Merge branch 'jdaw/prevent-same-mods-being-called' into 'master'

Prevent users from running models with overlapping mods

Closes DOR-486

See merge request machine-learning/dorado!781

dorado/api/runner_creation.cpp

@@ -1,6 +1,7 @@
 #include "runner_creation.h"
 
 #include "basecall/crf_utils.h"
+#include "modbase/ModBaseModelConfig.h"
 
 #if DORADO_GPU_BUILD
 #ifdef __APPLE__
@@ -142,6 +143,8 @@ std::vector<modbase::RunnerPtr> create_modbase_runners(
         return {};
     }
 
+    modbase::check_modbase_multi_model_compatibility(remora_models);
+
     // generate model callers before nodes or it affects the speed calculations
     std::vector<modbase::RunnerPtr> remora_runners;
     std::vector<std::string> modbase_devices;

dorado/modbase/ModBaseModelConfig.cpp

@@ -130,4 +130,29 @@ ModBaseInfo get_modbase_info(
     return result;
 }
 
+void check_modbase_multi_model_compatibility(
+        const std::vector<std::filesystem::path>& modbase_models) {
+    std::string err_msg = "";
+    for (size_t i = 0; i < modbase_models.size(); i++) {
+        auto ref_model = load_modbase_model_config(modbase_models[i]);
+        const auto& ref_motif = ref_model.motif[ref_model.motif_offset];
+        for (size_t j = i + 1; j < modbase_models.size(); j++) {
+            auto query_model = load_modbase_model_config(modbase_models[j]);
+            const auto& query_motif = query_model.motif[query_model.motif_offset];
+
+            if (ref_motif == query_motif) {
+                err_msg += modbase_models[i].string() + " and " + modbase_models[j].string() +
+                           " have overlapping canonical motif: " + ref_motif;
+            }
+        }
+    }
+
+    if (!err_msg.empty()) {
+        throw std::runtime_error(
+                "Following are incompatible modbase models. Please select only one of them to "
+                "run:\n" +
+                err_msg);
+    }
+}
+
 }  // namespace dorado::modbase

dorado/modbase/ModBaseModelConfig.h

@@ -26,10 +26,13 @@ struct ModBaseModelConfig {
     bool reverse_signal{false};           ///< Reverse model data before processing (rna model)
 };
 
-ModBaseModelConfig load_modbase_model_config(const std::filesystem::path & model_path);
+ModBaseModelConfig load_modbase_model_config(const std::filesystem::path& model_path);
 
 // Determine the modbase alphabet from parameters and calculate offset positions for the results
 ModBaseInfo get_modbase_info(
-        const std::vector<std::reference_wrapper<const ModBaseModelConfig>> & base_mod_params);
+        const std::vector<std::reference_wrapper<const ModBaseModelConfig>>& base_mod_params);
+
+void check_modbase_multi_model_compatibility(
+        const std::vector<std::filesystem::path>& modbase_models);
 
 }  // namespace dorado::modbase

tests/test_simple_basecaller_execution.sh

@@ -15,6 +15,7 @@ dorado_bin=$(cd "$(dirname $1)"; pwd -P)/$(basename $1)
 model_name=${2:-dna_r10.4.1_e8.2_400bps_hac@v4.1.0}
 batch=${3:-384}
 model_name_5k=${4:-dna_r10.4.1_e8.2_400bps_hac@v4.2.0}
+model_name_5k_v43=${4:-dna_r10.4.1_e8.2_400bps_hac@v4.3.0}
 data_dir=$test_dir/data
 output_dir_name=$(echo $RANDOM | head -c 10)
 output_dir=${test_dir}/${output_dir_name}
@@ -27,6 +28,8 @@ $dorado_bin download --model ${model_name} --directory ${output_dir}
 model=${output_dir}/${model_name}
 $dorado_bin download --model ${model_name_5k} --directory ${output_dir}
 model_5k=${output_dir}/${model_name_5k}
+$dorado_bin download --model ${model_name_5k_v43} --directory ${output_dir}
+model_5k_v43=${output_dir}/${model_name_5k_v43}
 
 echo dorado basecaller test stage
 $dorado_bin basecaller ${model} $data_dir/pod5 -b ${batch} --emit-fastq > $output_dir/ref.fq
@@ -56,6 +59,10 @@ if $dorado_bin basecaller ${model} $data_dir/pod5 -b ${batch} --emit-fastq --mod
     echo  "Error: dorado basecaller should fail with combination of emit-fastq and modbase!"
     exit 1
 fi
+if $dorado_bin basecaller $model_5k_v43 $data_dir/duplex/pod5 --modified-bases 5mC_5hmC 5mCG_5hmCG > $output_dir/error_condition.fq; then 
+    echo  "Error: dorado basecaller should fail with multiple modbase configs having overlapping mods!"
+    exit 1
+fi
 set -e
 
 echo dorado summary test stage