Changeset 161 for trunk/corelib/src/org/bridgedb/webservice/biomart/IDMapperBiomart.java

Timestamp:

08/19/09 05:53:32 (3 years ago)

Author:

jgao

Message:

IDMapperBiomart refactor to clean the interface

Files:

• trunk/corelib/src/org/bridgedb/webservice/biomart/IDMapperBiomart.java (moved) (moved from trunk/corelib/src/org/bridgedb/webservice/IDMapperBiomart.java) (18 diffs)

trunk/corelib/src/org/bridgedb/webservice/biomart/IDMapperBiomart.java

@@ -16 +16 @@
 //
 
-package org.bridgedb.webservice;
-
-import java.io.BufferedReader;
+package org.bridgedb.webservice.biomart;
+
 import java.io.IOException;
 
 import java.util.HashMap;
 import java.util.HashSet;
-import java.util.Iterator;
 import java.util.Map;
 import java.util.Set;
-import java.util.Vector;
-
-import javax.xml.parsers.ParserConfigurationException;
 
 import org.bridgedb.AbstractIDMapperCapabilities;
@@ -37 +32 @@
 import org.bridgedb.IDMapperException;
 import org.bridgedb.Xref;
-import org.bridgedb.file.IDMappingReaderFromDelimitedReader;
-import org.bridgedb.webservice.biomart.Attribute;
-import org.bridgedb.webservice.biomart.BiomartStub;
-import org.bridgedb.webservice.biomart.Filter;
-import org.bridgedb.webservice.biomart.XMLQueryBuilder;
-
-import org.xml.sax.SAXException;
+import org.bridgedb.webservice.IDMapperWebservice;
+import org.bridgedb.webservice.biomart.util.BiomartClient;
 
 /**
@@ -61 +51 @@
         /** {@inheritDoc} */
         public IDMapper connect(String location) throws IDMapperException  {
-            // e.g.: transitivity=false,id-type-filter=true@dataset=oanatinus_gene_ensembl
+            // e.g.: id-type-filter=true@dataset=oanatinus_gene_ensembl
             // e.g.: http://www.biomart.org/biomart/martservice?dataset=oanatinus_gene_ensembl
-            String baseURL = BiomartStub.defaultBaseURL;
-            boolean transitivity = false;
+            String baseURL = BiomartClient.defaultBaseURL;
             boolean idTypeFilter = true;
             
@@ -78 +67 @@
 
             if (config!=null) {
-                String transitivityTag = "transitivity=";
-                idx = config.indexOf(transitivityTag);
-                if (idx!=-1) {
-                    String tran = config.substring(idx+transitivityTag.length());
-                    if (tran.toLowerCase().startsWith("true")) {
-                        transitivity = true;
-                    } else if (tran.toLowerCase().startsWith("false")) {
-                        transitivity = false;
-                    } else {
-                        throw new IDMapperException(
-                                "transivity can only be true or false");
-                    }
-                }
-
-
                 String idTypeFilterTag = "id-type-filter=";
                 idx = config.indexOf(idTypeFilterTag);
@@ -117 +91 @@
             String martTag = "mart=";
             idx = param.indexOf(martTag);
-            String martName = param.substring(idx+martTag.length());
-
-            idx = martName.indexOf("&");
+            String mart = param.substring(idx+martTag.length());
+
+            idx = mart.indexOf("&");
             if (idx>-1) {
-                martName = martName.substring(0,idx);
+                mart = mart.substring(0,idx);
             }
 
