support JSON report

src/common.h

@@ -1,7 +1,7 @@
 #ifndef COMMON_H
 #define COMMON_H
 
-#define FUSIONSCAN_VER "0.4.0"
+#define FUSIONSCAN_VER "0.5.0"
 
 #define _DEBUG true
 

src/fusionscan.cpp

@@ -6,23 +6,24 @@
 #include "pescanner.h"
 #include "util.h"
 
-FusionScan::FusionScan(string fusionFile, string refFile, string read1File, string read2File, string html, int threadNum){
+FusionScan::FusionScan(string fusionFile, string refFile, string read1File, string read2File, string html, string json, int threadNum){
     mRead1File = read1File;
     mRead2File = read2File;
     mFusionFile = fusionFile;
     mRefFile = refFile;
     mHtmlFile = html;
+    mJsonFile = json;
     mThreadNum = threadNum;
 }
 
 bool FusionScan::scan(){
     vector<Fusion> fusions = Fusion::parseCsv(mFusionFile);
     if(mRead2File != ""){
-        PairEndScanner pescanner( mFusionFile, mRefFile, mRead1File, mRead2File, mHtmlFile, mThreadNum);
+        PairEndScanner pescanner( mFusionFile, mRefFile, mRead1File, mRead2File, mHtmlFile, mJsonFile, mThreadNum);
         return pescanner.scan();
     }
     else{
-        SingleEndScanner sescanner( mFusionFile, mRefFile, mRead1File, mHtmlFile, mThreadNum);
+        SingleEndScanner sescanner( mFusionFile, mRefFile, mRead1File, mHtmlFile, mJsonFile, mThreadNum);
         return sescanner.scan();
     }
 }

src/fusionscan.h

@@ -12,14 +12,15 @@ using namespace std;
 
 class FusionScan{
 public:
-    FusionScan(string fusionFile, string refFile, string read1File, string read2File, string html, int threadNum);
+    FusionScan(string fusionFile, string refFile, string read1File, string read2File, string html, string json, int threadNum);
     bool scan();
 
 private:
     string mFusionFile;
     string mRead1File;
     string mRead2File;
     string mHtmlFile;
+    string mJsonFile;
     string mRefFile;
     int mThreadNum;
 };

src/jsonreporter.cpp

@@ -0,0 +1,82 @@
+#include "jsonreporter.h"
+#include "common.h"
+#include <chrono>
+#include "globalsettings.h"
+
+JsonReporter::JsonReporter(string filename, FusionMapper* mapper){
+    mFusionMapper = mapper;
+    mFusionResults = mapper->mFusionResults;
+    mFilename = filename;
+    mFile.open(mFilename.c_str(), ifstream::out);
+}
+
+JsonReporter::~JsonReporter(){
+    mFile.close();
+}
+
+extern string getCurrentSystemTime();
+extern string command;
+
+void JsonReporter::run() {
+    mFile << "{" << endl;
+    mFile << "\t\"command\":\"" << command << "\"," << endl;
+    mFile << "\t\"version\":\"" << FUSIONSCAN_VER << "\"," << endl;
+    mFile << "\t\"time\":\"" << getCurrentSystemTime() << "\"," << endl;
+    mFile << "\t\"mutations\":{";
+
+    bool isFirstMut = true;
+    for(int i=0;i<mFusionResults.size();i++){
+        FusionResult fusion = mFusionResults[i];
+        vector<Match*> matches = fusion.mMatches;
+
+        if(isFirstMut) {
+            mFile << endl;
+            isFirstMut = false;
+        }
+        else
+            mFile << "," << endl;
+
+        mFile << "\t\t\"" <<  fusion.mTitle << "\":{" << endl;
+            mFile << "\t\t\t\"" <<  "left" << "\":{" << endl;
+                mFile << "\t\t\t\t\"" <<  "gene_name" << "\":" << "\"" << fusion.mLeftGene.mName << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "gene_chr" << "\":" << "\"" << fusion.mLeftGene.mChr << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "position" << "\":" << fusion.mLeftGP.position << "," << endl;
+                mFile << "\t\t\t\t\"" <<  "reference" << "\":" << "\"" << fusion.mLeftRef << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "ref_ext" << "\":" << "\"" << fusion.mLeftRefExt << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "pos_str" << "\":" << "\"" << fusion.mLeftPos << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "exon_or_intron" << "\":" << "\"" << (fusion.mLeftIsExon?"exon":"intron") << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "exon_or_intron_id" << "\":" << fusion.mLeftExonOrIntronID << "," << endl;
+                mFile << "\t\t\t\t\"" <<  "strand" << "\":" << "\"" << (fusion.isLeftProteinForward()?"forward":"reversed") << "\"" << endl;
+            mFile << "\t\t\t}, " << endl;
+            mFile << "\t\t\t\"" <<  "right" << "\":{" << endl;
+                mFile << "\t\t\t\t\"" <<  "gene_name" << "\":" << "\"" << fusion.mRightGene.mName << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "gene_chr" << "\":" << "\"" << fusion.mRightGene.mChr << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "position" << "\":" << fusion.mRightGP.position << "," << endl;
+                mFile << "\t\t\t\t\"" <<  "reference" << "\":" << "\"" << fusion.mRightRef << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "ref_ext" << "\":" << "\"" << fusion.mRightRefExt << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "pos_str" << "\":" << "\"" << fusion.mRightPos << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "exon_or_intron" << "\":" << "\"" << (fusion.mRightIsExon?"exon":"intron") << "\"," << endl;
+                mFile << "\t\t\t\t\"" <<  "exon_or_intron_id" << "\":" << fusion.mRightExonOrIntronID << "," << endl;
+                mFile << "\t\t\t\t\"" <<  "strand" << "\":" << "\"" << (fusion.isRightProteinForward()?"forward":"reversed") << "\"" << endl;
+            mFile << "\t\t\t}, " << endl;
+
+        mFile << "\t\t\t\"" <<  "unique" << "\":" << fusion.mUnique << "," << endl;
+        mFile << "\t\t\t\"" <<  "reads" << "\":[" << endl;
+        for(int m=0; m<matches.size(); m++){
+            mFile << "\t\t\t\t{" << endl;
+            mFile << "\t\t\t\t\t\"break\":" << matches[m]->mReadBreak << "," << endl;
+            mFile << "\t\t\t\t\t\"" << "strand" << "\":" << "\"" << (matches[m]->mReversed?"reversed":"forward") << "\"," << endl;
+            matches[m]->printReadToJson(mFile, "\t\t\t\t\t");
+            mFile << "\t\t\t\t}";
+            if(m!=matches.size()-1)
+                mFile << ",";
+            mFile << endl;
+        }
+        mFile << "\t\t\t]" << endl;
+        mFile << "\t\t}";
+    }
+
+    mFile << endl;
+    mFile << "\t}" << endl;
+    mFile << "}" << endl;
+}

