(no commit message)
[jalview.git] / forester / java / src / org / forester / util / SequenceAccessionTools.java
index 5ab000e..82bcae9 100644 (file)
@@ -38,38 +38,33 @@ import org.forester.phylogeny.data.Sequence;
 \r
 public final class SequenceAccessionTools {\r
 \r
-    // gb_ADF31344_1_segmented_worms_\r
-    // gb_AAA96518_1\r
-    // gb_EHB07727_1_rodents_\r
-    // dbj_BAF37827_1_turtles_\r
-    // emb_CAA73223_1_primates_\r
-    // lcl_91970_unknown_\r
-    // mites|ref_XP_002434188_1\r
-    // ref_XP_002434188_1_mites___ticks_\r
-    // ref_NP_001121530_1_frogs___toads_\r
     //The format for GenBank Accession numbers are:\r
     //Nucleotide: 1 letter + 5 numerals OR 2 letters + 6 numerals\r
     //Protein:    3 letters + 5 numerals\r
     //http://www.ncbi.nlm.nih.gov/Sequin/acc.html\r
-    public final static Pattern  GENBANK_NUC_PATTERN_1 = Pattern\r
-                                                               .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]\\d{5}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );\r
-    public final static Pattern  GENBANK_NUC_PATTERN_2 = Pattern\r
-                                                               .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}\\d{6}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );\r
-    public final static Pattern  GENBANK_PROT_PATTERN  = Pattern\r
-                                                               .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{3}\\d{5}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );\r
-    public final static Pattern  GI_PATTERN            = Pattern.compile( "(?:\\b|_)(?:GI|gi)[|_=:](\\d+)(?:\\b|_)" );\r
-    public final static Pattern  UNIPROT_KB_PATTERN_0  = Pattern\r
-                                                               .compile( "(?:\\b|_)([A-Z][0-9][A-Z0-9]{3}[0-9])(?:\\b|_)" );\r
-    public final static Pattern  UNIPROT_KB_PATTERN_1  = Pattern\r
-                                                               .compile( "(?:\\b|_)(?:sp|tr)[\\.|\\-_=/\\\\]([A-Z][0-9][A-Z0-9]{3}[0-9])(?:\\b|_)" );\r
-    public final static Pattern  UNIPROT_KB_PATTERN_2  = Pattern\r
-                                                               .compile( "(?:\\b|_)(?:[A-Z0-9]{2,5}|(?:[A-Z][0-9][A-Z0-9]{3}[0-9]))_(([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA)(?:\\b|_)" );\r
-    public final static Pattern  ENSEMBL_PATTERN       = Pattern.compile( "(?:\\b|_)(ENS[A-Z]*[0-9]+)(?:\\b|_)" );\r
-    // RefSeq accession numbers can be distinguished from GenBank accessions \r
+    public final static Pattern  GENBANK_NUC_PATTERN_1       = Pattern\r
+            .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]\\d{5}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );\r
+    public final static Pattern  GENBANK_NUC_PATTERN_2       = Pattern\r
+            .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}\\d{6}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );\r
+    public final static Pattern  GENBANK_PROT_PATTERN        = Pattern\r
+            .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{3}\\d{5}(?:\\.\\d+)?)(?:[^a-zA-Z0-9]|\\Z)" );\r
+    public final static Pattern  GI_PATTERN                  = Pattern\r
+            .compile( "(?:\\b|_)(?:GI|gi)[|_=:](\\d+)(?:\\b|_)" );\r
+    public final static String   UNIPROT_KB_BASE_PATTERN_STR = "((?:[OPQ][0-9][A-Z0-9]{3}[0-9])|(?:[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2}))";\r
+    public final static Pattern  UNIPROT_KB_PATTERN_0        = Pattern.compile( "(?:\\b|_)"\r
+            + UNIPROT_KB_BASE_PATTERN_STR + "(?:\\b|_)" );\r
+    public final static Pattern  UNIPROT_KB_PATTERN_1        = Pattern.compile( "(?:\\b|_)(?:sp|tr)[\\.|\\-_=/\\\\]"\r
+            + UNIPROT_KB_BASE_PATTERN_STR + "(?:\\b|_)" );\r
+    public final static Pattern  UNIPROT_KB_PATTERN_2        = Pattern\r
+            .compile( "(?:\\b|_)(?:[A-Z0-9]{2,5}|"\r
+                    + UNIPROT_KB_BASE_PATTERN_STR\r
+                    + ")_(([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA)(?:\\b|_)" );\r
+    public final static Pattern  ENSEMBL_PATTERN             = Pattern.compile( "(?:\\b|_)(ENS[A-Z]*[0-9]+)(?:\\b|_)" );\r
+    // RefSeq accession numbers can be distinguished from GenBank accessions\r
     // by their distinct prefix format of 2 characters followed by an\r
-    // underscore character ('_'). For example, a RefSeq protein accession is NP_015325. \r
-    private final static Pattern REFSEQ_PATTERN        = Pattern\r
-                                                               .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}_\\d{6,})(?:[^a-zA-Z0-9]|\\Z)" );\r
+    // underscore character ('_'). For example, a RefSeq protein accession is NP_015325.\r
+    private final static Pattern REFSEQ_PATTERN              = Pattern\r
+            .compile( "(?:\\A|.*[^a-zA-Z0-9])([A-Z]{2}_\\d{6,})(?:[^a-zA-Z0-9]|\\Z)" );\r
 \r
     private SequenceAccessionTools() {\r
         // Hiding the constructor.\r