+ * Note that species and assembly identifiers passed to this class must be valid
+ * in Ensembl. They are not case sensitive.
+ *
+ * @author gmcarstairs
+ * @see https://rest.ensembl.org/documentation/info/assembly_map
+ * @see https://rest.ensembl.org/info/assembly/human?content-type=text/xml
+ * @see https://rest.ensembl.org/info/species?content-type=text/xml
+ */
public class EnsemblMap extends EnsemblRestClient
{
+ private static final String MAPPED = "mapped";
+
+ private static final String MAPPINGS = "mappings";
+
+ private static final String CDS = "cds";
+
+ private static final String CDNA = "cdna";
/**
* Default constructor (to use rest.ensembl.org)
@@ -48,20 +93,36 @@ public class EnsemblMap extends EnsemblRestClient
return null; // not used
}
- protected URL getUrl(String species, String chromosome, String fromRef,
+ /**
+ * Constructs a URL of the format
+ * http://rest.ensembl.org/map/human/GRCh38/17:45051610..45109016:1/GRCh37?content-type=application/json
+ *
+ *
+ * @param species
+ * @param chromosome
+ * @param fromRef
+ * @param toRef
+ * @param startPos
+ * @param endPos
+ * @return
+ * @throws MalformedURLException
+ */
+ protected URL getAssemblyMapUrl(String species, String chromosome, String fromRef,
String toRef, int startPos, int endPos)
throws MalformedURLException
{
- String url = getDomain() + "/map/" + species + "/" + fromRef + "/"
- + chromosome + ":" + startPos + ".." + endPos + ":1/" + toRef
- + "?content-type=application/json";
- try
- {
- return new URL(url);
- } catch (MalformedURLException e)
- {
- return null;
- }
+ /*
+ * start-end might be reverse strand - present forwards to the service
+ */
+ boolean forward = startPos <= endPos;
+ int start = forward ? startPos : endPos;
+ int end = forward ? endPos : startPos;
+ String strand = forward ? "1" : "-1";
+ String url = String.format(
+ "%s/map/%s/%s/%s:%d..%d:%s/%s?content-type=application/json",
+ getDomain(), species, fromRef, chromosome, start, end, strand,
+ toRef);
+ return new URL(url);
}
@Override
@@ -71,24 +132,24 @@ public class EnsemblMap extends EnsemblRestClient
}
@Override
- protected String getRequestMimeType(boolean multipleIds)
- {
- return "application/json";
- }
-
- @Override
- protected String getResponseMimeType()
- {
- return "application/json";
- }
-
- @Override
protected URL getUrl(List
* {"mappings":
@@ -122,7 +195,7 @@ public class EnsemblMap extends EnsemblRestClient
* @param br
* @return
*/
- protected int[] parseResponse(BufferedReader br)
+ protected int[] parseAssemblyMappingResponse(BufferedReader br)
{
int[] result = null;
JSONParser jp = new JSONParser();
@@ -130,17 +203,25 @@ public class EnsemblMap extends EnsemblRestClient
try
{
JSONObject parsed = (JSONObject) jp.parse(br);
- JSONArray mappings = (JSONArray) parsed.get("mappings");
+ JSONArray mappings = (JSONArray) parsed.get(MAPPINGS);
Iterator rvals = mappings.iterator();
while (rvals.hasNext())
{
// todo check for "mapped"
JSONObject val = (JSONObject) rvals.next();
- JSONObject mapped = (JSONObject) val.get("mapped");
- String start = mapped.get("start").toString();
- String end = mapped.get("end").toString();
- result = new int[] { Integer.parseInt(start), Integer.parseInt(end) };
+ JSONObject mapped = (JSONObject) val.get(MAPPED);
+ int start = Integer.parseInt(mapped.get("start").toString());
+ int end = Integer.parseInt(mapped.get("end").toString());
+ String strand = mapped.get("strand").toString();
+ if ("1".equals(strand))
+ {
+ result = new int[] { start, end };
+ }
+ else
+ {
+ result = new int[] { end, start };
+ }
}
} catch (IOException | ParseException | NumberFormatException e)
{
@@ -149,4 +230,193 @@ public class EnsemblMap extends EnsemblRestClient
return result;
}
+ /**
+ * Calls the REST /map/cds/id service, and returns a DBRefEntry holding the
+ * returned chromosomal coordinates, or returns null if the call fails
+ *
+ * @param division
+ * e.g. Ensembl, EnsemblMetazoa
+ * @param accession
+ * e.g. ENST00000592782, Y55B1AR.1.1
+ * @param start
+ * @param end
+ * @return
+ */
+ public GeneLociI getCdsMapping(String division, String accession,
+ int start, int end)
+ {
+ return getIdMapping(division, accession, start, end, CDS);
+ }
+
+ /**
+ * Calls the REST /map/cdna/id service, and returns a DBRefEntry holding the
+ * returned chromosomal coordinates, or returns null if the call fails
+ *
+ * @param division
+ * e.g. Ensembl, EnsemblMetazoa
+ * @param accession
+ * e.g. ENST00000592782, Y55B1AR.1.1
+ * @param start
+ * @param end
+ * @return
+ */
+ public GeneLociI getCdnaMapping(String division, String accession,
+ int start, int end)
+ {
+ return getIdMapping(division, accession, start, end, CDNA);
+ }
+
+ GeneLociI getIdMapping(String division, String accession, int start,
+ int end, String cdsOrCdna)
+ {
+ URL url = null;
+ BufferedReader br = null;
+
+ try
+ {
+ String domain = new EnsemblInfo().getDomain(division);
+ if (domain != null)
+ {
+ url = getIdMapUrl(domain, accession, start, end, cdsOrCdna);
+ br = getHttpResponse(url, null);
+ if (br != null)
+ {
+ return (parseIdMappingResponse(br, accession, domain));
+ }
+ }
+ return null;
+ } catch (Throwable t)
+ {
+ System.out.println("Error calling " + url + ": " + t.getMessage());
+ return null;
+ } finally
+ {
+ if (br != null)
+ {
+ try
+ {
+ br.close();
+ } catch (IOException e)
+ {
+ // ignore
+ }
+ }
+ }
+ }
+
+ /**
+ * Constructs a URL to the /map/cds/
+ * {"mappings":
+ * [
+ * {"assembly_name":"TAIR10","end":2501311,"seq_region_name":"1","gap":0,
+ * "strand":-1,"coord_system":"chromosome","rank":0,"start":2501114},
+ * {"assembly_name":"TAIR10","end":2500815,"seq_region_name":"1","gap":0,
+ * "strand":-1,"coord_system":"chromosome","rank":0,"start":2500714}
+ * ]
+ * }
+ *
+ *
+ * @param br
+ * @param accession
+ * @param domain
+ * @return
+ */
+ GeneLociI parseIdMappingResponse(BufferedReader br, String accession,
+ String domain)
+ {
+ JSONParser jp = new JSONParser();
+
+ try
+ {
+ JSONObject parsed = (JSONObject) jp.parse(br);
+ JSONArray mappings = (JSONArray) parsed.get(MAPPINGS);
+
+ Iterator rvals = mappings.iterator();
+ String assembly = null;
+ String chromosome = null;
+ int fromEnd = 0;
+ List