Merge branch 'DOR-526-improve-rbk-splitting' into 'master'

[DOR-526] Improve RBK splitting by looking for spikes between muA and adapter sequences

Closes DOR-526

See merge request machine-learning/dorado!856

CMakeLists.txt

@@ -116,36 +116,8 @@ if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
     endif()
 endif()
 
-# Bring in spdlog
-add_subdirectory(${DORADO_3RD_PARTY_SOURCE}/spdlog)
-# Avoid namespace clashes with static torch.
-target_compile_definitions(spdlog PUBLIC
-    "FMT_BEGIN_NAMESPACE=namespace fmt { inline namespace ont {"
-    "FMT_END_NAMESPACE=}}"
-)
-
-# ELZIP_DECOMPRESS_ONLY stops minizip from adding OpenSSL as a target, preventing use of three dylibs on osx.
-set(ELZIP_DECOMPRESS_ONLY ON)
-add_subdirectory(${DORADO_3RD_PARTY_SOURCE}/elzip)
-
-option(ONT_MM2_EXE "Build minimap2 exe" OFF)
-add_subdirectory(${DORADO_3RD_PARTY_SOURCE}/ont-minimap2 EXCLUDE_FROM_ALL)
-message(STATUS "build minimap2 exe: ${ONT_MM2_EXE}")
-message(STATUS "build minimap2 lib: ${ONT_MM2_LIB}")
-if (ONT_MM2_EXE)
-    install(TARGETS minimap2_exe)
-    add_custom_target(testing_deps ALL DEPENDS minimap2_exe)
-endif()
-# Disable warnings from minimap source
-disable_warnings(minimap2)
-
-set(BUILD_TESTING OFF)
-add_subdirectory(${DORADO_3RD_PARTY_SOURCE}/edlib EXCLUDE_FROM_ALL)
-# Disable warnings from edlib source
-disable_warnings(edlib)
-set(BUILD_TESTING ON)
-
-add_subdirectory(${DORADO_3RD_PARTY_SOURCE}/date EXCLUDE_FROM_ALL)
+# Bring in 3rdparty libs
+add_subdirectory(dorado/3rdparty)
 
 enable_testing()
 
@@ -164,7 +136,7 @@ endif()
 
 configure_file(dorado/Version.h.in dorado/Version.h)
 
-set(LIB_SOURCE_FILES
+add_library(dorado_lib
     dorado/alignment/alignment_processing_items.cpp
     dorado/alignment/alignment_processing_items.h
     dorado/alignment/IndexFileAccess.cpp
@@ -247,6 +219,8 @@ set(LIB_SOURCE_FILES
     dorado/splitter/DuplexReadSplitter.h
     dorado/splitter/RNAReadSplitter.cpp
     dorado/splitter/RNAReadSplitter.h
+    dorado/splitter/myers.cpp
+    dorado/splitter/myers.h
     dorado/splitter/splitter_utils.cpp
     dorado/splitter/splitter_utils.h
     dorado/demux/AdapterDetector.cpp
@@ -273,9 +247,7 @@ set(LIB_SOURCE_FILES
     dorado/poly_tail/rna_poly_tail_calculator.h
     dorado/summary/summary.cpp
     dorado/summary/summary.h
- )
-
-add_library(dorado_lib ${LIB_SOURCE_FILES})
+)
 
 enable_warnings_as_errors(dorado_lib)
 

dorado/3rdparty/CMakeLists.txt

@@ -0,0 +1,33 @@
+# Bring in spdlog
+add_subdirectory(spdlog EXCLUDE_FROM_ALL)
+# Avoid namespace clashes with static torch.
+target_compile_definitions(spdlog PUBLIC
+    "FMT_BEGIN_NAMESPACE=namespace fmt { inline namespace ont {"
+    "FMT_END_NAMESPACE=}}"
+)
+
+# ELZIP_DECOMPRESS_ONLY stops minizip from adding OpenSSL as a target, preventing use of three dylibs on osx.
+set(ELZIP_DECOMPRESS_ONLY ON)
+add_subdirectory(elzip EXCLUDE_FROM_ALL)
+
+# minimap
+option(ONT_MM2_EXE "Build minimap2 exe" OFF)
+add_subdirectory(ont-minimap2 EXCLUDE_FROM_ALL)
+message(STATUS "build minimap2 exe: ${ONT_MM2_EXE}")
+message(STATUS "build minimap2 lib: ${ONT_MM2_LIB}")
+if (ONT_MM2_EXE)
+    install(TARGETS minimap2_exe)
+    add_custom_target(testing_deps ALL DEPENDS minimap2_exe)
+endif()
+# Disable warnings from minimap source
+disable_warnings(minimap2)
+
+# edlib
+set(BUILD_TESTING OFF)
+add_subdirectory(edlib EXCLUDE_FROM_ALL)
+# Disable warnings from edlib source
+disable_warnings(edlib)
+set(BUILD_TESTING ON)
+
+# date
+add_subdirectory(date EXCLUDE_FROM_ALL)