Changeset 3410 for trunk/yat/omic/BamHeader.cc

Timestamp:

Apr 22, 2015, 1:05:43 AM (8 years ago)

Author:

Peter

Message:

add two new functions read_group and program_group to access data in BamHeader?. closes #817

File:

• trunk/yat/omic/BamHeader.cc (modified) (5 diffs)

trunk/yat/omic/BamHeader.cc

@@ -25 +25 @@
 
 #include "yat/utility/Exception.h"
+#include "yat/utility/split.h"
 
 // we need to include 'sam_header.h' when compiling against libbam
@@ -67 +68 @@
 
   BamHeader::BamHeader(const BamHeader& other)
+    : read_group_(other.read_group_), program_group_(other.program_group_)
   {
 #ifndef YAT_HAVE_HTSLIB
@@ -72 +74 @@
 #endif
     header_ = bam_hdr_dup(other.header_);
+  }
+
+
+  const std::string& BamHeader::group(strMap2& table,const std::string& type,
+                                      const std::string& id,
+                                      const std::string& key) const
+  {
+    assert(type.size()==2);
+    if (table.empty()) { // parse header
+      std::istringstream ss(text());
+      std::string line;
+      while (getline(ss, line)) {
+        if (line.substr(1,2)!=type)
+          continue;
+        // find ID value
+        size_t pos = line.find("\tID:", 3);
+        size_t end = line.find('\t', pos+4);
+        std::string id_val = line.substr(pos+4, end-(pos+4));
+        std::vector<std::string> vec;
+        utility::split(vec, line, '\t');
+        strMap& m = table[id_val];
+        for (size_t i=1; i<vec.size(); ++i) {
+          m[vec[i].substr(0,2)] = vec[i].substr(3);
+        }
+      }
+      if (table.empty())
+        table[""][""]; // to avoid re-parsing
+    }
+    return table[id][key];
   }
 
@@ -105 +136 @@
 
 
+  const std::string& BamHeader::program_group(const std::string& id,
+                                              const std::string& key) const
+  {
+    return group(program_group_, "PG", id, key);
+  }
+
+
+  const std::string& BamHeader::read_group(const std::string& id,
+                                           const std::string& key) const
+  {
+    return group(read_group_, "RG", id, key);
+  }
+
+
   void BamHeader::swap(BamHeader& other)
   {
     std::swap(header_, other.header_);
+    std::swap(read_group_, other.read_group_);
+    std::swap(program_group_, other.program_group_);
   }
 
@@ -134 +181 @@
   void BamHeader::text(const std::string& txt)
   {
+    read_group_.clear();
+    program_group_.clear();
 #ifdef YAT_HAVE_HTSLIB
     // destroy the old