JAL-1950 first stab at Hmmer3 JSON parser and alignment decorator

diff --git a/examples/testdata/hmmer3/alignment_frag.fa.gz b/examples/testdata/hmmer3/alignment_frag.fa.gz
new file mode 100644 (file)
index 0000000..a4e79b2
Binary files /dev/null and b/examples/testdata/hmmer3/alignment_frag.fa.gz differ

diff --git a/examples/testdata/hmmer3/alignment_res.fa.gz b/examples/testdata/hmmer3/alignment_res.fa.gz
new file mode 100644 (file)
index 0000000..03206f4
Binary files /dev/null and b/examples/testdata/hmmer3/alignment_res.fa.gz differ

diff --git a/examples/testdata/hmmer3/hit_fragment.json.gz b/examples/testdata/hmmer3/hit_fragment.json.gz
new file mode 100644 (file)
index 0000000..bbc8405
Binary files /dev/null and b/examples/testdata/hmmer3/hit_fragment.json.gz differ

diff --git a/examples/testdata/hmmer3/hmmeresult.json.gz b/examples/testdata/hmmer3/hmmeresult.json.gz
new file mode 100644 (file)
index 0000000..c3fa9a2
Binary files /dev/null and b/examples/testdata/hmmer3/hmmeresult.json.gz differ

diff --git a/src/jalview/ws/ebi/HmmerJSONProcessor.java b/src/jalview/ws/ebi/HmmerJSONProcessor.java
new file mode 100644 (file)
index 0000000..611f99e
--- /dev/null
+++ b/src/jalview/ws/ebi/HmmerJSONProcessor.java
@@ -0,0 +1,122 @@
+package jalview.ws.ebi;
+
+import jalview.datamodel.AlignmentAnnotation;
+import jalview.datamodel.AlignmentI;
+import jalview.datamodel.SequenceI;
+import jalview.io.FileParse;
+
+import java.io.IOException;
+
+import org.json.simple.JSONArray;
+import org.json.simple.JSONObject;
+import org.json.simple.parser.JSONParser;
+
+public class HmmerJSONProcessor
+{
+  AlignmentI resultAl;
+
+  public HmmerJSONProcessor(AlignmentI searchResult)
+  {
+    resultAl = searchResult;
+  }
+
+  public void parseFrom(FileParse jsonsource) throws IOException,
+          OutOfMemoryError
+  {
+    JSONParser hmmerResultParser = new JSONParser();
+    Object jsonResults = null;
+    try
+    {
+      jsonResults = hmmerResultParser.parse(jsonsource.getReader());
+    } catch (Exception p)
+    {
+      throw new IOException("While parsing from " + jsonsource.getInFile(),
+              p);
+    }
+    if (jsonResults == null)
+    {
+      throw new IOException("No data at" + jsonsource.getInFile());
+    }
+    if (!(jsonResults instanceof JSONObject))
+    {
+      throw new IOException("Unexpected JSON model at "
+              + jsonsource.getInFile());
+    }
+    try
+    {
+      JSONObject hmmsearchr = (JSONObject) ((JSONObject) jsonResults)
+              .get("results");
+      // now process the hits
+      addStatistics((JSONObject) hmmsearchr.get("stats"));
+      JSONArray jsonArray = (JSONArray) hmmsearchr.get("hits");
+      long p = 1;
+      for (Object hit : jsonArray)
+      {
+        JSONObject hmmhit = (JSONObject) hit;
+        addHit(hmmhit, p++);
+      }
+    } catch (ClassCastException q)
+    {
+      throw new IOException("Unexpected JSON model content at "
+              + jsonsource.getInFile(), q);
+    }
+  }
+
+  /**
+   * 
+   * @param object
+   *          - actually a JSONObject key value set of search statistics.
+   */
+  public void addStatistics(JSONObject stats)
+  {
+    for (Object stat : stats.keySet())
+    {
+      String key = (String) stat;
+      Object val = stats.get(key);
+      resultAl.setProperty(key, "" + val);
+    }
+  }
+
+  //
+  // mapping of keys to types of property on sequence
+  //
+  public void addHit(JSONObject hmmrhit, long p)
+  {
+    String sname = (String) hmmrhit.get("name");
+    SequenceI[] hits = resultAl.findSequenceMatch(sname);
+    if (hits == null)
+    {
+      System.err.println("No seq for " + sname);
+    }
+    double pvalue = (Double) hmmrhit.get("pvalue");
+
+    double evalue = Double.valueOf("" + hmmrhit.get("evalue"));
+    for (Object domainhit : ((JSONArray) hmmrhit.get("domains")))
+    {
+      JSONObject dhit=(JSONObject) domainhit;
+      // dhit.get(key)
+
+      // alihmmfrom,alihmmto alimodel
+      // alisqfrom,alisqto,aliaseq
+      // alippline
+      // 
+      for (SequenceI hitseq : hits)
+      {
+        // match alisqfrom,alisqto,seq
+        // overlay ppline as seqannotation
+        // ievalue
+        // cevalue
+        // 
+
+        AlignmentAnnotation pval = new AlignmentAnnotation("p-value",
+                "hmmer3 pvalue", pvalue);
+        AlignmentAnnotation eval = new AlignmentAnnotation("e-value",
+                "hmmer3 evalue", evalue);
+        hitseq.addAlignmentAnnotation(pval);
+        hitseq.addAlignmentAnnotation(eval);
+        
+
+      }
+    }
+  }
+}

