add unique stringdb<->uniprot ids loader

meringlab · May 16, 2014 · c426cb9 · c426cb9
1 parent e2e0b98
commit c426cb9
Show file tree

Hide file tree

Showing 6 changed files with 11,267 additions and 11,239 deletions.
diff --git a/src/main/java/org/string_db/ProteinRepository.java b/src/main/java/org/string_db/ProteinRepository.java
@@ -65,4 +65,11 @@ public interface ProteinRepository {
      */
     Map<Integer, String> loadProteinSequences(Integer speciesId);
 
+    /**
+     * load unique (1:1) mapping of protein_ids-UniProt_ids  for this species
+     *
+     * @param speciesId
+     * @return {protein_id -> UniprotAC} map
+     */
+    Map<Integer, UniprotAC> loadUniqueUniProtIds(Integer speciesId);
 }
diff --git a/src/main/java/org/string_db/UniprotAC.java b/src/main/java/org/string_db/UniprotAC.java
@@ -60,7 +60,7 @@ public static void isValidUniprotAC(String ac) {
             throw new IllegalArgumentException("null AC");
         }
         if (ac.length() != 6) {
-            throw new IllegalArgumentException("AC must be exactly 6 chars long, not: " + ac.length());
+            throw new IllegalArgumentException("AC must be exactly 6 chars long, not: " + ac.length() + ": " + ac);
         }
         if (STARTS_WITH_OPQ_FORMAT.matcher(ac).matches() || OTHER_FORMAT.matcher(ac).matches()) {
             return;

diff --git a/src/main/java/org/string_db/jdbc/Postgres2HSQLDB.java b/src/main/java/org/string_db/jdbc/Postgres2HSQLDB.java
@@ -42,11 +42,6 @@
  */
 public class Postgres2HSQLDB {
     final JdbcTemplate jdbcTemplate;
-
-    public Postgres2HSQLDB(JdbcTemplate jdbcTemplate) {
-        this.jdbcTemplate = jdbcTemplate;
-    }
-
     private final String selectIndexDef =
             "SELECT pg_get_indexdef(i.indexrelid) \n" +
                     "FROM   pg_class c, pg_namespace n, pg_index i, pg_class t\n" +
@@ -58,6 +53,9 @@ public Postgres2HSQLDB(JdbcTemplate jdbcTemplate) {
                     "   AND n.nspname = ?\n" +
                     "   AND t.relname = ?";
 
+    public Postgres2HSQLDB(JdbcTemplate jdbcTemplate) {
+        this.jdbcTemplate = jdbcTemplate;
+    }
 
     public static void main(final String[] args) throws FileNotFoundException, UnsupportedEncodingException, SQLException {
         final ApplicationContext ctx = new AnnotationConfigApplicationContext(DbConfig.class, DriverDataSourceConfig.class);
@@ -68,9 +66,9 @@ public static void main(final String[] args) throws FileNotFoundException, Unsup
         // skip checksum (not used) and annotation_word_vectors columns (postgresql specific data type - tsvector)
         converter.dumpTable("items", "proteins", "species_id in (511145)", "hsql-data.sql",
                 new String[]{"protein_id", "protein_external_id", "species_id", "annotation", "preferred_name"});
-        //m.pneumoniae has fewest names:
+//        m.pneumoniae has fewest names:
         converter.dumpTable("items", "proteins_names", "species_id in (272634)", "hsql-data.sql",
-                new String[]{"protein_id", "protein_name", "species_id", /*escape keyword*/"\"source\""});
+                new String[]{"protein_id", "protein_name", "species_id", /*escape keyword*/"\"source\"", "linkout"});
     }
 
     private void dumpTable(String schema, String table, String filter, String fileName, String[] columnsToInclude) {

diff --git a/src/main/java/org/string_db/jdbc/ProteinRepositoryJdbc.java b/src/main/java/org/string_db/jdbc/ProteinRepositoryJdbc.java
@@ -18,11 +18,13 @@
 
 package org.string_db.jdbc;
 
+import org.apache.log4j.Logger;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.jdbc.core.namedparam.MapSqlParameterSource;
 import org.springframework.stereotype.Component;
 import org.string_db.ProteinExternalId;
 import org.string_db.ProteinRepository;
+import org.string_db.UniprotAC;
 
 import java.util.Map;
 import java.util.Set;
@@ -33,16 +35,25 @@
 @Component
 public class ProteinRepositoryJdbc implements ProteinRepository {
 
-    @Autowired
-    GenericQueryProcessor queryProcessor;
-
+    private static final Logger log = Logger.getLogger(ProteinRepositoryJdbc.class);
     protected final TwoColumnRowMapper<Integer, String, Set<String>> multiValSqlRowMapper = TwoColumnRowMapper.multiValMapper();
     protected TwoColumnRowMapper<Integer, String, ProteinExternalId> idExternalIdMapper = new TwoColumnRowMapper<Integer, String, ProteinExternalId>() {
         @Override
-        public void addToMap(Integer firstColumn, String secondColumn, Map<Integer, ProteinExternalId> map) {
-            map.put(firstColumn, new ProteinExternalId(secondColumn));
+        public void addToMap(Integer protein_id, String protein_external_id, Map<Integer, ProteinExternalId> map) {
+            map.put(protein_id, new ProteinExternalId(protein_external_id));
+        }
+    };
+    protected TwoColumnRowMapper<Integer, String, UniprotAC> uniprotAcMapper = new TwoColumnRowMapper<Integer, String, UniprotAC>() {
+        @Override
+        public void addToMap(Integer proteinId, String linkout, Map<Integer, UniprotAC> map) {
+            if (map.containsKey(proteinId)) {
+                log.warn("duplicate uniprotAc for " + proteinId);
+            }
+            map.put(proteinId, new UniprotAC(linkout));
         }
     };
+    @Autowired
+    GenericQueryProcessor queryProcessor;
 
     @Override
     public Map<Integer, ProteinExternalId> loadExternalIds(Integer speciesId) {
@@ -84,6 +95,12 @@ public Map<Integer, String> loadProteinSequences(Integer speciesId) {
                 new MapSqlParameterSource("species_id", speciesId));
     }
 
+    @Override
+    public Map<Integer, UniprotAC> loadUniqueUniProtIds(Integer speciesId) {
+        return queryProcessor.selectTwoColumns("protein_id", "protein_name", "items.proteins_names", uniprotAcMapper,
+                "linkout = 'UniProt' AND species_id = :species_id", new MapSqlParameterSource("species_id", speciesId));
+    }
+
 }
 
 
diff --git a/src/test/java/org/string_db/jdbc/ProteinRepositoryJdbcTest.java b/src/test/java/org/string_db/jdbc/ProteinRepositoryJdbcTest.java
@@ -21,6 +21,7 @@
 import com.google.common.collect.ImmutableSet;
 import org.junit.Test;
 import org.string_db.ProteinExternalId;
+import org.string_db.UniprotAC;
 
 import java.util.Map;
 import java.util.Set;
@@ -73,4 +74,9 @@ public void test_protein_preferred_name() throws Exception {
         assertEquals("thrA", names.get(4735233));
     }
 
+    @Test
+    public void test_loadUniqueUniProtIds() throws Exception {
+        final Map<Integer, UniprotAC> ids = repo.loadUniqueUniProtIds(272634);
+        assertEquals(new UniprotAC("P11311"), ids.get(2815147));
+    }
 }