Merge pull request #137 from andersen-lab/issue_132

addressing issue #132

.github/workflows/ccpp.yml

@@ -57,7 +57,7 @@ jobs:
 
   build-macos:
 
-    runs-on: [ macos-latest ]
+    runs-on: [ macos-11 ]
 
     steps:
     - uses: actions/checkout@master

data/primer_pair_test/test_primer_pair.bed

@@ -0,0 +1,17 @@
+MN908947.3	3507	3531	nCoV-2019_11_RIGHT	1	-
+MN908947.3	3460	3482	nCoV-2019_12_LEFT	2	+
+MN908947.3	3826	3853	nCoV-2019_12_RIGHT	2	-
+MN908947.3	3771	3795	nCoV-2019_13_LEFT	1	+
+MN908947.3	4142	4164	nCoV-2019_13_RIGHT	1	-
+MN908947.3	4054	4077	nCoV-2019_14_LEFT	2	+
+MN908947.3	4044	4068	nCoV-2019_14_LEFT_alt4	2	+
+MN908947.3	4428	4450	nCoV-2019_14_RIGHT	2	-
+MN908947.3	4402	4424	nCoV-2019_14_RIGHT_alt2	2	-
+MN908947.3	4294	4321	nCoV-2019_15_LEFT	1	+
+MN908947.3	4295	4322	nCoV-2019_15_LEFT_alt1_ssi1	1	+
+MN908947.3	4296	4322	nCoV-2019_15_LEFT_alt1	1	+
+MN908947.3	4674	4696	nCoV-2019_15_RIGHT	1	-
+MN908947.3	4666	4689	nCoV-2019_15_RIGHT_alt3	1	-
+MN908947.3	4636	4658	nCoV-2019_16_LEFT	2	+
+MN908947.3	4995	5017	nCoV-2019_16_RIGHT	2	-
+MN908947.3	4939	4966	nCoV-2019_17_LEFT	1	+

data/primer_pair_test/test_primer_pair_trim.tsv

@@ -0,0 +1,7 @@
+nCoV-2019_12_LEFT	nCoV-2019_12_RIGHT
+nCoV-2019_13_LEFT	nCoV-2019_13_RIGHT
+nCoV-2019_14_LEFT	nCoV-2019_14_RIGHT
+nCoV-2019_14_LEFT_alt4	nCoV-2019_14_RIGHT_alt2
+nCoV-2019_15_LEFT	nCoV-2019_15_RIGHT
+nCoV-2019_15_LEFT_alt1	nCoV-2019_15_RIGHT_alt3
+nCoV-2019_16_LEFT	nCoV-2019_16_RIGHT

src/call_consensus_pileup.cpp

@@ -32,7 +32,7 @@ ret_t get_consensus_allele(std::vector<allele> ad, uint8_t min_qual, double thre
   std::vector<allele> nuc_pos;
   allele tmp_a;
   char n;
-  uint32_t max_l = 0, max_depth = 0, tmp_depth = 0, cur_depth = 0, total_max_depth = 0, gap_depth = 0, total_indel_depth = 0;
+  uint32_t max_l = 0, max_depth = 0, cur_depth = 0, tmp_depth = 0, total_max_depth = 0, gap_depth = 0, total_indel_depth = 0;
   uint8_t ambg_n = 1, ctr = 0;
   double q = 0, tq = 0, cur_threshold = 0;
   std::vector<allele>::iterator it;

src/interval_tree.cpp

@@ -83,7 +83,7 @@ void IntervalTree::inOrder(ITNode *root){
 
 // A stand-alone function to create a tree containing the coordinates of each amplicon
 // based on user-specified primer pairs
-IntervalTree populate_amplicons(std::string pair_info_file, std::vector<primer> primers){
+IntervalTree populate_amplicons(std::string pair_info_file, std::vector<primer> &primers){
   int amplicon_start = -1;
   int amplicon_end = -1;
   IntervalTree tree = IntervalTree();

src/interval_tree.h

@@ -48,6 +48,6 @@ class IntervalTree{
   void inOrder() {inOrder(_root);}
 };
 
-IntervalTree populate_amplicons(std::string pair_info_file, std::vector<primer> primers);
+IntervalTree populate_amplicons(std::string pair_info_file, std::vector<primer> &primers);
 
 #endif

src/primer_bed.cpp

@@ -233,22 +233,40 @@ std::vector<primer> get_primers(std::vector<primer> p, unsigned int pos){
   return primers_with_mismatches;
 }
 
+//function to trim trailing spaces 
+std::string& rtrim(std::string& str, const std::string& chars = "\t\n\v\f\r ")
+{
+    str.erase(str.find_last_not_of(chars) + 1);
+    return str;
+}
+
 // Assumes unique primer names in BED file
+//returns the index of a primer by name
 int get_primer_indice(std::vector<primer> p, std::string name){
+  //iterate through the primers from the bed file
   for(std::vector<primer>::iterator it = p.begin(); it != p.end(); ++it) {
-    if(it->get_name().compare(name) == 0){
+    //check if the two strings are the same
+    if(it->get_name().compare(rtrim(name)) == 0){
       return it - p.begin();
     }
   }
   return -1;
 }
 