             String datasetTag = "dataset=";
             idx = param.indexOf(datasetTag);
-            String datasetName = param.substring(idx+datasetTag.length());
-
-            idx = datasetName.indexOf("&");
+            String dataset = param.substring(idx+datasetTag.length());
+
+            idx = dataset.indexOf("&");
             if (idx>-1) {
-                datasetName = datasetName.substring(0,idx);
-            }
-
-            return new IDMapperBiomart(martName, datasetName, baseURL, 
-                    idTypeFilter, transitivity);
-        }
-    }
-
-    private String martName;
-    private String datasetName;
+                dataset = dataset.substring(0,idx);
+            }
+
+            return new IDMapperBiomart(mart, dataset, baseURL,
+                    idTypeFilter);
+        }
+    }
+
+    private String mart;
+    private String dataset;
     private BiomartStub stub;
-    private boolean transitivity;
     private boolean idOnlyForTgtDataSource;
 
     private String baseURL;
 
-    private Map<DataSource, Filter> mapSrcDSFilter;
-    private Map<DataSource, Attribute> mapSrcDSAttr;
+    private Set<DataSource> supportedSrcDs;
+    private Set<DataSource> supportedTgtDs;
 
     /**
      * Transitivity is unsupported.ID only. ID only for target data sources.
      * Use default url of BiMart.
-     * @param martName name of mart
-     * @param datasetName name of dataset
+     * @param mart name of mart
+     * @param dataset name of dataset
      * @throws IDMapperException if failed to link to the dataset
      */
-    public IDMapperBiomart(String martName, String datasetName) throws IDMapperException {
-        this(martName, datasetName, null);
+    public IDMapperBiomart(String mart, String dataset)
+            throws IDMapperException {
+        this(mart, dataset, null);
     }
 
     /**
      * Use default url of BiMart.
-     * @param martName name of mart
-     * @param datasetName name of dataset
+     * @param mart name of mart
+     * @param dataset name of dataset
      * @param idOnlyForTgtDataSource id-only option, filter data source ends
      *        with 'ID' or 'Accession'.
@@ -169 +143 @@
      * @throws IDMapperException if failed to link to the dataset
      */
-    public IDMapperBiomart(String martName, String datasetName, boolean idOnlyForTgtDataSource,
-                boolean transitivity) throws IDMapperException {
-        this(martName, datasetName, null, idOnlyForTgtDataSource, transitivity);
+    public IDMapperBiomart(String mart, String dataset,
+            boolean idOnlyForTgtDataSource) throws IDMapperException {
+        this(mart, dataset, null, idOnlyForTgtDataSource);
     }
 
     /**
      * Transitivity is unsupported.ID only. ID only for target data sources.
-     * @param martName name of mart
-     * @param datasetName name of dataset
+     * @param mart name of mart
+     * @param dataset name of dataset
      * @param baseURL base url of BioMart
      * @throws IDMapperException if failed to link to the dataset
      */
-    public IDMapperBiomart(String martName, String datasetName, String baseURL)
+    public IDMapperBiomart(String mart, String dataset, String baseURL)
                 throws IDMapperException {
-        this(martName, datasetName, baseURL, true);
-    }
-
-    /**
-     * Transitivity is unsupported.
-     * @param martName name of mart
-     * @param datasetName name of dataset
-     * @param baseURL base url of BioMart
-     * @param idOnlyForTgtDataSource id-only option, filter data source ends
-     *        with 'ID' or 'Accession'.
-     * @throws IDMapperException if failed to link to the dataset
-     */
-    public IDMapperBiomart(String martName, String datasetName, String baseURL,
-            boolean idOnlyForTgtDataSource) throws IDMapperException {
-        this(martName, datasetName, baseURL, idOnlyForTgtDataSource, false);
+        this(mart, dataset, baseURL, true);
     }
 
