Merge branch 'mm2-v2.27' into 'master'

upgrade to minimap2-2.27 (r1193) See merge request machine-learning/dorado!885
nanoporetech · Mar 20, 2024 · 9b49ae5 · 9b49ae5
2 parents c0075a0 + 944b54b
commit 9b49ae5
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 11 deletions.
diff --git a/dorado/3rdparty/ont-minimap2 b/dorado/3rdparty/ont-minimap2
diff --git a/dorado/alignment/Minimap2Index.cpp b/dorado/alignment/Minimap2Index.cpp
@@ -101,8 +101,11 @@ bool Minimap2Index::initialise(Minimap2Options options) {
     m_mapping_options = std::make_optional<mm_mapopt_t>();
 
     mm_set_opt(0, &m_index_options.value(), &m_mapping_options.value());
-    // Setting options to map-ont default till relevant args are exposed.
-    mm_set_opt("map-ont", &m_index_options.value(), &m_mapping_options.value());
+    if (mm_set_opt(options.mm2_preset.c_str(), &m_index_options.value(),
+                   &m_mapping_options.value()) != 0) {
+        spdlog::error("Cannot set mm2 options with preset: {}", options.mm2_preset);
+        return false;
+    }
 
     set_index_options(options);
     set_mapping_options(options);
@@ -180,4 +183,4 @@ const mm_mapopt_t& Minimap2Index::mapping_options() const {
 
 const Minimap2Options& Minimap2Index::get_options() const { return m_options; }
 
-}  // namespace dorado::alignment
+}  // namespace dorado::alignment
diff --git a/dorado/alignment/Minimap2Options.h b/dorado/alignment/Minimap2Options.h
@@ -1,6 +1,7 @@
 #pragma once
 
 #include <cstdint>
+#include <string>
 #include <tuple>
 
 namespace dorado::alignment {
@@ -9,11 +10,12 @@ struct Minimap2IndexOptions {
     short kmer_size;
     short window_size;
     uint64_t index_batch_size;
+    std::string mm2_preset;
 };
 
 inline bool operator<(const Minimap2IndexOptions& l, const Minimap2IndexOptions& r) {
-    return std::tie(l.kmer_size, l.window_size, l.index_batch_size) <
-           std::tie(r.kmer_size, r.window_size, r.index_batch_size);
+    return std::tie(l.kmer_size, l.window_size, l.index_batch_size, l.mm2_preset) <
+           std::tie(r.kmer_size, r.window_size, r.index_batch_size, r.mm2_preset);
 }
 
 inline bool operator>(const Minimap2IndexOptions& l, const Minimap2IndexOptions& r) {
@@ -29,8 +31,8 @@ inline bool operator>=(const Minimap2IndexOptions& l, const Minimap2IndexOptions
 }
 
 inline bool operator==(const Minimap2IndexOptions& l, const Minimap2IndexOptions& r) {
-    return std::tie(l.kmer_size, l.window_size, l.index_batch_size) ==
-           std::tie(r.kmer_size, r.window_size, r.index_batch_size);
+    return std::tie(l.kmer_size, l.window_size, l.index_batch_size, l.mm2_preset) ==
+           std::tie(r.kmer_size, r.window_size, r.index_batch_size, r.mm2_preset);
 }
 
 inline bool operator!=(const Minimap2IndexOptions& l, const Minimap2IndexOptions& r) {
@@ -87,7 +89,7 @@ inline bool operator==(const Minimap2Options& l, const Minimap2Options& r) {
 
 inline bool operator!=(const Minimap2Options& l, const Minimap2Options& r) { return !(l == r); }
 
-static constexpr Minimap2Options dflt_options{{15, 10, 16000000000ull},
-                                              {5, 500, 20000, false, false, true},
-                                              false};
+static const Minimap2Options dflt_options{{15, 10, 16000000000ull, "lr:hq"},
+                                          {5, 500, 20000, false, false, true},
+                                          false};
 }  // namespace dorado::alignment
diff --git a/dorado/cli/cli_utils.h b/dorado/cli/cli_utils.h
@@ -203,6 +203,12 @@ void add_minimap2_arguments(ArgParser& parser, const Options& dflt) {
                   "specified as NUM,[NUM]")
             .default_value(to_size(dflt.bandwidth) + "," + to_size(dflt.bandwidth_long));
 
+    // Setting options to lr:hq which is appropriate for high quality nanopore reads.
+    parser.visible.add_argument("--mm2-preset")
+            .help("minimap2 preset for indexing and mapping. Alias for the -x "
+                  "option in minimap2.")
+            .default_value(dflt.mm2_preset);
+
     parser.hidden.add_argument("--secondary-seq")
             .help("minimap2 output seq/qual for secondary and supplementary alignments")
             .default_value(false)
@@ -251,6 +257,7 @@ Options process_minimap2_arguments(const ArgParser& parser, const Options& dflt)
         throw std::runtime_error("Wrong number of arguments for option '-r'.");
     }
     res.soft_clipping = parser.visible.get<bool>("Y");
+    res.mm2_preset = parser.visible.get<std::string>("mm2-preset");
     res.secondary_seq = parser.hidden.get<bool>("secondary-seq");
     res.print_aln_seq = parser.hidden.get<bool>("print-aln-seq");
     return res;