diff --git a/test/jalview/ws/ebi/HmmerJSONProcessTest.java b/test/jalview/ws/ebi/HmmerJSONProcessTest.java
new file mode 100644 (file)
index 0000000..0c07584
--- /dev/null
+++ b/test/jalview/ws/ebi/HmmerJSONProcessTest.java
@@ -0,0 +1,89 @@
+package jalview.ws.ebi;
+
+import jalview.datamodel.AlignmentI;
+import jalview.io.FileParse;
+import jalview.io.FormatAdapter;
+
+import java.io.File;
+
+import org.json.simple.JSONObject;
+import org.json.simple.parser.JSONParser;
+import org.testng.Assert;
+import org.testng.annotations.Test;
+
+public class HmmerJSONProcessTest {
+  public static File alignmentFragFile = new File(
+          "examples/testdata/hmmer3/alignment_frag.fa.gz");
+
+  private AlignmentI getSearchResultFragmentAlignment() throws Exception
+  {
+    AlignmentI alf = new FormatAdapter().readFile(
+            alignmentFragFile.getAbsolutePath(), FormatAdapter.FILE,
+            "FASTA");
+
+    return alf;
+  }
+
+  public static File alignmentResultFile = new File(
+          "examples/testdata/hmmer3/alignment_res.fa.gz");
+
+  private AlignmentI getSearchResultAlignment() throws Exception
+  {
+    AlignmentI alf = new FormatAdapter().readFile(
+            alignmentResultFile.getAbsolutePath(), FormatAdapter.FILE,
+            "FASTA");
+
+    return alf;
+  }
+
+  public static String hitTestFile = "examples/testdata/hmmer3/hit_fragment.json.gz",
+          hmmerResultFile = "examples/testdata/hmmer3/hmmeresult.json.gz";
+
+
+  @Test(groups = { "Functional" })
+  public void parseHitTest() throws Exception
+  {
+
+    Assert.assertTrue(new File(hitTestFile).exists(),
+            "Can't find test data.\n"
+            + hitTestFile);
+    JSONParser jp = new JSONParser();
+    // read JSON in same way - via fileparse
+    Object hitfragment = jp.parse(new FileParse(hitTestFile,
+            FormatAdapter.FILE).getReader());
+    Assert.assertTrue((hitfragment instanceof JSONObject),
+            "Didn't find a JSON object map in " + hitTestFile);
+    AlignmentI searchResult = getSearchResultFragmentAlignment();
+
+    Assert.assertTrue(searchResult != null && searchResult.getHeight() > 0,
+            "Didn't read search result alignment from " + alignmentFragFile);
+
+    HmmerJSONProcessor hjsp = new HmmerJSONProcessor(searchResult);
+    hjsp.addHit((JSONObject) hitfragment, 1);
+    // check that
+    // scores, posterior probabilities and stuff exist.
+  }
+
+  @Test(groups = { "Functional" })
+  public void parseJsonResultTest() throws Exception
+  {
+
+    Assert.assertTrue(new File(hmmerResultFile).exists(),
+            "Can't find test data.\n" + hmmerResultFile);
+
+    AlignmentI searchResult = getSearchResultAlignment();
+
+    Assert.assertTrue(searchResult != null && searchResult.getHeight() > 0,
+            "Didn't read search result alignment from " + alignmentFragFile);
+
+    HmmerJSONProcessor hjsp = new HmmerJSONProcessor(searchResult);
+    hjsp.parseFrom(new FileParse(hmmerResultFile, FormatAdapter.FILE));
+    // check that
+    // scores, posterior probabilities and stuff exist.
+  }
+  // Groovy test
+  // def al = Jalview.getAlignFrames()[0].getViewport().getAlignment()
+  // new jalview.ws.ebi.HmmerJSONProcessor(al).parseFrom(new
+  // jalview.io.FileParse("examples/testdata/hmmer3/hmmeresult.json.gz","File"))
+
+}