Some improvements

CMakeLists.txt

@@ -1,6 +1,7 @@
 cmake_minimum_required(VERSION 3.14)
 
 include(ExternalProject)
+include(FetchContent)
 
 set(CMAKE_CXX_STANDARD 14)
 set(CMAKE_CXX_STANDARD_REQUIRED True)
@@ -43,13 +44,8 @@ if(NOT DEFINED CONDAPREFIX)
   message(STATUS "libdeflate will be built from source")
   ExternalProject_Add(deflate
     GIT_REPOSITORY https://github.com/ebiggers/libdeflate.git
-    GIT_TAG 047aa84e01b38d82f3612810e357bd40f14a3d39
+    GIT_TAG 020133854ff73b8506fe59f92a9b5b622d360716 # v1.15
     BUILD_IN_SOURCE 1
-    UPDATE_COMMAND ""
-    CONFIGURE_COMMAND ""
-    BUILD_COMMAND make PREFIX="."
-      "CC=${CMAKE_C_COMPILER}"
-      "CFLAGS=-fPIC -O3" libdeflate.a
     INSTALL_COMMAND "")
   ExternalProject_Get_Property(deflate SOURCE_DIR)
   SET(DEFLATE_SOURCE_DIR ${SOURCE_DIR})
@@ -59,15 +55,15 @@ if(NOT DEFINED CONDAPREFIX)
 
   # HTSLIB
   #########
-  SET(CONF_CMD autoheader && autoreconf && ./configure --disable-libcurl --disable-gcs --with-libdeflate "CC=${CMAKE_C_COMPILER}" "CFLAGS=-O3 -I${DEFLATE_SOURCE_DIR}" "LDFLAGS=-L${DEFLATE_SOURCE_DIR}")
+  SET(CONF_CMD autoheader && autoreconf -i && ./configure --disable-libcurl --disable-gcs --with-libdeflate "CC=${CMAKE_C_COMPILER}" "CFLAGS=-O3 -I${DEFLATE_SOURCE_DIR}" "LDFLAGS=-L${DEFLATE_SOURCE_DIR}")
   if (HOLYBUILD)
     # we have to include local libraries (libbz2 and liblzma)
-    SET(CONF_CMD autoheader && autoreconf && ./configure --disable-libcurl --disable-gcs --with-libdeflate "CC=${CMAKE_C_COMPILER}" "CFLAGS=-O3 -I${DEFLATE_SOURCE_DIR} -I${BZIP_LIB_DIR}" "LDFLAGS=-L${DEFLATE_SOURCE_DIR} -L${BZIP_LIB_DIR} -L${LZMA_LIB_DIR}")
+    SET(CONF_CMD autoheader && autoreconf -i && ./configure --disable-libcurl --disable-gcs --with-libdeflate "CC=${CMAKE_C_COMPILER}" "CFLAGS=-O3 -I${DEFLATE_SOURCE_DIR} -I${BZIP_LIB_DIR}" "LDFLAGS=-L${DEFLATE_SOURCE_DIR} -L${BZIP_LIB_DIR} -L${LZMA_LIB_DIR}")
   endif ()
   message(STATUS "htslib will be built from source")
   ExternalProject_Add(htslib
     GIT_REPOSITORY https://github.com/samtools/htslib.git
-    GIT_TAG 2cd99cf333938a9eb70393bc59379e1d6ea66b37
+    GIT_TAG 4e61c128238f3e7cbb3b1f4e9c0fdb4880aa9a10 # v1.16
     UPDATE_COMMAND ""
     BUILD_IN_SOURCE 1
     CONFIGURE_COMMAND ${CONF_CMD}
@@ -98,7 +94,7 @@ else ()
 endif ()
 ExternalProject_Add(abpoa
   GIT_REPOSITORY https://github.com/yangao07/abPOA.git
-  GIT_TAG 16c64e5163cbd0502792ca20f1069fe40c929b0d
+  GIT_TAG 16c64e5163cbd0502792ca20f1069fe40c929b0d # v1.41
   PATCH_COMMAND ${PATCH_CMD}
   INSTALL_COMMAND ""
 )
@@ -114,7 +110,7 @@ SET(ABPOA_INCLUDE_DIR ${ABPOA_SOURCE_DIR}/include)
 message(STATUS "parasail will be built from source")
 ExternalProject_Add(parasail
   GIT_REPOSITORY https://github.com/jeffdaily/parasail.git
-  GIT_TAG e236a6feec7bc84fb81923d783acb345e538a281
+  GIT_TAG d23aed253fb5ecc7aeab0f81b6d64fcc26448089 # v2.6
   UPDATE_COMMAND ""
   BUILD_IN_SOURCE 1
   CMAKE_ARGS -DBUILD_SHARED_LIBS=OFF
@@ -128,30 +124,24 @@ set_target_properties(PARASAIL PROPERTIES IMPORTED_LOCATION ${PARASAIL_SOURCE_DI
 
