X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fdatamodel%2FDBRefSource.java;h=2d2ae4ff582c24e59cabafade39de76660b9a353;hb=9d2408483e451285fd555c3cd6e0273977acbaa7;hp=8783e4f3a712ce7f61efe14a9c919c2da94522e0;hpb=36de360de19fa715542db1cebe98ab65b83db540;p=jalview.git

diff --git a/src/jalview/datamodel/DBRefSource.java b/src/jalview/datamodel/DBRefSource.java
index 8783e4f..2d2ae4f 100755
--- a/src/jalview/datamodel/DBRefSource.java
+++ b/src/jalview/datamodel/DBRefSource.java
@@ -21,76 +21,146 @@
 package jalview.datamodel;
 
 /**
+ * BH 2018 SwingJS note: If additional final static Strings are added to this
+ * file, they should be added to public static final String[] allTypes.
+ * 
  * Defines internal constants for unambiguous annotation of DbRefEntry source
  * strings and describing the data retrieved from external database sources (see
- * jalview.ws.DbSourcProxy)
+ * jalview.ws.DbSourcProxy) <br/>
+ * TODO: replace with ontology to allow recognition of particular attributes
+ * (e.g. protein coding, alignment (ortholog db, paralog db, domain db),
+ * genomic, transcriptomic, 3D structure providing (PDB, MODBASE, etc) ..).
+ * 
+ * 
  * 
  * @author JimP
  * 
  */
+import java.util.Locale;
+
 public class DBRefSource
 {
+  
+  
+  
+  public static final String UNIPROT = "UNIPROT";
+  public static final String UP_NAME = "UNIPROT_NAME".toUpperCase(Locale.ROOT);
   /**
-   * UNIPROT Accession Number
+   * Uniprot Knowledgebase/TrEMBL as served from EMBL protein products.
    */
-  public static String UNIPROT = "UNIPROT";
+  public static final String UNIPROTKB = "UniProtKB/TrEMBL".toUpperCase(Locale.ROOT);
 
-  /**
-   * UNIPROT Entry Name
-   */
-  public static String UP_NAME = "UNIPROT_NAME".toUpperCase();
+  public static final String ENSEMBL        = "ENSEMBL";
+  public static final String ENSEMBLGENOMES = "ENSEMBLGENOMES";
+  
+  public static final String EMBL           = "EMBL";
+  public static final String EMBLCDS        = "EMBLCDS";
+  public static final String EMBLCDSProduct = "EMBLCDSProtein".toUpperCase(Locale.ROOT);
 
-  /**
-   * Uniprot Knowledgebase/TrEMBL as served from EMBL protein products.
-   */
-  public static final String UNIPROTKB = "UniProtKB/TrEMBL".toUpperCase();
+  public static final String PDB    = "PDB";
+  public static final String PFAM   = "PFAM";
+  public static final String RFAM   = "RFAM";
+  public static final String GENEDB = "GeneDB".toUpperCase(Locale.ROOT);
 
-  public static final String EMBLCDSProduct = "EMBLCDSProtein"
-          .toUpperCase();
+  public static final String PDB_CANONICAL_NAME = PDB;
 
-  /**
-   * PDB Entry Code
-   */
-  public static String PDB = "PDB";
 
-  /**
-   * EMBL ID
-   */
-  public static String EMBL = "EMBL";
+  public static final String[] allSources = new String[] {
+      UNIPROT, 
+      UP_NAME, UNIPROTKB, 
+      ENSEMBL, ENSEMBLGENOMES, 
+      EMBL, EMBLCDS, EMBLCDSProduct, 
+      PDB, PFAM, RFAM, GENEDB 
+      };  
 
-  /**
-   * EMBLCDS ID
-   */
-  public static String EMBLCDS = "EMBLCDS";
+	public static final int UNIPROT_MASK          = 1<<0;
+	public static final int UP_NAME_MASK          = 1<<1;
+	public static final int UNIPROT_KB_MASK       = 1<<2;
+	public static final int ENSEMBL_MASK          = 1<<3;
+	public static final int ENSEMBL_GENOMES_MASK  = 1<<4;
+	public static final int EMBL_MASK             = 1<<5;
+	public static final int EMBL_CDS_MASK         = 1<<6;
+	public static final int EMBL_CDS_PRODUCT_MASK = 1<<7;
+	public static final int PDB_MASK              = 1<<8;
+	public static final int PFAM_MASK             = 1<<9;
+	public static final int RFAM_MASK             = 1<<10;
+	public static final int GENE_DB_MASK          = 1<<11;
 