src/jsonreporter.h

@@ -0,0 +1,29 @@
+#ifndef JSON_REPORTER_H
+#define JSON_REPORTER_H
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string>
+#include "read.h"
+#include "fusion.h"
+#include "match.h"
+#include <iostream>
+#include <fstream>
+#include "fusionmapper.h"
+
+using namespace std;
+
+class JsonReporter{
+public:
+    JsonReporter(string filename, FusionMapper* mapper);
+    ~JsonReporter();
+    void run();
+
+private:
+    string mFilename;
+    FusionMapper* mFusionMapper;
+    ofstream mFile;
+    vector<FusionResult> mFusionResults;
+};
+
+#endif

src/main.cpp

@@ -23,13 +23,15 @@ int main(int argc, char* argv[]){
     cmd.add<string>("ref", 'r', "reference fasta file name", true, "");
     cmd.add<int>("unique", 'u', "specify the least supporting read number is required to report a fusion, default is 2", false, 2);
     cmd.add<int>("deletion", 'd', "specify the least deletion length of a intra-gene deletion to report, default is 50", false, 50);
-    cmd.add<string>("html", 'h', "file name to store html report, default is genefuse.html", false, "genefuse.html");
+    cmd.add<string>("html", 'h', "file name to store HTML report, default is genefuse.html", false, "genefuse.html");
+    cmd.add<string>("json", 'j', "file name to store JSON report, default is genefuse.json", false, "genefuse.json");
     cmd.add<int>("thread", 't', "worker thread number, default is 4", false, 4);
     cmd.parse_check(argc, argv);
     string r1file = cmd.get<string>("read1");
     string r2file = cmd.get<string>("read2");
     string fusionFile = cmd.get<string>("fusion");
     string html = cmd.get<string>("html");
+    string json = cmd.get<string>("json");
     string refFile = cmd.get<string>("ref");
     int threadNum = cmd.get<int>("thread");
     int unique = cmd.get<int>("unique");
@@ -61,7 +63,7 @@ int main(int argc, char* argv[]){
 
     time_t t1 = time(NULL);
 
-    FusionScan fs(fusionFile, refFile, r1file, r2file, html, threadNum);
+    FusionScan fs(fusionFile, refFile, r1file, r2file, html, json, threadNum);
     fs.scan();
 
     time_t t2 = time(NULL);

src/match.cpp

@@ -87,3 +87,8 @@ int Match::countUnique(vector<Match*>& matches) {
     }
     return count;
 }
+
+void Match::printReadToJson(ofstream& file, string pad) {
+    file << pad << "\"seq\":" << "\"" <<  mRead->mSeq.mStr << "\"," << endl;
+    file << pad << "\"qual\":" << "\"" <<  mRead->mQuality << "\"" << endl;
+}

src/match.h

@@ -26,6 +26,7 @@ class Match{
     Match(Read* r, int readBreak, GenePos leftGP, GenePos rightGP, int gap, bool reversed = false);
     ~Match();
     void print();
+    void printReadToJson(ofstream& file, string pad);
     void printHtmlTD(ofstream& file);
     void printReadsToFile(ofstream& file);
     void setReversed(bool flag);

src/pescanner.cpp

@@ -7,13 +7,15 @@
 #include <thread>
 #include <memory.h>
 #include "util.h"
+#include "jsonreporter.h"
 