 # rapidfuzz-cpp
 ################
-ExternalProject_Add(rapidfuzz
-    GIT_REPOSITORY https://github.com/maxbachmann/rapidfuzz-cpp.git
-    GIT_TAG d1e82379395cafc6d439c1c1e2cbe7512eaf2518
-    BUILD_IN_SOURCE 1
-    UPDATE_COMMAND ""
-    INSTALL_COMMAND ""
-)
-ExternalProject_Get_Property(rapidfuzz SOURCE_DIR)
-SET(RAPIDFUZZ_SOURCE_DIR ${SOURCE_DIR})
-SET(RAPIDFUZZ_INCLUDE_DIR ${RAPIDFUZZ_SOURCE_DIR})
+FetchContent_Declare(rapidfuzz
+  GIT_REPOSITORY https://github.com/maxbachmann/rapidfuzz-cpp.git
+  GIT_TAG 5412d5d877518e7754394bdbec76e45c8187c631) # v1.10.4
+FetchContent_MakeAvailable(rapidfuzz)
 
 # interval-tree
 ################
 ExternalProject_Add(intervaltree
     GIT_REPOSITORY https://github.com/5cript/interval-tree.git
-    GIT_TAG 4d7c66ce4fb542a29a7d74039f8e7c92fae3d6f8
+    GIT_TAG 309b9c725191d4bb1d134f28a8a32ad2f68a8ffa
     UPDATE_COMMAND ""
     CONFIGURE_COMMAND ""
     BUILD_COMMAND ""
     INSTALL_COMMAND ""
 )
 ExternalProject_Get_Property(intervaltree SOURCE_DIR)
 SET(INTERVALTREE_SOURCE_DIR ${SOURCE_DIR})
-SET(INTERVALTREE_INCLUDE_DIR ${INTERVALTREE_SOURCE_DIR})
+SET(INTERVALTREE_INCLUDE_DIR ${INTERVALTREE_SOURCE_DIR}/include/interval-tree)
 
 # ropebwt2
 ##########
@@ -213,6 +203,7 @@ else()
 endif()
 
 target_link_libraries(SVDSS
+  PUBLIC rapidfuzz::rapidfuzz
   PUBLIC ${BINARY_DIR}/lib/libabpoa.a
   PUBLIC PARASAIL
   PUBLIC ROPEBWT

README.md

@@ -210,7 +210,7 @@ For inquiries on this software please open an [issue](https://github.com/Parsoa/
 
 ### Citation
 
-SVDSS is currently pending peer review. A pre-print is available on [BioRxiv](https://www.biorxiv.org/content/10.1101/2022.02.12.480198v1).
+SVDSS is now published in [Nature Methods](https://doi.org/10.1038/s41592-022-01674-1).
 