-  /**
-   * PFAM ID
-   */
-  public static String PFAM = "PFAM";
+    public static final int MASK_COUNT = 12;
+	  
+    public static final int ALL_MASKS = (1 << MASK_COUNT) - 1;
 
-  /**
-   * RFAM ID
-   */
-  public static String RFAM = "RFAM";
+	public static int getSourceKey(String name) {
+		  for (int i = 0; i < MASK_COUNT; i++) {
+			  if (name.equals(allSources[i]))
+        {
+          return 1<<i;
+        }
+		  }
+		  return 0;	  
+	  }
 
-  /**
-   * GeneDB ID
-   */
-  public static final String GENEDB = "GeneDB".toUpperCase();
+	public static final int PRIMARY_MASK = UNIPROT_MASK | ENSEMBL_MASK;
 
-  /**
-   * Ensembl
-   */
-  public static final String ENSEMBL = "ENSEMBL";
+	  /**
+	  * List of databases whose sequences might have coding regions annotated
+	  */
+	 public static final String[] DNACODINGDBS = { 
+		     ENSEMBL, ENSEMBLGENOMES,
+			 EMBL, EMBLCDS, GENEDB
+	     };
+	
+     public static final int DNA_CODING_MASK = 
+    		 ENSEMBL_MASK | ENSEMBL_GENOMES_MASK
+    	  | EMBL_MASK | EMBL_CDS_MASK | GENE_DB_MASK;
+
+    
+     
+    public static final String[] CODINGDBS = { EMBLCDS, GENEDB, ENSEMBL };
+
+    public static final int CODING_MASK = EMBL_CDS_MASK | GENE_DB_MASK | ENSEMBL_MASK;
+  
+   
+  
+    public static final String[] PROTEINDBS = { 
+    		UNIPROT, UNIPROTKB,
+    		ENSEMBL, EMBLCDSProduct }; // Ensembl ENSP* entries are protein
+
+    public static final int PROTEIN_MASK = 
+    		  UNIPROT_MASK | UNIPROT_KB_MASK 
+    		| ENSEMBL_MASK | EMBL_CDS_PRODUCT_MASK ;
+
+
+    // for SequenceAnnotationReport only
+    
+//    public static final String[][] PRIMARY_SOURCES = new String[][] {
+//  	  CODINGDBS, DNACODINGDBS, PROTEINDBS };
+//  	  
+  	public static final int PRIMARY_SOURCES_MASK = CODING_MASK | DNA_CODING_MASK | PROTEIN_MASK;
+      	
+    public static boolean isPrimarySource(String source)
+    {
+    	return ((PRIMARY_SOURCES_MASK & getSourceKey(source)) != 0);
+    }
+
+  public static boolean isPrimaryCandidate(String ucversion) {	
+    // tricky - this test really needs to search the sequence's set of dbrefs to
+    // see if there is a primary reference that derived this reference.
+    for (int i = allSources.length; --i >= 0;)
+    {
+      if (ucversion.startsWith(allSources[i])) // BH 2019.01.25 .toUpperCase(Locale.ROOT) unnecessary here for allSources
+      {
+        // by convention, many secondary references inherit the primary
+        // reference's
+        // source string as a prefix for any version information from the
+        // secondary reference.
+        return false;
+      }
+    }
+    return true;
+}
 
-  /**
-   * List of databases whose sequences might have coding regions annotated
-   */
-  public static final String[] DNACODINGDBS = { EMBL, EMBLCDS, GENEDB,
-      ENSEMBL };
 
-  public static final String[] CODINGDBS = { EMBLCDS, GENEDB, ENSEMBL };
 
-  public static final String[] PROTEINDBS = { UNIPROT, PDB, UNIPROTKB,
-      EMBLCDSProduct, ENSEMBL }; // Ensembl ENSP* entries are protein
+  
 }