@@ -203 +163 @@
      * Construct from a dataset, a database, id-only option and transitivity
      * option.
-     * @param martName name of mart
-     * @param datasetName name of dataset
+     * @param mart name of mart
+     * @param dataset name of dataset
      * @param baseURL base url of BioMart
      * @param idOnlyForTgtDataSource id-only option, filter data source ends
@@ -211 +171 @@
      * @throws IDMapperException if failed to link to the dataset
      */
-    public IDMapperBiomart(String martName, String datasetName, String baseURL,
-            boolean idOnlyForTgtDataSource, boolean transitivity) throws IDMapperException {
-        this.martName = martName;
-        this.datasetName = datasetName;
+    public IDMapperBiomart(String mart, String dataset, String baseURL,
+            boolean idOnlyForTgtDataSource) throws IDMapperException {
+        this.mart = mart;
+        this.dataset = dataset;
         if (baseURL!=null) {
             this.baseURL = baseURL;
         } else {
-            this.baseURL = BiomartStub.defaultBaseURL;
+            this.baseURL = BiomartClient.defaultBaseURL;
         }
 
@@ -227 +187 @@
         }
 
-        try {
-            if (!stub.getRegistry().containsKey(martName)) {
-                throw new IDMapperException("Mart not exist.");
-            }
-
-            if (!stub.getAvailableDatasets(martName).contains(stub.getDataset(datasetName))) {
-                throw new IDMapperException("dataset not exist.");
-            }
-        } catch (IOException e) {
-            throw new IDMapperException(e);
-        } catch (ParserConfigurationException e) {
-            throw new IDMapperException(e);
-        } catch (SAXException e) {
-            throw new IDMapperException(e);
-        }
-        
-        setIDOnlyForTgtDataSource(idOnlyForTgtDataSource);
-        setTransitivity(transitivity);
-
-        mapSrcDSFilter = new HashMap<DataSource, Filter>();
-        mapSrcDSAttr = new HashMap<DataSource, Attribute>();
+        if (!stub.availableMarts().contains(mart)) {
+            throw new IDMapperException("Mart not exist.");
+        }
+
+        if (!stub.availableDatasets(mart).contains(dataset)) {
+            throw new IDMapperException("dataset not exist.");
+        }
+
+        this.idOnlyForTgtDataSource = idOnlyForTgtDataSource;
+
+        supportedSrcDs = this.getSupportedSrcDataSources();
+        supportedTgtDs = this.getSupportedTgtDataSources();
         
         cap = new BiomartCapabilities();
@@ -253 +204 @@
 
     /**
-     * Filter target datasource ending with "ID" or "Accession".
-     * @param idOnlyForTgtDataSource ID-only if true
-     */
-    public void setIDOnlyForTgtDataSource(boolean idOnlyForTgtDataSource) {
-        this.idOnlyForTgtDataSource = idOnlyForTgtDataSource;
-    }
-
-    /**
      *
      * @return true if ID-only for target data sources.
@@ -269 +212 @@
 
     /**
-     * Set transitivity support.
-     * @param transitivity support transitivity if true.
-     */
-    public void setTransitivity(final boolean transitivity) {
-        this.transitivity = transitivity;
-    }
-
-    /**
-     * Get transitivity support.
-     * @return true if support transitivity; false otherwise.
-     */
-    public boolean getTransitivity() {
-        return transitivity;
-    }
-
-    /**
-     * Set base url of BioMart.
-     * @param baseURL URL of BioMart.
-     * @throws IDMapperException if failed to read local resources.
-     */
-    public void setBaseURL(final String baseURL) throws IDMapperException {
-        try {
-            stub = BiomartStub.getInstance(baseURL);
-        } catch (IOException e) {
-            throw new IDMapperException(e);
-        }
-        this.baseURL = baseURL;
-    }
-
-    /**
      *
      * @return base URL of BioMart.
@@ -308 +221 @@
     /**
      *
-     * @param mart mart name
-     */
-    public void setMart(final String mart) {
-        this.martName = mart;
-    }
-
-    /**
-     *
      * @return mart name
      */
     public String getMart() {
-        return martName;
-    }
-
-    /**
-     * Set dataset.
-     * @param dataset dataset from BioMart
-     */
-    public void setDataset(final String dataset) {
-        this.datasetName = dataset;
+        return mart;
     }
 
@@ -335 +232 @@
      */
     public String getDataset() {
-        return datasetName;
+        return dataset;
     }
 
