*/
package jalview.ext.ensembl;
+import jalview.bin.Cache;
import jalview.datamodel.AlignmentI;
+import jalview.datamodel.GeneLociI;
+import jalview.util.MapList;
import java.io.BufferedReader;
import java.io.IOException;
import java.net.MalformedURLException;
import java.net.URL;
import java.util.Arrays;
+import java.util.Collections;
import java.util.List;
import org.json.simple.JSONObject;
import org.json.simple.parser.ParseException;
/**
- * A client for the Ensembl lookup REST endpoint; used to find the Parent gene
- * identifier given a transcript identifier.
+ * A client for the Ensembl /lookup REST endpoint, used to find the gene
+ * identifier given a gene, transcript or protein identifier, or to extract the
+ * species or chromosomal coordinates from the same service response
*
* @author gmcarstairs
- *
*/
public class EnsemblLookup extends EnsemblRestClient
{
+ private static final String SPECIES = "species";
- private static final String OBJECT_TYPE_TRANSLATION = "Translation";
- private static final String PARENT = "Parent";
- private static final String OBJECT_TYPE_TRANSCRIPT = "Transcript";
- private static final String ID = "id";
- private static final String OBJECT_TYPE_GENE = "Gene";
- private static final String OBJECT_TYPE = "object_type";
-
+ /**
+ * keep track of last identifier retrieved to break loops
+ */
private String lastId;
/**
protected URL getUrl(List<String> ids) throws MalformedURLException
{
String identifier = ids.get(0);
- return getUrl(identifier);
+ return getUrl(identifier, null);
}
/**
+ * Gets the url for lookup of the given identifier, optionally with objectType
+ * also specified in the request
+ *
* @param identifier
+ * @param objectType
* @return
*/
- protected URL getUrl(String identifier)
+ protected URL getUrl(String identifier, String objectType)
{
String url = getDomain() + "/lookup/id/" + identifier
+ CONTENT_TYPE_JSON;
+ if (objectType != null)
+ {
+ url += "&" + OBJECT_TYPE + "=" + objectType;
+ }
+
try
{
return new URL(url);
}
/**
- * Calls the Ensembl lookup REST endpoint and retrieves the 'Parent' for the
- * given identifier, or null if not found
+ * Returns the gene id related to the given identifier (which may be for a
+ * gene, transcript or protein)
*
* @param identifier
* @return
*/
public String getGeneId(String identifier)
{
+ return getGeneId(identifier, null);
+ }
+
+ /**
+ * Returns the gene id related to the given identifier (which may be for a
+ * gene, transcript or protein)
+ *
+ * @param identifier
+ * @param objectType
+ * @return
+ */
+ public String getGeneId(String identifier, String objectType)
+ {
List<String> ids = Arrays.asList(new String[] { identifier });
BufferedReader br = null;
try
{
- URL url = getUrl(identifier);
- if (identifier.equals(lastId))
- {
- System.err.println("** Ensembl lookup " + url.toString()
- + " looping on Parent!");
- return null;
- }
- lastId = identifier;
+ URL url = getUrl(identifier, objectType);
if (url != null)
{
br = getHttpResponse(url, ids);
String type = val.get(OBJECT_TYPE).toString();
if (OBJECT_TYPE_GENE.equalsIgnoreCase(type))
{
- geneId = val.get(ID).toString();
+ // got the gene - just returns its id
+ geneId = val.get(JSON_ID).toString();
}
else if (OBJECT_TYPE_TRANSCRIPT.equalsIgnoreCase(type))
{
+ // got the transcript - return its (Gene) Parent
geneId = val.get(PARENT).toString();
}
else if (OBJECT_TYPE_TRANSLATION.equalsIgnoreCase(type))
{
+ // got the protein - get its Parent, restricted to type Transcript
String transcriptId = val.get(PARENT).toString();
+ geneId = getGeneId(transcriptId, OBJECT_TYPE_TRANSCRIPT);
+ }
+ } catch (ParseException e)
+ {
+ // ignore
+ }
+ return geneId;
+ }
+
+ /**
+ * Calls the Ensembl lookup REST endpoint and retrieves the 'species' for the
+ * given identifier, or null if not found
+ *
+ * @param identifier
+ * @return
+ */
+ public String getSpecies(String identifier)
+ {
+ String species = null;
+ JSONObject json = getResult(identifier, null);
+ if (json != null)
+ {
+ Object o = json.get(SPECIES);
+ if (o != null)
+ {
+ species = o.toString();
+ }
+ }
+ return species;
+ }
+
+ /**
+ * Calls the /lookup/id rest service and returns the response as a JSONObject,
+ * or null if any error
+ *
+ * @param identifier
+ * @param objectType
+ * (optional)
+ * @return
+ */
+ protected JSONObject getResult(String identifier, String objectType)
+ {
+ List<String> ids = Arrays.asList(new String[] { identifier });
+
+ BufferedReader br = null;
+ try
+ {
+
+ URL url = getUrl(identifier, objectType);
+
+ if (identifier.equals(lastId))
+ {
+ System.err.println("** Ensembl lookup " + url.toString()
+ + " looping on Parent!");
+ return null;
+ }
+
+ lastId = identifier;
+
+ if (url != null)
+ {
+ br = getHttpResponse(url, ids);
+ }
+ return br == null ? null : (JSONObject) (new JSONParser().parse(br));
+ } catch (IOException | ParseException e)
+ {
+ System.err.println("Error parsing " + identifier + " lookup response "
+ + e.getMessage());
+ return null;
+ } finally
+ {
+ if (br != null)
+ {
try
{
- geneId = getGeneId(transcriptId);
- } catch (StackOverflowError e)
+ br.close();
+ } catch (IOException e)
{
- /*
- * unlikely data condition error!