 ##### Experiments
 Instructions on how to reproduce the experiments described in the manuscript can be found [here](https://github.com/ldenti/SVDSS-experiments) (also provided as submodule of this repository).

caller.cpp

@@ -164,6 +164,9 @@ void Caller::print_vcf_header() {
   ovcf << "##INFO=<ID=IMPRECISE,Number=0,Type=Flag,Description=\"Imprecise "
           "structural variation\">"
        << endl;
+  ovcf << "##INFO=<ID=CIGAR,Number=A,Type=String,Description=\"CIGAR of "
+          "consensus\">"
+       << endl;
   ovcf << "##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">"
        << endl;
   ovcf << "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tDEFAULT"

cluster.cpp

@@ -27,13 +27,20 @@ string Cluster::poa() {
   abpt->out_msa = 0;
   abpt->out_cons = 1;
   abpt->out_gfa = 0;
-  // abpt->w = 6, abpt->k = 9;
-  // abpt->min_w = 10; // minimizer-based seeding and partition
   // abpt->is_diploid = 1;
-  // abpt->min_freq = 0.3;
   abpt->progressive_poa = 0;
   abpoa_post_set_para(abpt);
 
+  // abpt->match = 2;      // match score
+  // abpt->mismatch = 4;   // mismatch penalty
+  // abpt->gap_mode = ABPOA_CONVEX_GAP; // gap penalty mode
+  // abpt->gap_open1 = 4;  // gap open penalty #1
+  // abpt->gap_ext1 = 2;   // gap extension penalty #1
+  // abpt->gap_open2 = 24; // gap open penalty #2
+  // abpt->gap_ext2 = 1;   // gap extension penalty #2
+  // gap_penalty = min{gap_open1 + gap_len * gap_ext1, gap_open2 + gap_len *
+  // gap_ext2}
+
   int *seq_lens = (int *)malloc(sizeof(int) * n_seqs);
   uint8_t **bseqs = (uint8_t **)malloc(sizeof(uint8_t *) * n_seqs);
   for (uint i = 0; i < n_seqs; ++i) {

config.cpp

@@ -37,7 +37,10 @@ Configuration::Configuration()
       "b,binary", "", cxxopts::value<bool>()->default_value("false"))(
       "aggregate", "", cxxopts::value<bool>()->default_value("false"))(
       "selective", "", cxxopts::value<bool>()->default_value("true"))(
-      "version", "Print version information.")("help", "Print this help.");
+      "version", "Print version information.")("help", "Print this help.")(
+      "l", "", cxxopts::value<float>())(
+      "acc", "", cxxopts::value<float>())(
+      "verbose", "", cxxopts::value<bool>()->default_value("false"));
 }
 
 void Configuration::parse(int argc, char **argv) {
@@ -107,12 +110,19 @@ void Configuration::parse(int argc, char **argv) {
   if (results.count("min-cluster-weight")) {
     min_cluster_weight = results["min-cluster-weight"].as<int>();
   }
+  if (results.count("l")) {
+    min_ratio = results["l"].as<float>();
+  }
+  if (results.count("acc")) {
+    al_accuracy = results["acc"].as<float>();
+  }
   binary = results["binary"].as<bool>();
   clipped = results["clipped"].as<bool>();
   assemble = results["assemble"].as<bool>();
   putative = !(results["noputative"].as<bool>());
   aggregate = results["aggregate"].as<bool>();
   selective = results["selective"].as<bool>();
   version = results["version"].as<bool>();
+  verbose = results["verbose"].as<bool>();
   help = results["help"].as<bool>();
 }

config.hpp

@@ -29,6 +29,8 @@ class Configuration {
   int min_indel_length = 20;
   int aggregate_batches = 5;
   int min_cluster_weight = 2;
+  float min_ratio = 0.97; // FIXME: change name
+  float al_accuracy = 0.02;
 
   bool binary = false;
   bool clipped = false;
@@ -37,6 +39,7 @@ class Configuration {
   bool aggregate = false;
   bool selective = true;
   bool version = false;
+  bool verbose = false;
   bool help = false;
 
   std::string bed;