Changeset 5729

Timestamp:

Nov 18, 2019, 1:43:09 PM (3 years ago)

Author:

Nicklas Nordborg

Message:

References #1199: Implement Variant calling pipeline

Added support for parsing the data from INFO column in VCF files. An InfoFactory implementation is needed that knows how to extract data into a more usable format. A simple implementation that extract everything into string key/value pairs have been provided.

Location:

extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/vcf

Files:

• AllInfoFactory.java (added)
• InfoData.java (added)
• InfoFactory.java (added)
• SnpData.java (modified) (3 diffs)
• VcfParser.java (modified) (5 diffs)

extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/vcf/SnpData.java

@@ -23 +23 @@
   private String alt;
   
-
+  private InfoData info;
+  
   /**
     Creates a new instance.
@@ -107 +108 @@
   }
   
+  public void setInfo(InfoData info)
+  {
+    this.info = info;
+  }
+  
+  public InfoData getInfo()
+  {
+    return info;
+  }
+
   /**
     Get all information as a JSON object. A new JSON object 
@@ -119 +130 @@
     json.put("ref", getRef());
     json.put("alt", getAlt());
+    if (info != null) 
+    {
+      json.put("info", info.asJSONObject());
+    }
     return json;
   }

extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/vcf/VcfParser.java

@@ -30 +30 @@
   private final Map<String, SnpData> snpDef;
   private boolean useLineNoAsId;
+  private InfoFactory infoFactory;
   
   /**
@@ -40 +41 @@
   
   /**
-    Set this to TRUE to use line number as the id. Usefule if there
+    Set this to TRUE to use line number as the id. Useful if there
     are no ID values in the file, but note that this means that 
     different VCF files can't be compared.
+    @since 4.24
   */
   public void setUseLineNoAsId(boolean useLineNoAsId)
   {
     this.useLineNoAsId = useLineNoAsId;
+  }
+  
+  /**
+    Set the factory to use for extracting information from
+    the INFO column. If not set, the INFO column is not parsed.
+    @since 4.24
+  */
+  public void setInfoFactory(InfoFactory infoFactory)
+  {
+    this.infoFactory = infoFactory;
   }
   
@@ -76 +88 @@
       throw new IOException("File '" + fileName + "' line " + lineNo + ": Could not find header line starting with '#CHROM{tab}POS...'");
     }
+    
+    if (infoFactory != null)
+    {
+      for (String name : ffp.getHeaderNames())
+      {
+        infoFactory.addInfoHeader(name, ffp.getHeader(name));
+      }
+    }
+    
     List<String> headers = ffp.getColumnHeaders();
     
@@ -110 +131 @@
         snp.setAlt(altMapper.getString(line));
         snpDef.put(id, snp);
+        if (infoFactory != null)
+        {
+          snp.setInfo(infoFactory.getInfo(infoMapper.getString(line), snp));
+        }
       }
       
@@ -146 +171 @@
     ffp.setDataHeaderRegexp(Pattern.compile("#CHROM\\tPOS.*"));
     ffp.setDataSplitterRegexp(Pattern.compile("\\t"));
+    ffp.setHeaderRegexp(Pattern.compile("##INFO\\=\\<ID\\=(\\w+),(.*)\\>"));
     ffp.setIgnoreRegexp(Pattern.compile("##.*"));
     return ffp;