show GC content in summary table

#15
OpenGene · Dec 7, 2017 · a50468c · a50468c
1 parent 58c8636
commit a50468c
Show file tree

Hide file tree

Showing 4 changed files with 31 additions and 4 deletions.
diff --git a/src/htmlreporter.cpp b/src/htmlreporter.cpp
@@ -58,6 +58,10 @@ void HtmlReporter::printSummary(ofstream& ofs, FilterResult* result, Stats* preS
     if(preStats2)
         pre_q30_bases += preStats2->getQ30();
 
+    long pre_total_gc = preStats1->getGCNumber();
+    if(preStats2)
+        pre_total_gc += preStats2->getGCNumber();
+
     long post_total_reads = postStats1->getReads();
     if(postStats2)
         post_total_reads += postStats2->getReads();
@@ -74,6 +78,10 @@ void HtmlReporter::printSummary(ofstream& ofs, FilterResult* result, Stats* preS
     if(postStats2)
         post_q30_bases += postStats2->getQ30();
 
+    long post_total_gc = postStats1->getGCNumber();
+    if(postStats2)
+        post_total_gc += postStats2->getGCNumber();
+
     string sequencingInfo  = mOptions->isPaired()?"paired end":"single end";
     if(mOptions->isPaired()) {
         sequencingInfo += " (" + to_string(preStats1->getCycles()) + " cycles + " + to_string(preStats2->getCycles()) + " cycles)";
@@ -98,6 +106,7 @@ void HtmlReporter::printSummary(ofstream& ofs, FilterResult* result, Stats* preS
     outputRow(ofs, "total bases:", formatNumber(pre_total_bases));
     outputRow(ofs, "Q20 bases:", formatNumber(pre_q20_bases) + " (" + getPercents(pre_q20_bases,pre_total_bases) + "%)");
     outputRow(ofs, "Q30 bases:", formatNumber(pre_q30_bases) + " (" + getPercents(pre_q30_bases, pre_total_bases) + "%)");
+    outputRow(ofs, "GC content:", getPercents(pre_total_gc,pre_total_bases) + "%");
     ofs << "</table>\n";
 
     ofs << "<div class='subsection_title'>After filtering</div>\n";
@@ -106,6 +115,7 @@ void HtmlReporter::printSummary(ofstream& ofs, FilterResult* result, Stats* preS
     outputRow(ofs, "total bases:", formatNumber(post_total_bases));
     outputRow(ofs, "Q20 bases:", formatNumber(post_q20_bases) + " (" + getPercents(post_q20_bases, post_total_bases) + "%)");
     outputRow(ofs, "Q30 bases:", formatNumber(post_q30_bases) + " (" + getPercents(post_q30_bases, post_total_bases) + "%)");
+    outputRow(ofs, "GC content:", getPercents(post_total_gc,post_total_bases) + "%");
     ofs << "</table>\n";
 
     if(result) {

diff --git a/src/jsonreporter.cpp b/src/jsonreporter.cpp
@@ -30,6 +30,10 @@ void JsonReporter::report(FilterResult* result, Stats* preStats1, Stats* postSta
     if(preStats2)
         pre_q30_bases += preStats2->getQ30();
 
+    long pre_total_gc = preStats1->getGCNumber();
+    if(preStats2)
+        pre_total_gc += preStats2->getGCNumber();
+
     long post_total_reads = postStats1->getReads();
     if(postStats2)
         post_total_reads += postStats2->getReads();
@@ -46,6 +50,10 @@ void JsonReporter::report(FilterResult* result, Stats* preStats1, Stats* postSta
     if(postStats2)
         post_q30_bases += postStats2->getQ30();
 
+    long post_total_gc = postStats1->getGCNumber();
+    if(postStats2)
+        post_total_gc += postStats2->getGCNumber();
+
     // summary
     ofs << "\t" << "\"summary\": {" << endl;
 
@@ -55,7 +63,8 @@ void JsonReporter::report(FilterResult* result, Stats* preStats1, Stats* postSta
     ofs << "\t\t\t" << "\"q20_bases\":" << pre_q20_bases << "," << endl; 
     ofs << "\t\t\t" << "\"q30_bases\":" << pre_q30_bases << "," << endl; 
     ofs << "\t\t\t" << "\"q20_rate\":" << (pre_total_bases == 0?0.0:(double)pre_q20_bases / (double)pre_total_bases) << "," << endl; 
-    ofs << "\t\t\t" << "\"q30_rate\":" << (pre_total_bases == 0?0.0:(double)pre_q30_bases / (double)pre_total_bases)  << endl; 
+    ofs << "\t\t\t" << "\"q30_rate\":" << (pre_total_bases == 0?0.0:(double)pre_q30_bases / (double)pre_total_bases) << "," << endl; 
+    ofs << "\t\t\t" << "\"gc_content\":" << (pre_total_bases == 0?0.0:(double)pre_total_gc / (double)pre_total_bases)  << endl; 
     ofs << "\t\t" << "}," << endl;
 
     ofs << "\t\t" << "\"after_filtering\": {" << endl;
@@ -64,7 +73,8 @@ void JsonReporter::report(FilterResult* result, Stats* preStats1, Stats* postSta
     ofs << "\t\t\t" << "\"q20_bases\":" << post_q20_bases << "," << endl; 
     ofs << "\t\t\t" << "\"q30_bases\":" << post_q30_bases << "," << endl; 
     ofs << "\t\t\t" << "\"q20_rate\":" << (post_total_bases == 0?0.0:(double)post_q20_bases / (double)post_total_bases) << "," << endl; 
-    ofs << "\t\t\t" << "\"q30_rate\":" << (post_total_bases == 0?0.0:(double)post_q30_bases / (double)post_total_bases) << endl; 
+    ofs << "\t\t\t" << "\"q30_rate\":" << (post_total_bases == 0?0.0:(double)post_q30_bases / (double)post_total_bases) << "," << endl; 
+    ofs << "\t\t\t" << "\"gc_content\":" << (post_total_bases == 0?0.0:(double)post_total_gc / (double)post_total_bases)  << endl; 
     ofs << "\t\t" << "}" << endl;
 
     ofs << "\t" << "}," << endl;

diff --git a/src/stats.cpp b/src/stats.cpp
@@ -328,6 +328,12 @@ long Stats::getQ30() {
     return mQ30Total;
 }
 
+long Stats::getGCNumber() {
+    if(!summarized)
+        summarize();
+    return mBaseContents['G' & 0x07] + mBaseContents['C' & 0x07];
+}
+
 void Stats::print() {
     if(!summarized) {
         summarize();
@@ -474,8 +480,8 @@ void Stats::reportHtmlKMER(ofstream& ofs, string filteringType, string readName)
     string title = "";
 
     ofs << "<div class='subsection_title'>" + subsection + "</div>\n";
-    ofs << "<div class='sub_section_tips'>Darker background means higher counts. The count will be shown on mouse over.</div>\n";
-    ofs << "<table class='kmer_table'>\n";
+    ofs << "<div class='sub_section_tips'>Darker background means larger counts. The count will be shown on mouse over.</div>\n";
+    ofs << "<table class='kmer_table' style='width:680px;'>\n";
     ofs << "<tr>";
     ofs << "<td></td>";
     // the heading row

diff --git a/src/stats.h b/src/stats.h
@@ -20,6 +20,7 @@ class Stats{
     long getBases();
     long getQ20();
     long getQ30();
+    long getGCNumber();
     // by default the qualified qual score is Q20 ('5')
     void statRead(Read* r, int& lowQualNum, int& nBaseNum, char qualifiedQual = '5');