Remove disallow_inferior_barcodes as an option, it is now always on f…

…or dual-ended kits
nanoporetech · Nov 15, 2024 · cc32c93 · cc32c93
1 parent b8c9f98
commit cc32c93
Show file tree

Hide file tree

Showing 7 changed files with 17 additions and 40 deletions.
diff --git a/dorado/cli/basecaller.cpp b/dorado/cli/basecaller.cpp
@@ -225,11 +225,6 @@ void set_dorado_basecaller_args(utils::arg_parse::ArgParser& parser, int& verbos
                 .help("Require both ends of a read to be barcoded for a double ended barcode.")
                 .default_value(false)
                 .implicit_value(true);
-        parser.visible.add_argument("--disallow-inferior-barcodes")
-                .help("Declassify reads if a better barcode match over the threshold exists at "
-                      "either end of the read for double ended kits.")
-                .default_value(false)
-                .implicit_value(true);
         parser.visible.add_argument("--barcode-arrangement")
                 .help("Path to file with custom barcode arrangement. Requires --kit-name.");
         parser.visible.add_argument("--barcode-sequences")
@@ -751,8 +746,6 @@ int basecaller(int argc, char* argv[]) {
         barcoding_info = std::make_shared<demux::BarcodingInfo>();
         barcoding_info->kit_name = parser.visible.get<std::string>("--kit-name");
         barcoding_info->barcode_both_ends = parser.visible.get<bool>("--barcode-both-ends");
-        barcoding_info->disallow_inferior_barcodes =
-                parser.visible.get<bool>("--disallow-inferior-barcodes");
         barcoding_info->trim = !no_trim_barcodes;
 
         std::optional<std::string> custom_seqs =

diff --git a/dorado/cli/demux.cpp b/dorado/cli/demux.cpp
@@ -62,7 +62,6 @@ std::shared_ptr<const dorado::demux::BarcodingInfo> get_barcoding_info(
         return nullptr;
     }
     result->barcode_both_ends = parser.visible.get<bool>("--barcode-both-ends");
-    result->disallow_inferior_barcodes = parser.visible.get<bool>("--disallow-inferior-barcodes");
     result->trim = !parser.visible.get<bool>("--no-trim");
     if (sample_sheet) {
         result->allowed_barcodes = sample_sheet->get_barcode_values();
@@ -155,11 +154,6 @@ int demuxer(int argc, char* argv[]) {
             .help("Require both ends of a read to be barcoded for a double ended barcode.")
             .default_value(false)
             .implicit_value(true);
-    parser.visible.add_argument("--disallow-inferior-barcodes")
-            .help("Declassify reads if a better barcode match over the threshold exists at either "
-                  "end of the read for double ended kits.")
-            .default_value(false)
-            .implicit_value(true);
     parser.visible.add_argument("--no-trim")
             .help("Skip barcode trimming. If this option is not chosen, trimming is enabled. "
                   "Note that you should use this option if your input data is mapped and you "

diff --git a/dorado/demux/BarcodeClassifier.cpp b/dorado/demux/BarcodeClassifier.cpp
@@ -188,10 +188,9 @@ BarcodeClassifier::~BarcodeClassifier() = default;
 
 BarcodeScoreResult BarcodeClassifier::barcode(const std::string& seq,
                                               bool barcode_both_ends,
-                                              bool disallow_inferior_barcodes,
                                               const BarcodeFilterSet& allowed_barcodes) const {
-    auto best_barcode = find_best_barcode(seq, m_barcode_candidates, barcode_both_ends,
-                                          disallow_inferior_barcodes, allowed_barcodes);
+    auto best_barcode =
+            find_best_barcode(seq, m_barcode_candidates, barcode_both_ends, allowed_barcodes);
     return best_barcode;
 }
 
@@ -822,7 +821,6 @@ BarcodeScoreResult BarcodeClassifier::find_best_barcode(
         const std::string& read_seq,
         const std::vector<BarcodeCandidateKit>& candidates,
         bool barcode_both_ends,
-        bool disallow_inferior_barcodes,
         const BarcodeFilterSet& allowed_barcodes) const {
     if (read_seq.length() == 0) {
         return UNCLASSIFIED;
@@ -961,7 +959,7 @@ BarcodeScoreResult BarcodeClassifier::find_best_barcode(
         }
     }
 
-    if (kit.double_ends && disallow_inferior_barcodes) {
+    if (kit.double_ends) {
         // For more stringent classification, ensure that neither end of a read has a higher scoring
         //  barcode, if any of the barcodes at that end are better than the threshold.
         auto best_top_result = std::min_element(

diff --git a/dorado/demux/BarcodeClassifier.h b/dorado/demux/BarcodeClassifier.h
@@ -25,7 +25,6 @@ class BarcodeClassifier {
 