@@ -353 +250 @@
         if (srcXrefs==null) {
             throw new java.lang.IllegalArgumentException(
-                        "srcXrefs or tgtDataSources cannot be null");
-        }
-
-        Map<Xref, Set<Xref>> result = new HashMap();
-
-        // remove unsupported source datasources
-        Set<DataSource> supportedSrcDatasources
-                    = cap.getSupportedSrcDataSources();
-        Map<DataSource, String> queryFilters = getQueryFilters(srcXrefs);
-        Iterator<DataSource> it = queryFilters.keySet().iterator();
-        while (it.hasNext()) {
-            DataSource ds = it.next();
-            if (!supportedSrcDatasources.contains(ds)) {
-                it.remove();
-            }
-        }
-        if (queryFilters.isEmpty()) {
-            return result;
-        }
-
-        // remove unsupported target datasources
-        Set<DataSource> supportedTgtDatasources = cap.getSupportedTgtDataSources();
-        Vector<DataSource> tgtDss;
-        if (tgtDataSources == null)
-                tgtDss = new Vector(supportedSrcDatasources);
-        else
-        {
-                tgtDss = new Vector(tgtDataSources);
-                tgtDss.retainAll(supportedTgtDatasources);
-        }
-        if (tgtDss.isEmpty()) {
-            return result;
-        }
-
-        for (Map.Entry<DataSource, String> filter : queryFilters.entrySet()) {
-            DataSource srcDs = filter.getKey();
-
-            String srcAttr = mapSrcDSFilter.get(srcDs).getName();
-            Attribute[] attrs = getAttributes(tgtDss, srcAttr);
-
-            Map<String, String> queryFilter = new HashMap(1);
-            queryFilter.put(srcAttr, filter.getValue());
-
-            String query = XMLQueryBuilder.getQueryString(datasetName, attrs, queryFilter);
-
-            BufferedReader bfr = null;
-            try {
-                bfr = stub.sendQuery(query);
-                if (!bfr.ready())
-                    throw new IDMapperException("Query failed");
-            } catch (IOException e) {
-                throw new IDMapperException(e);
-            }
-
-            if (bfr==null) {
-                return result;
-            }
-            
-            IDMappingReaderFromDelimitedReader reader
-                    = new IDMappingReaderFromDelimitedReader(bfr,
-                                "\\t", null, transitivity);
-
-            Vector<DataSource> dss = new Vector(tgtDss.size()+1);
-            dss.addAll(tgtDss);
-            dss.add(srcDs);
-            reader.setDataSources(dss);
-            
-            Map<Xref,Set<Xref>> mapXrefs = reader.getIDMappings();
-            if (mapXrefs==null) {
-                return result;
-            }
-
-            for (Xref srcXref : srcXrefs) {
-                Set<Xref> refs = mapXrefs.get(srcXref);
-                if (refs==null) continue;
-
-                Set<Xref> tgtRefs = result.get(srcXref);
-                if (tgtRefs==null) {
-                    tgtRefs = new HashSet();
-                    result.put(srcXref, tgtRefs);
+                        "srcXrefs or tgtDataSources cannot be null.");
+        }
+
+        Map<Xref, Set<Xref>> result = new HashMap<Xref, Set<Xref>>();
+
+        // source datasources
+        Map<String, Map<String, Xref>> mapSrcTypeIDXrefs = new HashMap();
+        for (Xref xref : srcXrefs) {
+            DataSource ds = xref.getDataSource();
+            if (!supportedSrcDs.contains(ds)) continue;
+
+            String src = ds.getFullName();
+            Map<String, Xref> ids = mapSrcTypeIDXrefs.get(src);
+            if (ids==null) {
+                ids = new HashMap();
+                mapSrcTypeIDXrefs.put(src, ids);
+            }
+            ids.put(xref.getId(), xref);
+        }
+
+        // supported tgt datasources
+        Set<String> tgtTypes = new HashSet();
+        for (DataSource ds : tgtDataSources) {
+            if (supportedTgtDs.contains(ds)) {
+                tgtTypes.add(ds.getFullName());
+            }
+        }
+        String[] tgts = tgtTypes.toArray(new String[0]);
+
+        for (Map.Entry<String, Map<String, Xref>> entry :
+                mapSrcTypeIDXrefs.entrySet()) {
+            String src = entry.getKey();
+            Set<String> ids = entry.getValue().keySet();
+            Map<String,Set<String>[]> res =
+                        stub.translate(mart, dataset , src, tgts, ids);
+
+
+            for (Map.Entry<String,Set<String>[]> entryRes : res.entrySet()) {
+                String srcId = entryRes.getKey();
+                Set<String>[] tgtIds = entryRes.getValue();
+                if (tgtIds==null) { // source xref not exist
+                    continue;
                 }
 
