2727
2828import org .bridgedb .DataSource ;
2929import org .bridgedb .Xref ;
30- import org .bridgedb .bio .BioDataSource ;
3130import org .pathvisio .core .debug .Logger ;
3231
3332/**
34- * Exports to pathway format understood by the EuGene
35- * pathway statistics program.
36- * This format is basically a list of genes in a flat text file
33+ * Exports to pathway format understood by the EuGene pathway statistics
34+ * program. This format is basically a list of genes in a flat text file
3735 * preceded by 3 header lines.
3836 *
39- * EuGene supports several id systems but has its own naming
40- * for them, this exporter also handles the translation.
37+ * EuGene supports several id systems but has its own naming for them, this
38+ * exporter also handles the translation.
4139 */
42- public class EUGeneExporter implements PathwayExporter
43- {
40+ public class EUGeneExporter implements PathwayExporter {
4441 public String [] getExtensions () {
4542 return new String [] { "pwf" };
4643 }
@@ -53,7 +50,7 @@ public void doExport(File file, Pathway pathway) throws ConverterException {
5350 EUGenePathway eugPathway = new EUGenePathway (pathway );
5451 try {
5552 eugPathway .writeToEUGene (file );
56- } catch (Exception e ) {
53+ } catch (Exception e ) {
5754 throw new ConverterException (e );
5855 }
5956 }
@@ -62,11 +59,11 @@ private static class EUGenePathway {
6259 Logger log = Logger .log ;
6360 Pathway pathway ;
6461
65- DataSource system ; //The annotation system
62+ DataSource system ; // The annotation system
6663
6764 List <Xref > refs ;
6865
69- public EUGenePathway (Pathway p ) {
66+ public EUGenePathway (Pathway p ) {
7067 pathway = p ;
7168 read ();
7269 }
@@ -77,30 +74,27 @@ void writeToEUGene(File file) throws FileNotFoundException {
7774 StringBuilder missedGenes = new StringBuilder ();
7875 euGeneSystem = getEUGeneSystem ();
7976
80- for (Xref ref : refs )
81- {
77+ for (Xref ref : refs ) {
8278 DataSource ds = ref .getDataSource ();
8379 String id = ref .getId ();
84- if (ds == system )
85- { //Check if gene is of most occuring system
80+ if (ds == system ) { // Check if gene is of most occuring system
8681 geneString .append (id + "\n " );
87- }
88- else
89- {
82+ } else {
9083 missedGenes .append (id + "|" + ds .getSystemCode () + "; " );
9184 log .error ("id '" + id + "' differs from pathway annotation system" );
9285 }
9386 }
9487
95- //Write the file
88+ // Write the file
9689 PrintStream out = null ;
9790 out = new PrintStream (file );
9891
99- //Print the data
92+ // Print the data
10093 out .println ("//PATHWAY_NAME = " + pathway .getMappInfo ().getMapInfoName ());
10194 out .println ("//PATHWAY_SOURCE = GenMAPP" );
10295 out .println ("//PATHWAY_MARKER = " + euGeneSystem );
103- if (missedGenes .length () > 0 ) out .println ("//LOST_DURING_CONVERSION: " + missedGenes );
96+ if (missedGenes .length () > 0 )
97+ out .println ("//LOST_DURING_CONVERSION: " + missedGenes );
10498 out .print (geneString );
10599
106100 out .close ();
@@ -110,105 +104,79 @@ void read() {
110104 refs = new ArrayList <Xref >();
111105 Map <DataSource , Integer > codeCount = new HashMap <DataSource , Integer >();
112106
113- for (PathwayElement elm : pathway .getDataObjects ()) {
114- if (elm .getObjectType () != ObjectType .DATANODE ) {
115- continue ; //Skip non-datanodes
107+ for (PathwayElement elm : pathway .getDataObjects ()) {
108+ if (elm .getObjectType () != ObjectType .DATANODE ) {
109+ continue ; // Skip non-datanodes
116110 }
117111 Xref ref = elm .getXref ();
118112 DataSource ds = ref .getDataSource ();
119- if (ref == null || ref .getId ().equals ("" ) || ref .getDataSource () == null )
120- {
121- continue ; //Skip datanodes with incomplete annotation
113+ if (ref == null || ref .getId ().equals ("" ) || ref .getDataSource () == null ) {
114+ continue ; // Skip datanodes with incomplete annotation
122115 }
123- refs .add (ref );
116+ refs .add (ref );
124117
125- //Increase code count for this code
126- if (codeCount .containsKey (ref .getDataSource ()))
118+ // Increase code count for this code