- */
- System.err
- .println("** Ensembl lookup "
- + getUrl(transcriptId).toString()
- + " looping on Parent!");
+ // ignore
}
}
- } catch (ParseException e)
+ }
+ }
+
+ /**
+ * Parses the JSON response and returns the gene identifier, or null if not
+ * found. If the returned object_type is Gene, returns the id, if Transcript
+ * returns the Parent. If it is Translation (peptide identifier), then the
+ * Parent is the transcript identifier, so we redo the search with this value,
+ * specifying that object_type should be Transcript.
+ *
+ * @param jsonObject
+ * @return
+ */
+ protected String parseGeneId(JSONObject json)
+ {
+ if (json == null)
{
- // ignore
+ // e.g. lookup failed with 404 not found
+ return null;
}
+
+ String geneId = null;
+ String type = json.get(OBJECT_TYPE).toString();
+ if (OBJECT_TYPE_GENE.equalsIgnoreCase(type))
+ {
+ // got the gene - just returns its id
+ geneId = json.get(JSON_ID).toString();
+ }
+ else if (OBJECT_TYPE_TRANSCRIPT.equalsIgnoreCase(type))
+ {
+ // got the transcript - return its (Gene) Parent
+ geneId = json.get(PARENT).toString();
+ }
+ else if (OBJECT_TYPE_TRANSLATION.equalsIgnoreCase(type))
+ {
+ // got the protein - look up its Parent, restricted to type Transcript
+ String transcriptId = json.get(PARENT).toString();
+ geneId = parseGeneId(getResult(transcriptId, OBJECT_TYPE_TRANSCRIPT));
+ }
+
return geneId;
}
+ /**
+ * Calls the /lookup/id rest service for the given id, and if successful,
+ * parses and returns the gene's chromosomal coordinates
+ *
+ * @param geneId
+ * @return
+ */
+ public GeneLociI getGeneLoci(String geneId)
+ {
+ return parseGeneLoci(getResult(geneId, OBJECT_TYPE_GENE));
+ }
+
+ /**
+ * Parses the /lookup/id response for species, asssembly_name,
+ * seq_region_name, start, end and returns an object that wraps them, or null
+ * if unsuccessful
+ *
+ * @param json
+ * @return
+ */
+ GeneLociI parseGeneLoci(JSONObject json)
+ {
+ if (json == null)
+ {
+ return null;
+ }
+
+ try
+ {
+ final String species = json.get("species").toString();
+ final String assembly = json.get("assembly_name").toString();
+ final String chromosome = json.get("seq_region_name").toString();
+ String strand = json.get("strand").toString();
+ int start = Integer.parseInt(json.get("start").toString());
+ int end = Integer.parseInt(json.get("end").toString());
+ int fromEnd = end - start + 1;
+ boolean reverseStrand = "-1".equals(strand);
+ int toStart = reverseStrand ? end : start;
+ int toEnd = reverseStrand ? start : end;
+ List<int[]> fromRange = Collections.singletonList(new int[] { 1,
+ fromEnd });
+ List<int[]> toRange = Collections.singletonList(new int[] { toStart,
+ toEnd });
+ final MapList map = new MapList(fromRange, toRange, 1, 1);
+ return new GeneLociI()
+ {
+
+ @Override
+ public String getSpeciesId()
+ {
+ return species == null ? "" : species;
+ }
+
+ @Override
+ public String getAssemblyId()
+ {
+ return assembly;
+ }
+
+ @Override
+ public String getChromosomeId()
+ {
+ return chromosome;
+ }
+
+ @Override
+ public MapList getMap()
+ {
+ return map;
+ }
+ };
+ } catch (NullPointerException | NumberFormatException e)
+ {
+ Cache.log.error("Error looking up gene loci: " + e.getMessage());
+ e.printStackTrace();
+ }
+ return null;
+ }
+
}
git://source.jalview.org / jalview.git / blobdiff