+//function using the tab seperated primer pair file
 int populate_pair_indices(std::vector<primer> &primers, std::string path){
+  /*
+  * @param primers: the primer vector to add pair info to
+  * @param path: the path to the primer pair file
+  */
+
+  //load primer pair file
   std::ifstream fin(path.c_str());
   std::string line, cell, p1,p2;
   std::stringstream line_stream;
   std::vector<primer>::iterator it;
   int32_t indice;
+  //iterate the primer pair file line by line
   while (std::getline(fin, line)){
     line_stream << line;
     std::getline(line_stream, cell, '\t');
@@ -257,22 +275,31 @@ int populate_pair_indices(std::vector<primer> &primers, std::string path){
     std::getline(line_stream, cell, '\t');
     p2 = cell;
     line_stream.clear();
+
+    p1 = rtrim(p1);
+    p2 = rtrim(p2); 
+
     if(!p1.empty() && !p2.empty()){
       for(it = primers.begin(); it != primers.end(); ++it) {
-	if (it->get_name() == p1) {
-	  indice = get_primer_indice(primers, p2);
-	  if (indice != -1)
-	    it->set_pair_indice(indice);
-	  else
-	    std::cout << "Primer pair for " << p1 << " not found in BED file." <<std::endl;
-	} else if (it->get_name() == p2){
-	  indice = get_primer_indice(primers, p1);
+      //search for primer name in pair file
+	  if (it->get_name() == p1) {
+        //make sure it's pair exists
+	    indice = get_primer_indice(primers, p2);
+	    if (indice != -1){
+	      it->set_pair_indice(indice);
+        }else{
+	      std::cout << "Primer pair for " << p1 << " not found in BED file." <<std::endl;
+        }
+	  } else if (it->get_name() == p2){
+        indice = get_primer_indice(primers, p1);
 	  if(indice != -1)
 	    it->set_pair_indice(indice);
 	  else
 	    std::cout << "Primer pair for " << p2 << " not found in BED file." << std::endl;
-	}
+	  }
       }
+    } else {
+      std::cout << "Primer pair is empty." << std::endl;
     }
   }
   return 0;
@@ -287,4 +314,39 @@ primer get_min_start(std::vector<primer> primers){
 primer get_max_end(std::vector<primer> primers){
   auto minmax_start = std::minmax_element(primers.begin(), primers.end(), [] (primer lhs, primer rhs) {return lhs.get_end() < rhs.get_end();});
   return *(minmax_start.second);
-}
+}
+
+//give a primer index, find its pair and return the primer
+primer fetch_primer_pair(int16_t index, std::vector<primer> primers){
+  /*
+  * @param index: the pair index for the primer of interest
+  * @return pair_primer: the primer object that's pairs with the primer of interest
+  */
+  std::vector<primer>::iterator it;
+  primer pair_primer;
+  for(it = primers.begin(); it != primers.end(); ++it) {
+    if(it->get_indice() == index){
+        return *it;
+    }      
+  }
+  return(pair_primer);
+}
+
+void print_all_primer_info(std::vector<primer> primers){
+  std::vector<primer>::iterator it;
+  for(it = primers.begin(); it != primers.end(); ++it) {
+    std::cout << "Primer name: " << it->get_name() << std::endl;
+    std::cout<< "Primer start: " << it->get_start() << std::endl; 
+    std::cout<< "Primer end: " << it->get_end() << std::endl; 
+    std::cout<< "Indice: " << it->get_indice() << std::endl; 
+    std::cout<< "Pair indice: " << it->get_pair_indice() << std::endl; 
+  }
+}
+
+void print_primer_info(primer primer){
+    std::cout << "Primer name: " << primer.get_name() << std::endl;
+    std::cout<< "Primer start: " << primer.get_start() << std::endl; 
+    std::cout<< "Primer end: " << primer.get_end() << std::endl; 
+    std::cout<< "Indice: " << primer.get_indice() << std::endl; 
+    std::cout<< "Pair indice: " << primer.get_pair_indice() << std::endl; 
+}

src/primer_bed.h

@@ -50,7 +50,10 @@ std::vector<primer> populate_from_file(std::string path, int32_t offset);
 std::vector<primer> populate_from_file(std::string path);
 std::vector<primer> get_primers(std::vector<primer> p, unsigned int pos);
 int get_primer_indice(std::vector<primer> p, std::string name);
+primer fetch_primer_pair(int16_t index, std::vector<primer> primers);
 int populate_pair_indices(std::vector<primer> &primers, std::string path);
+void print_primer_info(primer primers);
+void print_all_primer_info(std::vector<primer> primers);
 primer get_min_start(std::vector<primer> primers);
 primer get_max_end(std::vector<primer> primers);