[cudamapper] Accuracy improvements through chaining

cudamapper/CMakeLists.txt

@@ -46,6 +46,7 @@ endif()
 
 cuda_add_library(${MODULE_NAME}
         src/application_parameters.cpp
+        src/chainer_utils.cu
         src/cudamapper.cpp
         src/index_batcher.cu
         src/index_descriptor.cpp
@@ -59,6 +60,8 @@ cuda_add_library(${MODULE_NAME}
         src/cudamapper_utils.cpp
         src/overlapper.cpp
         src/overlapper_triggered.cu
+        src/overlapper_anchmer.cu
+        src/overlapper_minimap.cu
         src/utils.cpp
         ${CMAKE_CURRENT_BINARY_DIR}/version.cpp)
 

cudamapper/include/claraparabricks/genomeworks/cudamapper/overlapper.hpp

@@ -38,6 +38,16 @@ namespace details
 namespace overlapper
 {
 
+// void filter_self_mappings(std::vector<Overlap>& overlaps,
+//                           const io::FastaParser& query_parser,
+//                           const io::FastaParser& target_parser,
+//                           const double max_percent_similarity);
+
+void filter_self_mappings(std::vector<Overlap>& overlaps,
+                          const io::FastaParser& query_parser,
+                          const io::FastaParser& target_parser,
+                          const double max_percent_overlap);
+
 /// \brief Extends a single overlap at its ends if the similarity of the query and target sequences is above a specified threshold.
 /// \param overlap An Overlap which is modified in place. Any of the query_start_position_in_read, query_end_position_in_read,
 /// target_start_position_in_read, and target_end_position_in_read fields may be modified.

cudamapper/src/application_parameters.cpp

@@ -51,6 +51,7 @@ ApplicationParameters::ApplicationParameters(int argc, char* argv[])
         {"min-overlap-fraction", required_argument, 0, 'z'},
         {"rescue-overlap-ends", no_argument, 0, 'R'},
         {"drop-fused-overlaps", no_argument, 0, 'D'},
+        {"drop-self-mappings", no_argument, 0, 'X'},
         {"query-indices-in-host-memory", required_argument, 0, 'Q'},
         {"query-indices-in-device-memory", required_argument, 0, 'q'},
         {"target-indices-in-host-memory", required_argument, 0, 'C'},
@@ -59,7 +60,7 @@ ApplicationParameters::ApplicationParameters(int argc, char* argv[])
         {"help", no_argument, 0, 'h'},
     };
 
-    std::string optstring = "k:w:d:m:i:t:F:a:r:l:b:z:RDQ:q:C:c:vh";
+    std::string optstring = "k:w:d:m:i:t:F:a:r:l:b:z:RDXQ:q:C:c:vh";
 
     bool target_indices_in_host_memory_set   = false;
     bool target_indices_in_device_memory_set = false;
@@ -117,6 +118,9 @@ ApplicationParameters::ApplicationParameters(int argc, char* argv[])
         case 'D':
             drop_fused_overlaps = true;
             break;
+        case 'X':
+            drop_self_mappings = true;
+            break;
         case 'Q':
             query_indices_in_host_memory = std::stoi(optarg);
             break;

cudamapper/src/application_parameters.hpp

@@ -55,6 +55,7 @@ class ApplicationParameters
     float min_overlap_fraction              = 0.8;   // z
     bool perform_overlap_end_rescue         = false; // R
     bool drop_fused_overlaps                = false; // D
+    bool drop_self_mappings                 = false; // X
     int32_t query_indices_in_host_memory    = 10;    // Q
     int32_t query_indices_in_device_memory  = 5;     // q
     int32_t target_indices_in_host_memory   = 10;    // C