Merge branch 'Jalview-JS/develop' into feature/JAL-3690_callback-based-web-services

[jalview.git] / src / jalview / ws / ebi / hmmerClient.java

diff --git a/src/jalview/ws/ebi/hmmerClient.java b/src/jalview/ws/ebi/hmmerClient.java
index 3db35be..41cd0d8 100644 (file)
--- a/src/jalview/ws/ebi/hmmerClient.java
+++ b/src/jalview/ws/ebi/hmmerClient.java
@@ -1,6 +1,13 @@
 package jalview.ws.ebi;
 
+import jalview.datamodel.AlignmentI;
+import jalview.io.AppletFormatAdapter;
+import jalview.io.DataSourceType;
+import jalview.io.FileFormat;
+import jalview.io.FileParse;
+
 import java.io.File;
+import java.io.IOException;
 import java.util.regex.Matcher;
 
 import org.apache.axis.transport.http.HTTPConstants;
@@ -50,6 +57,7 @@ public class hmmerClient
     System.out.println("Result\n" + res);
     return;
   }
+
   /**
    * 
    * @param input
@@ -154,8 +162,8 @@ public class hmmerClient
           if ("PEND".equals(st) || "RUN".equals("st"))
           {
             JSONArray iters = jobstate.getJSONArray("result");
-            lastiter = iters.getJSONObject(iters.length() - 1)
-                    .getString("uuid");
+            lastiter = iters.getJSONObject(iters.length() - 1).getString(
+                    "uuid");
             if (lastiter.length() > 0)
             {
               java.util.regex.Pattern p = java.util.regex.Pattern
@@ -204,7 +212,7 @@ public class hmmerClient
     // $ua->get( $rootUrl."/results/".$lastIteration->{uuid} . "/score"
     return lastiter;
     /*
-     *  * #Job should have finished, but we may have converged, so get the last
+     * * #Job should have finished, but we may have converged, so get the last
      * job. my $results = $json->decode( $response->content ); my $lastIteration
      * = pop( @{ $results->{result} } ); #Now fetch the results of the last
      * iteration my $searchResult = $ua->get( $rootUrl."/results/" .
@@ -219,4 +227,93 @@ public class hmmerClient
      * results......
      */
   }
+
+  /**
+   * retrieve an alignment annotated with scores from JackHmmer
+   * 
+   * @param jobid
+   * @param dataset
+   * @return
+   */
+  AlignmentI retrieveJackhmmerResult(String jobid, AlignmentI dataset)
+          throws OutOfMemoryError, IOException
+  {
+    AlignmentI searchResult = null;
+
+    // get results
+
+    searchResult = new AppletFormatAdapter().readFile(baseUrl
+            + "/download/" + jobid + "/score?format=afa&t=.gz",
+            DataSourceType.URL, FileFormat.Fasta);
+
+    // TODO extract gapped columns as '.' - inserts to query profile
+
+    // TODO match up jackhammer results to dataset.
+
+    // do scores
+    FileParse jsonsource = new FileParse(baseUrl + "/download/" + jobid
+            + "/score?format=json", DataSourceType.URL);
+    if (!jsonsource.isValid())
+    {
+      throw new IOException("Couldn't access scores for Jackhammer results");
+    }
+    readJackhmmerScores(searchResult, jsonsource);
+    return searchResult;
+  }
+
+  /**
+   * // 1gri_A 1gri_A 217 jackhmmer - 163 4.7e-62 212.4 0.1 1 2 4.4e-46 2.1e-43
+   * 
+   * // 151.758316040039 0.04 11 151 3 139 1 150 0.94 GROWTH FACTOR BOUND
+   * PROTEIN // 2 1cj1_J 1gri_B // 1gri_A 1gri_A 217 jackhmmer - 163 4.7e-62
+   * 212.4 0.1 2 2 1.6e-17 7.9e-15 // 58.8796501159668 0.01 7 66 157 215 153 216
+   * 0.95 GROWTH FACTOR BOUND // PROTEIN 2 1cj1_J 1gri_B // 4h1o_A 4h1o_A 560
+   * jackhmmer - 163 2.1e-57 197.3 0.0 1 2 7.5e-28 3.6e-25 // 92.4921493530273
+   * 0.00 65 161 20 122 17 124 0.95 Tyrosine-protein // phosphatase non-receptor
+   * typ 4h1o_A
+   */
+  private static String[] _hmmsearchcols = new String[] { "acc", "name", "" };
+
+  private void readJackhmmerScores(AlignmentI searchResult,
+          FileParse jsonsource) throws IOException, OutOfMemoryError
+  {
+    HmmerJSONProcessor hjp = new HmmerJSONProcessor(searchResult);
+    hjp.parseFrom(jsonsource);
+
+    // http://www.ebi.ac.uk/Tools/hmmer/download/60048B38-7CEC-11E5-A230-CED6D26C98AD.5/score?format=csv
+    // 1gri_A 1gri_A 217 jackhmmer - 163 4.7e-62 212.4 0.1 1 2 4.4e-46 2.1e-43
+    // 151.758316040039 0.04 11 151 3 139 1 150 0.94 GROWTH FACTOR BOUND PROTEIN
+    // 2 1cj1_J 1gri_B
+    // 1gri_A 1gri_A 217 jackhmmer - 163 4.7e-62 212.4 0.1 2 2 1.6e-17 7.9e-15
+    // 58.8796501159668 0.01 7 66 157 215 153 216 0.95 GROWTH FACTOR BOUND
+    // PROTEIN 2 1cj1_J 1gri_B
+    // 4h1o_A 4h1o_A 560 jackhmmer - 163 2.1e-57 197.3 0.0 1 2 7.5e-28 3.6e-25
+    // 92.4921493530273 0.00 65 161 20 122 17 124 0.95 Tyrosine-protein
+    // phosphatase non-receptor typ 4h1o_A
+    // each line scores a fragment
+    // so for a combined score ?
+
+    /**
+     * for a sequence q sort any t against q according to overallScore(q,t)
+     * maxFragment(q,t) in sequence features parlance: for alignment
+     * s.getFeature("overallScore",q) -> range on q and range on s
+     * 
+     * 
+     */
+
+    // 151.758316040039 0.04 11 151 3 139 1 150 0.94 GROWTH FACTOR BOUND PROTEIN
+    // 2 1cj1_J 1gri_B
+    // 1gri_A 1gri_A 217 jackhmmer - 163 4.7e-62 212.4 0.1 2 2 1.6e-17 7.9e-15
+    // 58.8796501159668 0.01 7 66 157 215 153 216 0.95 GROWTH FACTOR BOUND
+    // PROTEIN 2 1cj1_J 1gri_B
+    // 4h1o_A 4h1o_A 560 jackhmmer - 163 2.1e-57 197.3 0.0 1 2 7.5e-28 3.6e-25
+    // 92.4921493530273 0.00 65 161 20 122 17 124 0.95 Tyrosine-protein
+    // phosphatase non-receptor typ 4h1o_A
+    //
+    // 4h1o_A 4h1o_A 560 jackhmmer - 163 2.1e-57 197.3 0.0 2 2 7.6e-31 3.7e-28
+    // 102.219146728516 0.03 66 161 127 236 124 238 0.94 Tyrosine-protein
+    // phosphatase non-receptor typ 4h1o_A
+
+  }
+
 }