Merge pull request #102 from vgteam/i99_odgi_bin_nuc_pos

fixes #99
pangenome · Apr 10, 2020 · c09c983 · c09c983
2 parents 5e18490 + 49acd60
commit c09c983
Show file tree

Hide file tree

Showing 3 changed files with 30 additions and 12 deletions.
diff --git a/src/algorithms/bin_path_info.cpp b/src/algorithms/bin_path_info.cpp
@@ -71,10 +71,13 @@ void bin_path_info(const PathHandleGraph& graph,
                         }
                         bins[curr_bin].mean_pos += path_pos++;
                         nucleotide_count += 1;
-                        if(bins[curr_bin].first_nucleotide == 0){
-                            bins[curr_bin].first_nucleotide = nucleotide_count;
-                        }
-                        bins[curr_bin].last_nucleotide = nucleotide_count;
+						if((bins[curr_bin].ranges.size() == 0) ||
+						    ((nucleotide_count -  bins[curr_bin].ranges.back().second) > 1)){
+						    bins[curr_bin].ranges.push_back(std::make_pair(nucleotide_count, nucleotide_count));
+							// bins[curr_bin].first_nucleotide = nucleotide_count;
+						} else {
+                            bins[curr_bin].ranges.back().second = nucleotide_count;
+						}
                         last_bin = curr_bin;
                         last_is_rev = is_rev;
                         last_pos_in_bin = curr_pos_in_bin;

diff --git a/src/algorithms/bin_path_info.hpp b/src/algorithms/bin_path_info.hpp
@@ -23,8 +23,9 @@ namespace odgi {
             double mean_cov;
             double mean_inv;
             double mean_pos;
-            long int first_nucleotide;
-            long int last_nucleotide;
+            vector<std::pair<uint64_t,uint64_t>> ranges;
+            // long int first_nucleotide;
+            // long int last_nucleotide;
         };
 
         void bin_path_info(const PathHandleGraph &graph,

diff --git a/src/subcommand/bin_main.cpp b/src/subcommand/bin_main.cpp
@@ -83,7 +83,7 @@ int main_bin(int argc, char** argv) {
     }
 
     // ODGI JSON VERSION
-    const uint64_t ODGI_JSON_VERSION = 10;
+    const uint64_t ODGI_JSON_VERSION = 12; // this brings the exact nucleotide positions for each bin for each path referred to as ranges
 
     std::function<void(const uint64_t&, const uint64_t&)> write_header_tsv
     = [&] (const uint64_t pangenome_length, const uint64_t bin_width) {
@@ -130,6 +130,20 @@ int main_bin(int argc, char** argv) {
                 }
             };
 
+    std::function<void(const vector<std::pair<uint64_t , uint64_t >>&)> write_ranges_json
+        = [&](const vector<std::pair<uint64_t , uint64_t >>& ranges) {
+        std::cout << "[";
+        for (int i = 0; i < ranges.size(); i++) {
+            std::pair<uint64_t, uint64_t > range = ranges[i];
+            if (i == 0) {
+                std::cout << "[" << range.first << "," << range.second << "]";
+            } else {
+                std::cout << "," << "[" << range.first << "," << range.second << "]";
+            }
+        }
+        std::cout << "]";
+    };
+
     std::function<void(const std::string&,
                        const std::vector<std::pair<uint64_t, uint64_t>>&,
                        const std::map<uint64_t, algorithms::path_info_t>&)> write_json
@@ -151,9 +165,9 @@ int main_bin(int argc, char** argv) {
             std::cout << "[" << bin_id << ","
                       << info.mean_cov << ","
                       << info.mean_inv << ","
-                      << info.mean_pos << ","
-                      << info.first_nucleotide << ","
-                      << info.last_nucleotide << "]";
+                      << info.mean_pos << ",";
+            write_ranges_json(info.ranges);
+			std::cout  << "]";
             if (i+1 != bins.size()) {
                 std::cout << ",";
             }
@@ -193,8 +207,8 @@ int main_bin(int argc, char** argv) {
                           << info.mean_cov << "\t"
                           << info.mean_inv << "\t"
                           << info.mean_pos << "\t"
-                          << info.first_nucleotide << "\t"
-                          << info.last_nucleotide << std::endl;
+						  << info.ranges[0].first << "\t"
+						  << info.ranges[info.ranges.size() - 1].second << std::endl;
             }
         }
     };