From 06548fdf3d2dfa5532500b9c9f52f91b01ed87ad Mon Sep 17 00:00:00 2001
From: Shifu Chen <chen@haplox.com>
Date: Tue, 26 Jun 2018 18:19:56 +0800
Subject: [PATCH] report mean read length

---
 src/htmlreporter.cpp | 12 +++++++++++-
 src/jsonreporter.cpp |  6 ++++++
 src/stats.cpp        | 11 +++++++++++
 src/stats.h          |  2 ++
 4 files changed, 30 insertions(+), 1 deletion(-)

diff --git a/src/htmlreporter.cpp b/src/htmlreporter.cpp
index 23d1773..f84771d 100644
--- a/src/htmlreporter.cpp
+++ b/src/htmlreporter.cpp
@@ -113,6 +113,16 @@ void HtmlReporter::printSummary(ofstream& ofs, FilterResult* result, Stats* preS
     ofs << "<table class='summary_table'>\n";
     outputRow(ofs, "fastp version:", string(FASTP_VER)+ " (<a href='https://github.com/OpenGene/fastp'>https://github.com/OpenGene/fastp</a>)");
     outputRow(ofs, "sequencing:", sequencingInfo);
+
+    // report read length change
+    if(mOptions->isPaired()) {
+        outputRow(ofs, "mean length before filtering:", to_string(preStats1->getMeanLength()) + "bp, " + to_string(preStats2->getMeanLength()) + "bp");
+        outputRow(ofs, "mean length after filtering:", to_string(postStats1->getMeanLength()) + "bp, " + to_string(postStats2->getMeanLength()) + "bp");
+    } else  {
+        outputRow(ofs, "mean length before filtering:", to_string(preStats1->getMeanLength()) + "bp");
+        outputRow(ofs, "mean length after filtering:", to_string(postStats1->getMeanLength()) + "bp");
+    }
+
     if(mOptions->duplicate.enabled) {
         string dupStr = to_string(mDupRate*100) + "%";
         if(!mOptions->isPaired())
@@ -367,7 +377,7 @@ void HtmlReporter::printCSS(ofstream& ofs){
     ofs << "<style type=\"text/css\">" << endl;
     ofs << "td {border:1px solid #dddddd;padding:5px;font-size:12px;}" << endl;
     ofs << "table {border:1px solid #999999;padding:2x;border-collapse:collapse; width:800px}" << endl;
-    ofs << ".col1 {width:200px; font-weight:bold;}" << endl;
+    ofs << ".col1 {width:240px; font-weight:bold;}" << endl;
     ofs << ".adapter_col {width:500px; font-size:10px;}" << endl;
     ofs << "img {padding:30px;}" << endl;
     ofs << "#menu {font-family:Consolas, 'Liberation Mono', Menlo, Courier, monospace;}" << endl;
diff --git a/src/jsonreporter.cpp b/src/jsonreporter.cpp
index e4def40..25c5b36 100644
--- a/src/jsonreporter.cpp
+++ b/src/jsonreporter.cpp
@@ -76,6 +76,9 @@ void JsonReporter::report(FilterResult* result, Stats* preStats1, Stats* postSta
     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" << "\"read1_mean_length\":" << preStats1->getMeanLength() << "," << endl;
+    if(mOptions->isPaired())
+        ofs << "\t\t\t" << "\"read2_mean_length\":" << preStats2->getMeanLength() << "," << 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;
 
@@ -86,6 +89,9 @@ void JsonReporter::report(FilterResult* result, Stats* preStats1, Stats* postSta
     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" << "\"read1_mean_length\":" << postStats1->getMeanLength() << "," << endl;
+    if(mOptions->isPaired())
+        ofs << "\t\t\t" << "\"read2_mean_length\":" << postStats2->getMeanLength() << "," << 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" << "}";
 
diff --git a/src/stats.cpp b/src/stats.cpp
index a1ad703..63a7ac2 100644
--- a/src/stats.cpp
+++ b/src/stats.cpp
@@ -9,6 +9,7 @@ Stats::Stats(Options* opt, bool isRead2, int guessedCycles, int bufferMargin){
     mOptions = opt;
     mIsRead2 = isRead2;
     mReads = 0;
+    mLengthSum = 0;
 
     mEvaluatedSeqLen = mOptions->seqLen1;
     if(mIsRead2)
@@ -214,9 +215,18 @@ void Stats::summarize(bool forced) {
     summarized = true;
 }
 
+int Stats::getMeanLength() {
+    if(mReads == 0)
+        return 0.0;
+    else
+        return mLengthSum/mReads;
+}
+
 void Stats::statRead(Read* r) {
     int len = r->length();
 
+    mLengthSum += len;
+
     if(mBufLen < len) {
         extendBuffer(max(len + 100, (int)(len * 1.5)));
     }
@@ -888,6 +898,7 @@ Stats* Stats::merge(vector<Stats*>& list) {
         int curCycles =  list[t]->getCycles();
         // merge read number
         s->mReads += list[t]->mReads;
+        s->mLengthSum += list[t]->mLengthSum;
 
         // merge per cycle counting for different bases
         for(int i=0; i<8; i++){
diff --git a/src/stats.h b/src/stats.h
index a4f2b9c..f568b23 100644
--- a/src/stats.h
+++ b/src/stats.h
@@ -38,6 +38,7 @@ class Stats{
     void reportHtmlORA(ofstream& ofs, string filteringType, string readName);
     bool isLongRead();
     void initOverRepSeq();
+    int getMeanLength();
 
 public:
     static string list2string(double* list, int size);
@@ -93,6 +94,7 @@ class Stats{
     long mKmerMax;
     long mKmerMin;
     int mKmerBufLen;
+    int mLengthSum;
 };
 
 #endif
\ No newline at end of file