-                for (Xref tgtXref : refs) {
-                    if (tgtDataSources.contains(tgtXref.getDataSource())) {
-                        tgtRefs.add(tgtXref);
+                Xref srcXref = mapSrcTypeIDXrefs.get(src).get(srcId);
+                
+                Set<Xref> tgtXrefs = new HashSet();
+                for (int itgt=0; itgt<tgts.length; itgt++) {
+                    for (String tgtId : tgtIds[itgt]) {
+                        Xref tgtXref = new Xref(tgtId,
+                                DataSource.getByFullName(tgts[itgt]));
+                        tgtXrefs.add(tgtXref);
                     }
                 }
-            }
-        }
+                
+                result.put(srcXref, tgtXrefs);
+            }
+        }        
 
         return result;
-    }
-
-    /**
-     * Create filters from the source xrefs.
-     * @param srcXrefs source xrefs
-     * @return map from data source to IDs
-     */
-    protected Map<DataSource,String> getQueryFilters(Set<Xref> srcXrefs) {
-        Map<DataSource, Set<String>> mapNameValue = new HashMap();
-        for (Xref xref : srcXrefs) {
-            DataSource ds = xref.getDataSource();
-            Set<String> ids = mapNameValue.get(ds);
-            if (ids==null) {
-                ids = new HashSet();
-                mapNameValue.put(ds, ids);
-            }
-            ids.add(xref.getId());
-        }
-
-        Map<DataSource,String> filters = new HashMap();
-        for (Map.Entry<DataSource, Set<String>> entry : mapNameValue.entrySet()) {
-            DataSource ds = entry.getKey();
-            StringBuilder value = new StringBuilder();
-            for (String str : entry.getValue()) {
-                value.append(str);
-                value.append(",");
-            }
-
-            int len = value.length();
-            if (len>0) {
-                value.deleteCharAt(len-1);
-            }
-
-            filters.put(ds, value.toString());
-        }
-
-        return filters;
-    }
-
-    /**
-     * This code is bollowed from IDMapperClient from Cytoscape.
-     * @param tgtDataSources target data sources
-     * @param filterName filter name
-     * @return attributes
-     */
-    protected Attribute[] getAttributes(Vector<DataSource> tgtDataSources, 
-            String filterName) {
-        int n = tgtDataSources.size();
-        Attribute[] attrs = new Attribute[n+1];
-
-        int iattr = 0;
-        for (DataSource ds : tgtDataSources) {
-            //attrs[iattr++] = new Attribute(ds.getFullName());
-            attrs[iattr++] = this.mapSrcDSAttr.get(ds);
-        }
-
-        // Database-specific modification.
-        // This is not the best way, but cannot provide universal solution.
-        Attribute attr;
-        if (datasetName.contains("REACTOME")) {
-            attr = stub.filterToAttributeName(datasetName, "REACTOME", filterName);
-        } else if (datasetName.contains("UNIPROT")) {
-            attr = stub.filterToAttributeName(datasetName, "UNIPROT", filterName);
-        } else if (datasetName.contains("VARIATION")) {
-            attr = stub.getAttribute(datasetName, filterName + "_stable_id");
-        } else {
-            attr = stub.getAttribute(datasetName, filterName);
-        }
-
-        attrs[n] = attr;
-
-        return attrs;
     }
 