     BarcodeScoreResult barcode(const std::string& seq,
                                bool barcode_both_ends,
-                               bool disallow_inferior_barcodes,
                                const BarcodeFilterSet& allowed_barcodes) const;
 
 private:
@@ -57,7 +56,6 @@ class BarcodeClassifier {
     BarcodeScoreResult find_best_barcode(const std::string& read_seq,
                                          const std::vector<BarcodeCandidateKit>& adapter,
                                          bool barcode_both_ends,
-                                         bool disallow_inferior_barcodes,
                                          const BarcodeFilterSet& allowed_barcodes) const;
 };
 

diff --git a/dorado/demux/barcoding_info.h b/dorado/demux/barcoding_info.h
@@ -9,7 +9,6 @@ namespace dorado::demux {
 struct BarcodingInfo {
     std::string kit_name;
     bool barcode_both_ends{false};
-    bool disallow_inferior_barcodes{false};
     bool trim{false};
     BarcodeFilterSet allowed_barcodes;
 };

diff --git a/dorado/read_pipeline/BarcodeClassifierNode.cpp b/dorado/read_pipeline/BarcodeClassifierNode.cpp
@@ -89,7 +89,6 @@ void BarcodeClassifierNode::barcode(BamMessage& read, const demux::BarcodingInfo
     }
 
     auto bc_res = barcoder->barcode(seq, barcoding_info->barcode_both_ends,
-                                    barcoding_info->disallow_inferior_barcodes,
                                     barcoding_info->allowed_barcodes);
     auto bc = generate_barcode_string(bc_res);
     read.barcoding_result = std::make_shared<BarcodeScoreResult>(std::move(bc_res));
@@ -122,7 +121,6 @@ void BarcodeClassifierNode::barcode(SimplexRead& read) {
 
     // get the sequence to map from the record
     auto bc_res = barcoder->barcode(read.read_common.seq, barcoding_info->barcode_both_ends,
-                                    barcoding_info->disallow_inferior_barcodes,
                                     barcoding_info->allowed_barcodes);
     read.read_common.barcode = generate_barcode_string(bc_res);
     spdlog::trace("Barcode for {} is {}", read.read_common.read_id, read.read_common.barcode);

diff --git a/tests/BarcodeClassifierTest.cpp b/tests/BarcodeClassifierTest.cpp
@@ -32,15 +32,14 @@ namespace {
 std::shared_ptr<const demux::BarcodingInfo> create_barcoding_info(
         const std::string& kit_name,
         bool barcode_both_ends,
-        bool disallow_inferior_barcodes,
         bool trim_barcode,
         BarcodeFilterSet allowed_barcodes) {
     if (kit_name.empty()) {
         return {};
     }
 
-    auto result = demux::BarcodingInfo{kit_name, barcode_both_ends, disallow_inferior_barcodes,
-                                       trim_barcode, std::move(allowed_barcodes)};
+    auto result = demux::BarcodingInfo{kit_name, barcode_both_ends, trim_barcode,
+                                       std::move(allowed_barcodes)};
     return std::make_shared<demux::BarcodingInfo>(std::move(result));
 }
 