119+ if (codeCount .containsKey (ref .getDataSource ()))
127120 codeCount .put (ds , codeCount .get (ds ) + 1 );
128- else codeCount .put (ds , 1 );
121+ else
122+ codeCount .put (ds , 1 );
129123 }
130124
131- //Get most occuring systemcode
125+ // Get most occuring systemcode
132126 DataSource maxCode = null ;
133- for (DataSource ds : codeCount .keySet ())
134- {
135- if (maxCode == null || codeCount .get (ds ) > codeCount .get (maxCode ))
136- {
127+ for (DataSource ds : codeCount .keySet ()) {
128+ if (maxCode == null || codeCount .get (ds ) > codeCount .get (maxCode )) {
137129 maxCode = ds ;
138130 }
139131 }
140132 system = maxCode ;
141133
142- if (system == null ) { //May occur when no identifiers available
143- system = BioDataSource . ENSEMBL ;
134+ if (system == null ) { // May occur when no identifiers available
135+ system = DataSource . getByCompactIdentifierPrefix ( "ensembl" ) ;
144136 }
145137
146- if (codeCount .keySet ().size () > 1 ) {
147- log .warn ("\t This pathway contains genes with different SystemCodes; '" +
148- maxCode + "' has the highest occurence and is therefore chosen as PATHWAY_MARKER" +
149- " for the EUGene file\n \t Other SystemCodes found and their occurences: "
150- + codeCount );
138+ if (codeCount .keySet ().size () > 1 ) {
139+ log .warn ("\t This pathway contains genes with different SystemCodes; '" + maxCode
140+ + "' has the highest occurence and is therefore chosen as PATHWAY_MARKER"
141+ + " for the EUGene file\n \t Other SystemCodes found and their occurences: " + codeCount );
151142 }
152143
153144 }
154145
155146 String getEUGeneSystem () {
156- if (systemMappings .containsKey (system ))
157- {
147+ if (systemMappings .containsKey (system )) {
158148 return systemMappings .get (system );
159- }
160- else
161- {
149+ } else {
162150 return system .getFullName ();
163151 }
164152 }
165153 }
166154
167155 private static Map <DataSource , String > systemMappings ;
168- private static final String [] EU_GENE_SYSTEMS = new String []
169- {
170- "ENSEMBL_GENE_ID" ,
171- "UNIPROT" ,
172- "ENTREZ" ,
173- "UNIGENE" ,
174- "AFFYMETRIX" ,
175- "AGILENT" ,
176- "HGNC" ,
177- "PDB_ID" ,
178- "SGD_ID"
179- };
180- private static final DataSource [] GENMAPP_SYSTEMS = new DataSource []
181- {
182- BioDataSource .ENSEMBL ,
183- BioDataSource .UNIPROT ,
184- BioDataSource .ENTREZ_GENE ,
185- BioDataSource .UNIGENE ,
186- BioDataSource .AFFY ,
187- BioDataSource .AGILENT ,
188- BioDataSource .HUGO ,
189- BioDataSource .PDB ,
190- BioDataSource .SGD
191- };
192-
193- static
194- {
156+ private static final String [] EU_GENE_SYSTEMS = new String [] { "ENSEMBL_GENE_ID" , "UNIPROT" , "ENTREZ" , "UNIGENE" ,
157+ "AFFYMETRIX" , "AGILENT" , "HGNC" , "PDB_ID" , "SGD_ID" };
158+ private static final DataSource [] GENMAPP_SYSTEMS = new DataSource [] {
159+ DataSource .getByCompactIdentifierPrefix ("ensembl" ), DataSource .getByCompactIdentifierPrefix ("uniprot" ),
160+ DataSource .getByCompactIdentifierPrefix ("ncbigene" ), DataSource .getByCompactIdentifierPrefix ("unigene" ),
161+ DataSource .getByCompactIdentifierPrefix ("affy.probeset" ), DataSource .getExistingByFullName ("Agilent" ), // TODO
162+ DataSource .getExistingByFullName ("HGNC" ), // TODO
163+ DataSource .getByCompactIdentifierPrefix ("pdb" ), DataSource .getByCompactIdentifierPrefix ("sgd" ) };
164+
165+ static {
195166 systemMappings = new HashMap <DataSource , String >();
196- for (int i = 0 ; i < EU_GENE_SYSTEMS .length ; i ++)
197- {
167+ for (int i = 0 ; i < EU_GENE_SYSTEMS .length ; i ++) {
198168 systemMappings .put (GENMAPP_SYSTEMS [i ], EU_GENE_SYSTEMS [i ]);
199169 }
200170 }
201171
202172 @ Override
203- public List <String > getWarnings ()
204- {
173+ public List <String > getWarnings () {
205174 return Collections .emptyList ();
206175 }
207176
208177 @ Override
209- public void doExport (File file , Pathway pathway , int zoom )
210- throws ConverterException {
178+ public void doExport (File file , Pathway pathway , int zoom ) throws ConverterException {
211179 // TODO Auto-generated method stub
212-
180+
213181 }
214182}
0 commit comments