Merge branch 'jdaw/fix-read-rc' into 'release-v0.6'

Fix bug with loading reverse aligned reads See merge request machine-learning/dorado!989
nanoporetech · May 9, 2024 · a64492b · a64492b
2 parents 6cc278f + b037472
commit a64492b
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 0 deletions.
diff --git a/dorado/read_pipeline/AdapterDetectorNode.cpp b/dorado/read_pipeline/AdapterDetectorNode.cpp
@@ -4,6 +4,7 @@
 #include "demux/Trimmer.h"
 #include "utils/bam_utils.h"
 #include "utils/barcode_kits.h"
+#include "utils/sequence_utils.h"
 #include "utils/trim.h"
 #include "utils/types.h"
 
@@ -57,7 +58,11 @@ void AdapterDetectorNode::input_thread_fn() {
 
 void AdapterDetectorNode::process_read(BamPtr& read) {
     bam1_t* irecord = read.get();
+    bool is_input_reversed = irecord->core.flag & BAM_FREVERSE;
     std::string seq = utils::extract_sequence(irecord);
+    if (is_input_reversed) {
+        seq = utils::reverse_complement(seq);
+    }
     int seqlen = irecord->core.l_qseq;
 
     std::pair<int, int> adapter_trim_interval = {0, seqlen};

diff --git a/dorado/read_pipeline/BarcodeClassifierNode.cpp b/dorado/read_pipeline/BarcodeClassifierNode.cpp
@@ -4,6 +4,7 @@
 #include "demux/Trimmer.h"
 #include "utils/SampleSheet.h"
 #include "utils/bam_utils.h"
+#include "utils/sequence_utils.h"
 #include "utils/trim.h"
 #include "utils/types.h"
 
@@ -108,7 +109,11 @@ void BarcodeClassifierNode::barcode(BamPtr& read) {
     auto barcoder = m_barcoder_selector.get_barcoder(*m_default_barcoding_info);
 
     bam1_t* irecord = read.get();
+    bool is_input_reversed = irecord->core.flag & BAM_FREVERSE;
     std::string seq = utils::extract_sequence(irecord);
+    if (is_input_reversed) {
+        seq = utils::reverse_complement(seq);
+    }
 
     auto bc_res = barcoder->barcode(seq, m_default_barcoding_info->barcode_both_ends,
                                     m_default_barcoding_info->allowed_barcodes);
@@ -152,6 +157,10 @@ void BarcodeClassifierNode::barcode(SimplexRead& read) {
     }
 
     m_num_records++;
+    {
+        std::lock_guard lock(m_barcode_count_mutex);
+        m_barcode_count[read.read_common.barcode]++;
+    }
 }
 
 stats::NamedStats BarcodeClassifierNode::sample_stats() const {