Merge branch 'smalton/DOR-756-cherry-pick-demux-segfault' into 'relea…

…se-v0.7' DOR-756: Cherry pick demux segfault fix to 0.7 branch See merge request machine-learning/dorado!1072
nanoporetech · Jun 12, 2024 · 3621800 · 3621800
2 parents c670201 + 8a41fc8
commit 3621800
Show file tree

Hide file tree

Showing 5 changed files with 20 additions and 7 deletions.
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
@@ -613,6 +613,7 @@ test:macos:x64:
     BATCH: 0
     MODEL_SPEED: hac
     MODEL_VERSION: v4.2.0
+  timeout: 1h 30m
 
 # Test that you can run dorado in a clean cuda 20.04 environment
 test_archive:linux:x86:20.04_nvidia:

diff --git a/dorado/demux/AdapterDetector.cpp b/dorado/demux/AdapterDetector.cpp
@@ -1,7 +1,6 @@
 #include "AdapterDetector.h"
 
 #include "parse_custom_sequences.h"
-#include "utils/alignment_utils.h"
 #include "utils/parse_custom_kit.h"
 #include "utils/sequence_utils.h"
 #include "utils/types.h"
@@ -27,7 +26,7 @@ const int PRIMER_TRIM_LENGTH = 150;
 EdlibAlignConfig init_edlib_config_for_adapters() {
     EdlibAlignConfig placement_config = edlibDefaultAlignConfig();
     placement_config.mode = EDLIB_MODE_HW;
-    placement_config.task = EDLIB_TASK_PATH;
+    placement_config.task = EDLIB_TASK_LOC;
     // Currently none of our adapters or primers have Ns, but we should support them.
     static const EdlibEqualityPair additionalEqualities[4] = {
             {'N', 'A'}, {'N', 'T'}, {'N', 'C'}, {'N', 'G'}};
@@ -122,8 +121,13 @@ const std::vector<AdapterDetector::Query>& AdapterDetector::get_primer_sequences
 static SingleEndResult copy_results(const EdlibAlignResult& source,
                                     const std::string& name,
                                     size_t length) {
-    SingleEndResult dest;
+    SingleEndResult dest{};
     dest.name = name;
+
+    if (source.status != EDLIB_STATUS_OK || !source.startLocations || !source.endLocations) {
+        return dest;
+    }
+
     dest.score = 1.0f - float(source.editDistance) / length;
     dest.position = {source.startLocations[0], source.endLocations[0]};
     return dest;
@@ -219,8 +223,13 @@ AdapterScoreResult AdapterDetector::detect(const std::string& seq,
             }
         }
     }
-    result.front = front_results[best_front];
-    result.rear = rear_results[best_rear];
+
+    if (best_front != -1) {
+        result.front = front_results[best_front];
+    }
+    if (best_rear != -1) {
+        result.rear = rear_results[best_rear];
+    }
     return result;
 }
 

diff --git a/dorado/demux/Trimmer.cpp b/dorado/demux/Trimmer.cpp
@@ -75,9 +75,9 @@ std::pair<int, int> Trimmer::determine_trim_interval(const BarcodeScoreResult& r
         // window.
         if (trim_interval.second <= trim_interval.first) {
             if (res.use_top) {
-                return {res.top_barcode_pos.second, seqlen};
+                trim_interval = {res.top_barcode_pos.second, seqlen};
             } else {
-                return {0, res.bottom_barcode_pos.first};
+                trim_interval = {0, res.bottom_barcode_pos.first};
             }
         }
     } else {

diff --git a/tests/data/pod5/degenerate/overtrim.pod5 b/tests/data/pod5/degenerate/overtrim.pod5
diff --git a/tests/test_simple_basecaller_execution.sh b/tests/test_simple_basecaller_execution.sh
@@ -78,6 +78,9 @@ if $dorado_bin basecaller $model_5k_v43 $data_dir/duplex/pod5 --modified-bases 5
 fi
 set -e
 
+# Check that dorado handles degenerate reads without crashing
+$dorado_bin basecaller $model_5k_v43 $data_dir/pod5/degenerate/overtrim.pod5 -b ${batch} --skip-model-compatibility-check --kit-name EXP-NBD196 > $output_dir/error_condition.fq
+
 echo dorado summary test stage
 $dorado_bin summary $output_dir/calls.bam