Changeset 3477

Timestamp:

Mar 9, 2016, 9:26:38 AM (8 years ago)

Author:

Peter

Message:

add functionality to access which RG IDs that are present in a BamHeader?

Location:

trunk

Files:

• test/bam_header.cc (modified) (1 diff)
• yat/omic/BamHeader.cc (modified) (3 diffs)
• yat/omic/BamHeader.h (modified) (5 diffs)

trunk/test/bam_header.cc

@@ -133 +133 @@
     }
 
+    typedef BamHeader::read_group_iterator iterator;
+    iterator it = hdr.read_group_begin();
+    if (it == hdr.read_group_end()) {
+      suite.add(false);
+      suite.err() << "it == hdr.read_group_end()\n";
+    }
+    else {
+      if (!suite.add(*it == "foo"))
+        suite.err() << "*it != foo\n";
+      std::map<std::string, std::string> m = hdr.read_group(*it);
+      std::string val = m["SM"];
+      if (!suite.add(val == "Tumour"))
+        suite.err() << "incorrect val: " << val << " expected Tumour\n";
+      val = m["PL"];
+      if (!suite.add(val == "ILLUMINA"))
+        suite.err() << "incorrect val: " << val << " expected ILLUMINA\n";
+    }
+
     std::string bwa_version = hdr.program_group("bwa", "VN");
     if (bwa_version!="0.6.1-r104") {

trunk/yat/omic/BamHeader.cc

@@ -81 +81 @@
                                       const std::string& id,
                                       const std::string& key) const
+  {
+    update_group(table, type);
+    strMap2::const_iterator line_table = table.find(id);
+    if (line_table == table.end()) {
+      std::string msg = "@" + type + ": unknown ID: '" + id + "'";
+      throw utility::runtime_error(msg);
+    }
+    strMap::const_iterator it = line_table->second.find(key);
+    if (it == line_table->second.end()) {
+      std::string msg = "@" + type + ": unknown KEY: '" + key + "'";
+      throw utility::runtime_error(msg);
+    }
+    return it->second;
+  }
+
+
+  void BamHeader::update_group(strMap2& table, const std::string& type) const
   {
     assert(type.size()==2);
@@ -101 +118 @@
       }
     }
-    strMap2::const_iterator line_table = table.find(id);
-    if (line_table == table.end()) {
-      std::string msg = "@" + type + ": unknown ID: '" + id + "'";
-      throw utility::runtime_error(msg);
-    }
-    strMap::const_iterator it = line_table->second.find(key);
-    if (it == line_table->second.end()) {
-      std::string msg = "@" + type + ": unknown KEY: '" + key + "'";
-      throw utility::runtime_error(msg);
-    }
-    return it->second;
   }
 
@@ -156 +162 @@
   {
     return group(read_group_, "RG", id, key);
+  }
+
+
+  const std::map<std::string, std::string>&
+  BamHeader::read_group(const std::string& id) const
+  {
+    update_group(read_group_, "RG");
+    strMap2::const_iterator line_table = read_group_.find(id);
+    if (line_table == read_group_.end()) {
+      std::string msg = "@RG: unknown ID: '" + id + "'";
+      throw utility::runtime_error(msg);
+    }
+    return line_table->second;
+  }
+
+
+  BamHeader::read_group_iterator BamHeader::read_group_begin(void) const
+  {
+    update_group(read_group_, "RG");
+    BamHeader::strMap2::const_iterator it(read_group_.begin());
+    return utility::pair_first_iterator(it);
+  }
+
+
+  BamHeader::read_group_iterator BamHeader::read_group_end(void) const
+  {
+    update_group(read_group_, "RG");
+    BamHeader::strMap2::const_iterator it(read_group_.end());
+    return utility::pair_first_iterator(it);
   }
 

trunk/yat/omic/BamHeader.h

@@ -27 +27 @@
 #include "yat/utility/config_public.h"
 
+#include "yat/utility/stl_utility.h"
+
 #include YAT_SAM_HEADER
+
+#include <boost/iterator/transform_iterator.hpp>
 
 #include <map>
@@ -56 +60 @@
   class BamHeader
   {
+    typedef std::map<std::string, std::string> strMap;
+    typedef std::map<std::string, strMap> strMap2;
   public:
     /**
@@ -112 +118 @@
        and for this line \c read_group("foo", "SM") returns \c "Tumour"
 
+       To see which IDs are available in the header, iterate through
+       the set of IDs with read_group_begin() (and read_group_end()).
+
        \return value for \a key for line with ID \a id.
 
        \since New in yat 0.13
     */
-    //
     const std::string& read_group(const std::string& id,
                                   const std::string& key) const;
+
+    /**
+       \returns a map describing an \@RG line such as
+
+       \code @RG  ID:foo  PL:ILLUMINA SM:Tumour \endcode
+
+       for which the returned map contains two entries: one with key
+       \c PL and value \c ILLUMINA, and one with key \c SM and value
+       \c Tumour.
+
+       \since New in yat 0.14
+     */
+    const std::map<std::string, std::string>&
+    read_group(const std::string& id) const;
+
+    /**
+       read_group_iterator is used to iterate through read groups
+
+       \see read_group_begin
+       \see read_group_end
+
+       \since New in yat 0.14
+     */
+    typedef boost::transform_iterator<
+      utility::PairFirst<const strMap2::value_type>,
+      strMap2::const_iterator>
+    read_group_iterator;
+
+    /**
+       \return first of available RG IDs
+
+       \since New in yat 0.14
+     */
+    read_group_iterator read_group_begin(void) const;
+
+    /**
+       \return end of range of available RG IDs
+
+       \see read_group_begin()
+
+       \since New in yat 0.14
+     */
+    read_group_iterator read_group_end(void) const;
 
     /**
@@ -179 +230 @@
 #endif
     bam_hdr_t* header_;
-    typedef std::map<std::string, std::string> strMap;
-    typedef std::map<std::string, strMap> strMap2;
     mutable strMap2 read_group_;
     mutable strMap2 program_group_;
@@ -191 +240 @@
                              const std::string& key) const;
 
+    void update_group(strMap2& map, const std::string& type) const;
+
     // using compiler generated copy and assignment
   };