-PairEndScanner::PairEndScanner(string fusionFile, string refFile, string read1File, string read2File, string html, int threadNum){
+PairEndScanner::PairEndScanner(string fusionFile, string refFile, string read1File, string read2File, string html, string json, int threadNum){
     mRead1File = read1File;
     mRead2File = read2File;
     mFusionFile = fusionFile;
     mRefFile = refFile;
     mHtmlFile = html;
+    mJsonFile = json;
     mProduceFinished = false;
     mThreadNum = threadNum;
     mFusionMapper = NULL;
@@ -53,6 +55,7 @@ bool PairEndScanner::scan(){
     mFusionMapper->clusterMatches();
 
     htmlReport();
+    jsonReport();
 
     mFusionMapper->freeMatches();
     return true;
@@ -266,3 +269,11 @@ void PairEndScanner::htmlReport() {
     HtmlReporter reporter(mHtmlFile, mFusionMapper);
     reporter.run();
 }
+
+void PairEndScanner::jsonReport() {
+    if(mJsonFile == "")
+        return;
+
+    JsonReporter reporter(mJsonFile, mFusionMapper);
+    reporter.run();
+}

src/pescanner.h

@@ -38,11 +38,12 @@ typedef struct ReadPairRepository ReadPairRepository;
 
 class PairEndScanner{
 public:
-    PairEndScanner(string fusionFile, string refFile, string read1File, string read2File, string html="", int threadnum=1);
+    PairEndScanner(string fusionFile, string refFile, string read1File, string read2File, string html, string json, int threadnum);
     ~PairEndScanner();
     bool scan();
     void textReport();
     void htmlReport();
+    void jsonReport();
 
 private:
     bool scanPairEnd(ReadPairPack* pack);
@@ -60,6 +61,7 @@ class PairEndScanner{
     string mRead1File;
     string mRead2File;
     string mHtmlFile;
+    string mJsonFile;
     ReadPairRepository mRepo;
     bool mProduceFinished;
     std::mutex mFusionMtx;

src/read.cpp

@@ -229,6 +229,8 @@ Read* ReadPair::fastMerge(){
 			} else {
 				// add the quality of the pair to make a high qual
 				mergedQual[offset+i] =  qual1[offset+i] + qual2[i] - 33;
+				if(mergedQual[offset+i] >= 'Z')
+					mergedQual[offset+i] = 'Z';
 			}
 		}
 		delete rcRight;

src/sescanner.cpp

@@ -7,12 +7,14 @@
 #include <thread>
 #include <memory.h>
 #include "util.h"
+#include "jsonreporter.h"
 
-SingleEndScanner::SingleEndScanner(string fusionFile, string refFile, string read1File, string html, int threadNum){
+SingleEndScanner::SingleEndScanner(string fusionFile, string refFile, string read1File, string html, string json, int threadNum){
     mRead1File = read1File;
     mFusionFile = fusionFile;
     mRefFile = refFile;
     mHtmlFile = html;
+    mJsonFile = json;
     mProduceFinished = false;
     mThreadNum = threadNum;
     mFusionMapper = NULL;
@@ -51,8 +53,8 @@ bool SingleEndScanner::scan(){
     mFusionMapper->sortMatches();
     mFusionMapper->clusterMatches();
 
-    textReport();
     htmlReport();
+    jsonReport();
 
     mFusionMapper->freeMatches();
 
@@ -224,3 +226,11 @@ void SingleEndScanner::htmlReport() {
     HtmlReporter reporter(mHtmlFile, mFusionMapper);
     reporter.run();
 }
+
+void SingleEndScanner::jsonReport() {
+    if(mJsonFile == "")
+        return;
+
+    JsonReporter reporter(mJsonFile, mFusionMapper);
+    reporter.run();
+}

src/sescanner.h

@@ -38,11 +38,12 @@ typedef struct ReadRepository ReadRepository;
 
 class SingleEndScanner{
 public:
-    SingleEndScanner(string fusionFile, string refFile, string read1File, string html="", int threadnum=1);
+    SingleEndScanner(string fusionFile, string refFile, string read1File, string html, string json, int threadnum);
     ~SingleEndScanner();
     bool scan();
     void textReport();
     void htmlReport();
+    void jsonReport();
 
 private:
     bool scanSingleEnd(ReadPack* pack);
@@ -60,6 +61,7 @@ class SingleEndScanner{
     string mRead1File;
     string mRead2File;
     string mHtmlFile;
+    string mJsonFile;
     ReadRepository mRepo;
     bool mProduceFinished;
     std::mutex mFusionMtx;