Major update to BamTools version 1.0

[bamtools.git] / src / toolkit / bamtools_stats.cpp

diff --git a/src/toolkit/bamtools_stats.cpp b/src/toolkit/bamtools_stats.cpp
index 68ca6d3e185a29ce1e95d08caa5943a1cf6155db..42e7cbceb4d3a14a4bdefebd9a0f3fe617e1edda 100644 (file)
--- a/src/toolkit/bamtools_stats.cpp
+++ b/src/toolkit/bamtools_stats.cpp
@@ -3,11 +3,17 @@
 // Marth Lab, Department of Biology, Boston College
 // All rights reserved.
 // ---------------------------------------------------------------------------
-// Last modified: 22 July 2010
+// Last modified: 21 March 2011
 // ---------------------------------------------------------------------------
 // Prints general alignment statistics for BAM file(s).
 // ***************************************************************************
 
+#include "bamtools_stats.h"
+
+#include <api/BamMultiReader.h>
+#include <utils/bamtools_options.h>
+using namespace BamTools;
+
 #include <cmath>
 #include <algorithm>
 #include <functional>
@@ -15,12 +21,7 @@
 #include <numeric>
 #include <string>
 #include <vector>
-
-#include "bamtools_stats.h"
-#include "bamtools_options.h"
-#include "BamMultiReader.h"
 using namespace std;
-using namespace BamTools;
 
 // ---------------------------------------------
 // StatsSettings implementation
@@ -99,19 +100,20 @@ StatsTool::StatsToolPrivate::StatsToolPrivate(StatsTool::StatsSettings* _setting
 
 StatsTool::StatsToolPrivate::~StatsToolPrivate(void) { }
 
-bool StatsTool::StatsToolPrivate::CalculateMedian(vector<int>& data, double& median) { // median is double in case of even data size, need to return average of middle 2 elements
+// median is of type double because in the case of even number of data elements,
+// we need to return the average of middle 2 elements
+bool StatsTool::StatsToolPrivate::CalculateMedian(vector<int>& data, double& median) { 
   
-    // check that data exists
+    // skip if data empty
     if ( data.empty() ) return false;
-  
-    size_t dataSize = data.size();
-    size_t middleIndex = dataSize / 2;
-    
+
+    // find middle element
+    size_t middleIndex = data.size() / 2;
     vector<int>::iterator target = data.begin() + middleIndex;
     nth_element(data.begin(), target, data.end());
     
     // odd number of elements
-    if ( (dataSize % 2) != 0) {
+    if ( (data.size() % 2) != 0) {
         median = (double)(*target);
         return true;
     }
@@ -215,25 +217,22 @@ void StatsTool::StatsToolPrivate::ProcessAlignment(const BamAlignment& al) {
 
 bool StatsTool::StatsToolPrivate::Run() {
   
-    // opens the BAM files without checking for indexes 
+    // open the BAM files
     BamMultiReader reader;
-    if ( !reader.Open(settings->InputFiles, false, true) ) {
-        cerr << "Could not open input BAM file(s)... quitting." << endl;
+    if ( !reader.Open(settings->InputFiles) ) {
+        cerr << "bamtools stats ERROR: could not open input BAM file(s)... Aborting." << endl;
         reader.Close();
         return false;
     }
     
-    // plow through file, keeping track of stats
+    // plow through alignments, keeping track of stats
     BamAlignment al;
-    while ( reader.GetNextAlignmentCore(al) ) {
+    while ( reader.GetNextAlignmentCore(al) )
         ProcessAlignment(al);
-    }
+    reader.Close();
     
-    // print stats
+    // print stats & exit
     PrintStats();
-    
-    // clean and exit
-    reader.Close();
     return true; 
 }
 
@@ -246,7 +245,7 @@ StatsTool::StatsTool(void)
     , m_impl(0)
 {
     // set program details
-    Options::SetProgramInfo("bamtools stats", "prints general alignment statistics", "[-in <filename> -in <filename> ... ]");
+    Options::SetProgramInfo("bamtools stats", "prints general alignment statistics", "[-in <filename> -in <filename> ...] [statsOptions]");
     
     // set up options 
     OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
@@ -273,12 +272,12 @@ int StatsTool::Run(int argc, char* argv[]) {
   
     // parse command line arguments
     Options::Parse(argc, argv, 1);
-    
+
     // set to default input if none provided
     if ( !m_settings->HasInput ) 
         m_settings->InputFiles.push_back(Options::StandardIn());
     
-    // run internal SortTool implementation, return success/fail
+    // run internal StatsTool implementation, return success/fail
     m_impl = new StatsToolPrivate(m_settings);
     
     if ( m_impl->Run() ) return 0;