Added custom models as submodule, added pipleline for trap species handling (up to alignment). Initial progress on custom model rules

workflow/config/dependencies.yaml

@@ -1,3 +1,4 @@
 arch: "amd64"
 
 datasets_binary : "binaries/datasets"
+ont_dcs_fasta : '../reference/ont_control_sequence.fasta'

workflow/config/main.yaml

@@ -1,5 +1,7 @@
 tmp_dir: "/tmp"
 
+seed : 0
+
 # dorado
 dorado_model_dir : "../dorado_models"
 hac_model_name: "dna_r10.4.1_e8.2_400bps_hac@v6.0.0"
@@ -13,4 +15,9 @@ fast_min_q: 8 # empty string or 0 to disable
 # squiqqle dataset
 pod5_dataset: "nomiss_96BC_P2I_SUP_2026"
 pod5_dataset_size : 293
-pod5_stride : 12 # Use 1 in x pod5 files from the ONT NO-MISS dataset to reduce dataset size / compute requirements
+pod5_stride : 600 # Use 1 in x pod5 files from the ONT NO-MISS dataset to reduce dataset size / compute requirements
+
+# custom ML
+torch_env : '../envs/torch_cpu.yaml'
+budget_tolerance : 0.1
+reads_per_genus : 5000