@@ -535 +329 @@
      * free text search is not supported for BioMart-based IDMapper.
      */
-    public Set<Xref> freeSearch (String text, int limit) throws IDMapperException {
+    public Set<Xref> freeSearch (String text, int limit)
+            throws IDMapperException {
         throw new UnsupportedOperationException();
     }
@@ -544 +339 @@
      * @throws IOException if failed to read the filters
      */
-    protected Set<DataSource> getSupportedSrcDataSources() throws IOException {
+    protected Set<DataSource> getSupportedSrcDataSources()
+            throws IDMapperException {
         Set<DataSource> dss = new HashSet();
-        Map<String, Filter> filters = stub.getFilters(datasetName);
-        for (Filter filter : filters.values()) {
-            //String fullName = filter.getDisplayName()+" ("+filter.getName()+")";
-            String fullName = filter.getDisplayName();
-            if (fullName.endsWith("(s)")) {
-                fullName = fullName.substring(0, fullName.length()-3);
-            }
-            //TODO: mapping to bridgedb system code
-            DataSource ds = DataSource.getByFullName(fullName);
-            dss.add(ds);
-            mapSrcDSFilter.put(ds, filter);
+        Set<String> filters = stub.availableFilters(mart, dataset);
+        for (String filter : filters) {
+            DataSource ds = DataSource.getByFullName(filter);
+            if (ds!=null) {
+                dss.add(ds);
+            }
         }
         return dss;
@@ -566 +357 @@
      * @throws IOException if failed to read the filters
      */
-    protected Set<DataSource> getSupportedTgtDataSources() throws IOException {
-        Map<String, Attribute> attributeVals = stub.getAttributes(datasetName);
+    protected Set<DataSource> getSupportedTgtDataSources() 
+            throws IDMapperException {
         Set<DataSource> dss = new HashSet();
-        for (Attribute attr : attributeVals.values()) {
-            String displayName = attr.getDisplayName();
-            String name = attr.getName();
-            if (idOnlyForTgtDataSource) {
-                if (!displayName.endsWith("ID")
-                        && !displayName.endsWith("Accession")
-                        && !name.endsWith("id")
-                        && !name.endsWith("accession")) {
-                    continue;
-                }
-            }
-            //String fullName = displayName + " ("+name+")";
-            String fullName = displayName;
-            //TODO: mapping to bridgedb system code
-            DataSource ds = DataSource.getByFullName(fullName);
-            dss.add(ds);
-            mapSrcDSAttr.put(ds, attr);
+        Set<String> attributes = stub.availableAttributes(mart, dataset,
+                idOnlyForTgtDataSource);
+
+        for (String attr : attributes) {
+            DataSource ds = DataSource.getByFullName(attr);
+            if (ds!=null) {
+                dss.add(ds);
+            }
         }
         return dss;
@@ -603 +385 @@
         /** {@inheritDoc} */
         public Set<DataSource> getSupportedSrcDataSources() throws IDMapperException {
-        try {
-            return IDMapperBiomart.this.getSupportedSrcDataSources();
-        } catch (IOException ex) {
-            throw new IDMapperException(ex);
-        }
+            return IDMapperBiomart.this.supportedSrcDs;
         }
 
         /** {@inheritDoc} */
         public Set<DataSource> getSupportedTgtDataSources() throws IDMapperException {
-            try {
-            return IDMapperBiomart.this.getSupportedTgtDataSources();
-        } catch (IOException ex) {
-            throw new IDMapperException(ex);
-        }
+            return IDMapperBiomart.this.supportedTgtDs;
         }