adding initial code to turn cigar strings into variants

src/primer_bed.cpp

@@ -5,6 +5,8 @@ std::string primer::get_region() { return region; }
 
 std::vector<std::vector<uint32_t>> cigarotype::get_cigarotypes() { return cigarotypes; }
 
+std::vector<uint32_t> cigarotype::get_count_cigarotypes() { return count_cigarotypes; }
+
 std::vector<uint32_t> cigarotype::get_nlengths() { return nlengths; }
 
 int primer::get_score() { return score; }
@@ -287,7 +289,17 @@ int populate_pair_indices(std::vector<primer>& primers, std::string path) {
   }
   return 0;
 }
-//this must be passed by pointer due to array decay
+
+void primer::transform_mutations() {
+  /*
+   * Take all recorded cigarotypes and transform them into positions relative to primer
+   */
+  std::vector<std::vector<uint32_t>> cigarotypes = this->get_cigarotypes(); 
+  std::cerr << cigarotypes.size() << std::endl;
+  std::cerr << "hello world" << std::endl;;
+}
+
+//cigar must be passed by pointer due to array decay
 void cigarotype::add_cigarotype(uint32_t *cigar , uint32_t start_pos, uint32_t nlength){
   bool found = false; //have we seen this before
   std::vector<uint32_t> cig;

src/primer_bed.h

@@ -9,15 +9,16 @@
 
 class cigarotype{
   protected:
-    std::vector<std::vector<uint32_t>> cigarotypes; //unique cigar value x = cigar, y = operation z, = length
+    std::vector<std::vector<uint32_t>> cigarotypes; //unique cigar value 
     std::vector<uint32_t> nlengths; //this is just for printing, record length og cig ops
     std::vector<uint32_t> ncigarotypes; //starting pos of these cigars
     std::vector<uint32_t> count_cigarotypes; //count of amount found
+
   public:
     std::vector<std::vector<uint32_t>> get_cigarotypes();
+    std::vector<uint32_t> get_count_cigarotypes();
     std::vector<uint32_t> get_nlengths();
     void add_cigarotype(uint32_t *cigar, uint32_t start_pos, uint32_t nlength);
-    //void transform_mutations(); //this is will process unique cigar strings into mutations/indels
     int test = 12;
 
 };
@@ -55,6 +56,7 @@ class primer : public cigarotype{
   void set_indice(int16_t i);
   void set_read_count(uint32_t rc);
   void add_read_count(uint32_t rc);
+  void transform_mutations();
   bool operator==(const primer& p) const {
     return (indice == p.get_indice()) ? true : false;
   }
@@ -71,5 +73,7 @@ primer get_min_start(std::vector<primer> primers);
 primer get_max_end(std::vector<primer> primers);
 void add_cigarotype(uint32_t *cigar, uint32_t start_pos, uint32_t nlength);
 std::vector<std::vector<uint32_t>> get_cigarotypes();
+std::vector<uint32_t> get_count_cigarotypes();
 std::vector<uint32_t> get_nlengths();
+void transform_mutations();
 #endif

src/saga.cpp

@@ -152,24 +152,26 @@ int preprocess_reads(std::string bam, std::string bed, std::string bam_out,
     }   
   }
 
-  for(uint32_t i=0; i < primers.size(); i++){
+  //sanity check not intended for final version
+  /*for(uint32_t i=0; i < primers.size(); i++){
     std::cout << "primer " << i << std::endl;
     primer tmp = primers[i];
     std::vector<std::vector<uint32_t>> otmp = tmp.get_cigarotypes();
     std::vector<uint32_t> nlengths = tmp.get_nlengths();
+    std::vector<uint32_t> counts = tmp.get_count_cigarotypes();
     for(uint32_t j=0; j < otmp.size(); j++){
-      std::cerr << "cigarotype " << j << std::endl;
+      std::cerr << "cigarotype " << j << " counts " << counts[j] << std::endl;
       std::cerr << nlengths[j] << std::endl;
 
-      for(uint32_t k = 0; k < nlengths[j]; k++){
-        std::cerr << "k " << k << std::endl;
-        std:: cerr << bam_cigar_op(otmp[j][k]) << " " << bam_cigar_oplen(otmp[j][k]) << std::endl;
-      }
     }
     if( i > 10) break;
-  }
+  }*/
 
-  //PRIMER METHOD calculate mutations from unique cigars per primer (ie. count mutations per primer efficiently)
+  //PRIMER METHOD calculate mutations from unique cigars per primer, outputing variant frequencies
+  for(uint32_t i=0; i < primers.size(); i++){
+    primers[i].transform_mutations();
+    if(i > 10) break;
+  }
   //AMPLICON METHOD translate this into amplicon haplotype obj of mutations per primer (ie. variant freq per amplicon)
   //detect fluctuating variants - iterate every position and look for fluctuation between every amplicon objects, flag these
   //combine amplicon counts to get total variants