHPA(fileName,version)0001 function hpaData=parseHPA(fileName, version) 0002 % parseHPA 0003 % Parses a database dump of the Human Protein Atlas (HPA) 0004 % 0005 % Input: 0006 % fileName comma- or tab-separated database dump of HPA. For details 0007 % regarding the format, see 0008 % http://www.proteinatlas.org/about/download. 0009 % version version of HPA [optional, default=19] 0010 % 0011 % 0012 % Output: 0013 % hpaData 0014 % genes cell array with the unique gene names. In 0015 % version >=18 this is the ensemble name, see 0016 % geneNames below for the names in ver >=18 0017 % geneNames cell array with the gene names, indexed the 0018 % same way as genes. 0019 % tissues cell array with the tissue names. The list may not be 0020 % unique, as there can be multiple cell types per tissue 0021 % celltypes cell array with the cell type names for each tissue 0022 % levels cell array with the unique expression levels 0023 % types cell array with the unique evidence types 0024 % reliabilities cell array with the unique reliability levels 0025 % 0026 % gene2Level gene-to-expression level mapping in sparse matrix form. 0027 % The value for element i,j is the index in 0028 % hpaData.levels of gene i in cell type j 0029 % gene2Type gene-to-evidence type mapping in sparse matrix form. 0030 % The value for element i,j is the index in 0031 % hpaData.types of gene i in cell type j. Doesn't 0032 % exist in version >=18. 0033 % gene2Reliability gene-to-reliability level mapping in sparse matrix form. 0034 % The value for element i,j is the index in 0035 % hpaData.reliabilities of gene i in cell type j 0036 % 0037 % Usage: hpaData=parseHPA(fileName,version) 0038 0039 if nargin<2 0040 version=19; %Change this and add code for more versions when the current HPA version is increased and the format is changed 0041 end 0042 0043 fileName=char(fileName); 0044 if ~isfile(fileName) 0045 error('HPA file %s cannot be found',string(fileName)); 0046 end 0047 0048 if (version >= 17) 0049 fid=fopen(fileName,'r'); 0050 hpa=textscan(fid,'%q %q %q %q %q %q','Delimiter','\t'); 0051 fclose(fid); 0052 0053 %Go through and see if the headers match what was expected 0054 headers={'Gene' 'Gene name' 'Tissue' 'Cell type' 'Level' 'Reliability'}; 0055 for i=1:numel(headers) 0056 if ~strcmpi(headers(i),hpa{i}(1)) 0057 EM=['Could not find the header "' headers{i} '". Make sure that the input file matches the format specified at http://www.proteinatlas.org/about/download']; 0058 dispEM(EM); 0059 end 0060 %Remove the header line here 0061 hpa{i}(1)=[]; 0062 end 0063 0064 %Get the unique values of each data type 0065 [hpaData.genes, P, I]=unique(hpa{1}); 0066 hpaData.geneNames=hpa{2}(P); %make this vector use the index as genes 0067 [~, J, K]=unique(strcat(hpa{3},'€',hpa{4})); 0068 hpaData.tissues=hpa{3}(J); 0069 hpaData.celltypes=hpa{4}(J); 0070 [hpaData.levels, ~, L]=unique(hpa{5}); 0071 [hpaData.reliabilities, ~, N]=unique(hpa{6}); 0072 0073 %Map the data to be sparse matrises instead 0074 hpaData.gene2Level=sparse(I,K,L,numel(hpaData.genes),numel(hpaData.tissues)); 0075 hpaData.gene2Reliability=sparse(I,K,N,numel(hpaData.genes),numel(hpaData.tissues)); 0076 else 0077 fid=fopen(fileName,'r'); 0078 hpa=textscan(fid,'%q %q %q %q %q %q','Delimiter',','); 0079 fclose(fid); 0080 0081 %Go through and see if the headers match what was expected 0082 headers={'Gene' 'Tissue' 'Cell type' 'Level' 'Expression type' 'Reliability'}; 0083 for i=1:numel(headers) 0084 if ~strcmpi(headers(i),hpa{i}(1)) 0085 EM=['Could not find the header "' headers{i} '". Make sure that the input file matches the format specified at http://www.proteinatlas.org/about/download']; 0086 dispEM(EM); 0087 end 0088 %Remove the header line here 0089 hpa{i}(1)=[]; 0090 end 0091 0092 %Get the unique values of each data type 0093 [hpaData.genes, ~, I]=unique(hpa{1}); 0094 [~, J, K]=unique(strcat(hpa{2},'€',hpa{3})); 0095 hpaData.tissues=hpa{2}(J); 0096 hpaData.celltypes=hpa{3}(J); 0097 [hpaData.levels, ~, L]=unique(hpa{4}); 0098 [hpaData.types, ~, M]=unique(hpa{5}); 0099 [hpaData.reliabilities, ~, N]=unique(hpa{6}); 0100 0101 %Map the data to be sparse matrises instead 0102 hpaData.gene2Level=sparse(I,K,L,numel(hpaData.genes),numel(hpaData.tissues)); 0103 hpaData.gene2Type=sparse(I,K,M,numel(hpaData.genes),numel(hpaData.tissues)); 0104 hpaData.gene2Reliability=sparse(I,K,N,numel(hpaData.genes),numel(hpaData.tissues)); 0105 end 0106 end