germline.cpp : add function to load germline from germlines.data

algo/core/germline.cpp

 
 #include "germline.h"
+#include <fstream>
 #include <ctype.h>
 
 void Germline::init(string _code, char _shortcut,
@@ -86,6 +87,49 @@ Germline::Germline(string _code, char _shortcut,
   rep_3 = _rep_3 ;
 }
 
+Germline::Germline(string code, char shortcut, string path, json json_recom, int max_indexing)
+{
+    
+  int delta_min = 0;
+  
+  if (json_recom.find("4") != json_recom.end()) {
+      delta_min = -10;
+  }
+  
+  init(code, shortcut, delta_min, max_indexing);
+  
+  rep_5 = Fasta(2, "|") ;
+  rep_4 = Fasta(2, "|") ;
+  rep_3 = Fasta(2, "|") ;
+
+  for (json::iterator it = json_recom["5"].begin();
+       it != json_recom["5"].end(); ++it) 
+  {
+    string filename = *it;
+    f_reps_5.push_back(path + filename);
+    rep_5.add(path + filename);
+  }
+  
+  if (json_recom.find("4") != json_recom.end()) {
+    for (json::iterator it = json_recom["4"].begin();
+        it != json_recom["4"].end(); ++it) 
+    {
+        string filename = *it;
+        f_reps_4.push_back(path + filename);
+        rep_4.add(path + filename);
+    }
+  }
+  
+  for (json::iterator it = json_recom["3"].begin();
+       it != json_recom["3"].end(); ++it) 
+  {
+    string filename = *it;
+    f_reps_3.push_back(path + filename);
+    rep_3.add(path + filename);
+  }
+  
+}
+
 void Germline::new_index(string seed)
 {
   bool rc = true ;
@@ -177,6 +221,37 @@ void MultiGermline::add_germline(Germline *germline, string seed)
   germlines.push_back(germline);
 }
 
+void MultiGermline::build(string path, int max_indexing)
+{
+  //parse germlines.data
+  ifstream germline_data(path + "/germlines.data");
+  string content( (std::istreambuf_iterator<char>(germline_data) ),
+                  (std::istreambuf_iterator<char>()    ) );
+
+  json j = json::parse(content);
+  
+  //for each germline
+  for (json::iterator it = j.begin(); it != j.end(); ++it) {
+      
+    json recom = it.value()["recombinations"];
+    char shortcut = it.value()["shortcut"].dump()[1];
+    string code = it.key();
+    string seed = it.value()["parameters"]["seed"];
+    
+    map<string, string> seedMap;
+    seedMap["13s"] = SEED_S13;
+    seedMap["12s"] = SEED_S12;
+    seedMap["10s"] = SEED_S10;
+    seedMap["9s"] = SEED_9;
+    
+    //for each set of recombination 3/4/5
+    for (json::iterator it2 = recom.begin(); it2 != recom.end(); ++it2) {
+      add_germline(new Germline(code, shortcut, path, *it2 , max_indexing), seedMap[seed]);
+    }
+  }
+  
+}
+
 void MultiGermline::build_default_set(string path, int max_indexing)
 {
   // Should parse 'data/germlines.data'

algo/core/germline.h

@@ -8,10 +8,12 @@
 #include "kmerstore.h"
 #include "stats.h"
 #include "tools.h"
+#include "../lib/json.hpp"
 
 #define PSEUDO_GERMLINE_MAX12 "xxx"
 
 using namespace std;
+using json = nlohmann::json;
 
 class Germline {
  private:
@@ -47,6 +49,8 @@ class Germline {
   Germline(string _code, char _shortcut,
 	   int _delta_min,
             int max_indexing=0);
+  
+  Germline(string _code, char shortcut, string path, json json_recom, int max_indexing=0);
 
   ~Germline();
 
@@ -99,6 +103,7 @@ class MultiGermline {
   void add_germline(Germline *germline, string seed);
   void build_default_set(string path, int max_indexing);
   void build_incomplete_set(string path, int max_indexing);
+  void build(string path, int max_indexing);
 
   // Creates and update an unique index for all the germlines
   // If 'set_index' is set, set this index as the index for all germlines

algo/core/segment.h

@@ -2,6 +2,7 @@
 #define SEGMENT_H
 
 #include <string>
+#include <fstream>
 #include <iostream>
 #include "fasta.h"
 #include "dynprog.h"

algo/vidjil.cpp

@@ -773,7 +773,8 @@ int main (int argc, char **argv)
     
       if (multi_germline)
 	{
-	  multigermline->build_default_set(multi_germline_file, trim_sequences);
+	  multigermline->build_default_set(multi_germline_file, trim_sequences);      
+      //multigermline->build(multi_germline_file, trim_sequences);
 	}
       else
 	{