split multi-alt variant to multiple single-alt variants

OpenGene · Oct 22, 2017 · 4fddf2f · 4fddf2f
1 parent 7e2e3eb
commit 4fddf2f
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 20 deletions.
diff --git a/src/pescanner.cpp b/src/pescanner.cpp
@@ -283,6 +283,7 @@ void PairEndScanner::consumerTask()
 void PairEndScanner::textReport(vector<Mutation>& mutationList, vector<Match*> *mutationMatches) {
     //output result
     bool found = false;
+    int undetected = 0;
     for(int i=0;i<mutationList.size();i++){
         vector<Match*> matches = mutationMatches[i];
         if(matches.size()>=GlobalSettings::minReadSupport){
@@ -293,14 +294,16 @@ void PairEndScanner::textReport(vector<Mutation>& mutationList, vector<Match*> *
                 cout<<m+1<<", ";
                 matches[m]->print(mutationList[i].mLeft.length(), mutationList[i].mCenter.length(), mutationList[i].mRight.length());
             }
+        } else {
+            undetected++;
         }
     }
     if(found == false) {
         cout << "MutScan didn't find any mutation" << endl;
     }
     // if processing VCF, output those with no supporting reads found
-    if(GlobalSettings::processingVCF) {
-        cerr << "Following mutations are not detected" << endl;
+    if(GlobalSettings::processingVCF && undetected>0) {
+        cerr << undetected << " mutations of this VCF are not detected" << endl;
         for(int i=0;i<mutationList.size();i++){
             vector<Match*> matches = mutationMatches[i];
             if(matches.size()<GlobalSettings::minReadSupport){

diff --git a/src/sescanner.cpp b/src/sescanner.cpp
@@ -255,6 +255,7 @@ void SingleEndScanner::consumerTask()
 void SingleEndScanner::textReport(vector<Mutation>& mutationList, vector<Match*> *mutationMatches) {
     //output result
     bool found = false;
+    int undetected = 0;
     for(int i=0;i<mutationList.size();i++){
         vector<Match*> matches = mutationMatches[i];
         if(matches.size()>=GlobalSettings::minReadSupport){
@@ -265,14 +266,16 @@ void SingleEndScanner::textReport(vector<Mutation>& mutationList, vector<Match*>
                 cout<<m+1<<", ";
                 matches[m]->print(mutationList[i].mLeft.length(), mutationList[i].mCenter.length(), mutationList[i].mRight.length());
             }
+        } else {
+            undetected++;
         }
     }
     if(found == false) {
         cout << "MutScan didn't find any mutation" << endl;
     }
     // if processing VCF, output those with no supporting reads found
-    if(GlobalSettings::processingVCF) {
-        cerr << "Following mutations are not detected" << endl;
+    if(GlobalSettings::processingVCF && undetected>0) {
+        cerr << undetected << " mutations of this VCF are not detected" << endl;
         for(int i=0;i<mutationList.size();i++){
             vector<Match*> matches = mutationMatches[i];
             if(matches.size()<GlobalSettings::minReadSupport){

diff --git a/src/vcfreader.cpp b/src/vcfreader.cpp
@@ -54,22 +54,27 @@ bool VcfReader::readNext()
     if(items[0][0]=='#')
         return false;
 
-    int pos =  atoi(items[1].c_str());
-    Variant var;
-    var.chrom = trim(items[0]);
-    var.pos = pos;
-    var.id = trim(items[2]);
-    var.ref = trim(items[3]);
-    var.alt = trim(items[4]);
-    var.qual = trim(items[5]);
-    var.filter = trim(items[6]);
-    var.info = trim(items[7]);
-
-    // format is not required
-    if(items.size()>=9)
-        var.format = trim(items[8]);
-
-    mVariants.push_back(var);
+    //split the alt by comma to make multiple variants, GATK usually output such kind of variant like C>T,AT
+    vector<string> alts;
+    split(trim(items[4]), alts, ",");
+    for(int a=0; a<alts.size(); a++){
+        int pos =  atoi(items[1].c_str());
+        Variant var;
+        var.chrom = trim(items[0]);
+        var.pos = pos;
+        var.id = trim(items[2]);
+        var.ref = trim(items[3]);
+        var.alt = alts[a];
+        var.qual = trim(items[5]);
+        var.filter = trim(items[6]);
+        var.info = trim(items[7]);
+
+        // format is not required
+        if(items.size()>=9)
+            var.format = trim(items[8]);
+
+        mVariants.push_back(var);
+    }
 
     return true;
 }