@@ -82,7 +81,7 @@ TEST_CASE("BarcodeClassifier: test single ended barcode", TEST_GROUP) {
         while (reader.read()) {
             auto seqlen = reader.record->core.l_qseq;
             std::string seq = utils::extract_sequence(reader.record.get());
-            auto res = classifier.barcode(seq, false, false, std::nullopt);
+            auto res = classifier.barcode(seq, false, std::nullopt);
             if (res.barcode_name == "unclassified") {
                 CHECK(bc == res.barcode_name);
             } else {
@@ -109,7 +108,7 @@ TEST_CASE("BarcodeClassifier: test double ended barcode", TEST_GROUP) {
         while (reader.read()) {
             auto seqlen = reader.record->core.l_qseq;
             std::string seq = utils::extract_sequence(reader.record.get());
-            auto res = classifier.barcode(seq, false, false, std::nullopt);
+            auto res = classifier.barcode(seq, false, std::nullopt);
             if (res.barcode_name == "unclassified") {
                 CHECK(bc == res.barcode_name);
             } else {
@@ -138,7 +137,7 @@ TEST_CASE("BarcodeClassifier: test double ended barcode with different variants"
         while (reader.read()) {
             auto seqlen = reader.record->core.l_qseq;
             std::string seq = utils::extract_sequence(reader.record.get());
-            auto res = classifier.barcode(seq, false, false, std::nullopt);
+            auto res = classifier.barcode(seq, false, std::nullopt);
             if (res.barcode_name == "unclassified") {
                 CHECK(bc == res.barcode_name);
             } else {
@@ -165,8 +164,8 @@ TEST_CASE("BarcodeClassifier: check barcodes on both ends - failing case", TEST_
     HtsReader reader(bc_file.string(), std::nullopt);
     while (reader.read()) {
         std::string seq = utils::extract_sequence(reader.record.get());
-        auto single_end_res = classifier.barcode(seq, false, false, std::nullopt);
-        auto double_end_res = classifier.barcode(seq, true, false, std::nullopt);
+        auto single_end_res = classifier.barcode(seq, false, std::nullopt);
+        auto double_end_res = classifier.barcode(seq, true, std::nullopt);
         CHECK(double_end_res.barcode_name == "unclassified");
         CHECK(single_end_res.barcode_name == "BC01");
     }
@@ -182,8 +181,8 @@ TEST_CASE("BarcodeClassifier: check barcodes on both ends - passing case", TEST_
     HtsReader reader(bc_file.string(), std::nullopt);
     while (reader.read()) {
         std::string seq = utils::extract_sequence(reader.record.get());
-        auto single_end_res = classifier.barcode(seq, false, false, std::nullopt);
-        auto double_end_res = classifier.barcode(seq, true, false, std::nullopt);
+        auto single_end_res = classifier.barcode(seq, false, std::nullopt);
+        auto double_end_res = classifier.barcode(seq, true, std::nullopt);
         CHECK(double_end_res.barcode_name == single_end_res.barcode_name);
         CHECK(single_end_res.barcode_name == "BC01");
     }
@@ -199,7 +198,7 @@ TEST_CASE("BarcodeClassifier: check presence of midstrand barcode double ended k
     HtsReader reader(bc_file.string(), std::nullopt);
     while (reader.read()) {
         std::string seq = utils::extract_sequence(reader.record.get());
-        auto res = classifier.barcode(seq, false, false, std::nullopt);
+        auto res = classifier.barcode(seq, false, std::nullopt);
         CHECK(res.barcode_name == "unclassified");
         CHECK(res.found_midstrand);
     }
@@ -215,7 +214,7 @@ TEST_CASE("BarcodeClassifier: check presence of midstrand barcode single ended k
     HtsReader reader(bc_file.string(), std::nullopt);
     while (reader.read()) {
         std::string seq = utils::extract_sequence(reader.record.get());
-        auto res = classifier.barcode(seq, false, false, std::nullopt);
+        auto res = classifier.barcode(seq, false, std::nullopt);
         CHECK(res.barcode_name == "unclassified");
         CHECK(res.found_midstrand);
     }
@@ -255,8 +254,7 @@ TEST_CASE(
     read->read_common.model_stride = stride;
 
     auto client_info = std::make_shared<dorado::DefaultClientInfo>();
-    auto barcoding_info =
-            create_barcoding_info(kit, barcode_both_ends, false, !no_trim, std::nullopt);
+    auto barcoding_info = create_barcoding_info(kit, barcode_both_ends, !no_trim, std::nullopt);
     client_info->contexts().register_context<const demux::BarcodingInfo>(std::move(barcoding_info));
     read->read_common.client_info = client_info;
 
@@ -404,8 +402,7 @@ TEST_CASE("BarcodeClassifierNode: test for proper trimming and alignment data st
     reader.read();
 
     auto client_info = std::make_shared<dorado::DefaultClientInfo>();
-    auto barcoding_info =
-            create_barcoding_info(kit, barcode_both_ends, false, !no_trim, std::nullopt);
+    auto barcoding_info = create_barcoding_info(kit, barcode_both_ends, !no_trim, std::nullopt);
     client_info->contexts().register_context<const demux::BarcodingInfo>(std::move(barcoding_info));
 
     BamPtr read1(bam_dup1(reader.record.get()));
@@ -512,7 +509,7 @@ TEST_CASE("BarcodeClassifier: test custom kit with double ended barcode", TEST_G
         while (reader.read()) {
             auto seqlen = reader.record->core.l_qseq;
             std::string seq = utils::extract_sequence(reader.record.get());
-            auto res = classifier.barcode(seq, false, false, std::nullopt);
+            auto res = classifier.barcode(seq, false, std::nullopt);
             if (res.barcode_name == "unclassified") {
                 CHECK(bc == res.barcode_name);
             } else {