[hicap] expose optional parameters (#724)

* Enable optional parameters in hicap * Remove "--full_sequence" parameter and other comments that are no longer needed. * Update md5sum for wf_merlin_magic in illumina PE and SE test workflows * Updated documentation to reflect additional optional parameters in the merlin_magic workflow * Update md5sum for wf_merlin_magic in test workflows
theiagen · Jan 17, 2025 · 204aff5 · 204aff5
1 parent 2a8c304
commit 204aff5
Show file tree

Hide file tree

Showing 5 changed files with 27 additions and 22 deletions.
diff --git a/docs/workflows/genomic_characterization/theiaprok.md b/docs/workflows/genomic_characterization/theiaprok.md
@@ -330,7 +330,11 @@ All input reads are processed through "[core tasks](#core-tasks-performed-for-al
 | merlin_magic | **ectyper_verify** | Boolean | Set to true to enable E. coli species verification | False  | Optional | FASTA, ONT, PE, SE |
 | merlin_magic | **emmtypingtool_docker_image** | String | The Docker container to use for the task | us-docker.pkg.dev/general-theiagen/staphb/emmtypingtool:0.0.1 | Optional | FASTA, ONT, PE, SE |
 | merlin_magic | **genotyphi_docker_image** | String | The Docker container to use for the task | us-docker.pkg.dev/general-theiagen/staphb/mykrobe:0.11.0 | Optional | FASTA, ONT, PE, SE |
+| merlin_magic | **hicap_broken_gene_identity** | Float | Minimum percentage identity to consider a broken gene | 0.80 | Optional | FASTA, ONT, PE, SE |
+| merlin_magic | **hicap_broken_gene_length** | Int | Minimum length to consider a broken gene | 60 | Optional | FASTA, ONT, PE, SE |
 | merlin_magic | **hicap_docker_image** | String | The Docker container to use for the task | us-docker.pkg.dev/general-theiagen/biocontainers/hicap:1.0.3--py_0 | Optional | FASTA, ONT, PE, SE |
+| merlin_magic | **hicap_gene_coverage** | Float | Minimum percentage coverage to consider a single gene complete | 0.80 | Optional | FASTA, ONT, PE, SE |
+| merlin_magic | **hicap_gene_identity** | Float | Minimum percentage identity to consider a single gene complete | 0.70 | Optional | FASTA, ONT, PE, SE |
 | merlin_magic | **kaptive_docker_image** | String | The Docker container to use for the task | us-docker.pkg.dev/general-theiagen/staphb/kaptive:2.0.3 | Optional | FASTA, ONT, PE, SE |
 | merlin_magic | **kaptive_low_gene_id** | Float | Percent identity threshold for what counts as a low identity match in the gene BLAST search | 95 | Optional | FASTA, ONT, PE, SE |
 | merlin_magic | **kaptive_min_coverage** | Float | Minimum required percent identity for the gene BLAST search via tBLASTn | 80 | Optional | FASTA, ONT, PE, SE |

diff --git a/tasks/species_typing/haemophilus/task_hicap.wdl b/tasks/species_typing/haemophilus/task_hicap.wdl
@@ -7,27 +7,14 @@ task hicap {
   input {
     File assembly
     String samplename
+    Float gene_coverage = 0.80 #Minimum percentage coverage to consider a single gene complete. [default: 0.80]
+    Float gene_identity = 0.70 #Minimum percentage identity to consider a single gene complete. [default: 0.70]
+    Float broken_gene_identity = 0.80 #Minimum percentage identity to consider a broken gene. [default: 0.80]
+    Int broken_gene_length = 60 #Minimum length to consider a broken gene. [default: 60]
     String docker = "us-docker.pkg.dev/general-theiagen/biocontainers/hicap:1.0.3--py_0"
     Int cpu = 2
     Int memory = 8
     Int disk_size = 50
-
-    #Parameters
-    #-q QUERY_FP, --query_fp QUERY_FP              Input FASTA query
-    #-o OUTPUT_DIR, --output_dir OUTPUT_DIR        Output directory
-    #-d DATABASE_DIR, --database_dir DATABASE_DIR  Directory containing locus database. [default: /usr/local/lib/python3.6/site-
-    #                                              packages/hicap/database]
-    #-m MODEL_FP, --model_fp MODEL_FP              Path to prodigal model. [default: /usr/local/lib/python3.6/site-
-    #                                              packages/hicap/model/prodigal_hi.bin]
-    #-s, --full_sequence                           Write the full input sequence out to the genbank file rather than just the region
-    #                                              surrounding and including the locus
-    #--gene_coverage GENE_COVERAGE                 Minimum percentage coverage to consider a single gene complete. [default: 0.80]
-    #--gene_identity GENE_IDENTITY                 Minimum percentage identity to consider a single gene complete. [default: 0.70]
-    #--broken_gene_length BROKEN_GENE_LENGTH       Minimum length to consider a broken gene. [default: 60]
-    #--broken_gene_identity BROKEN_GENE_IDENTITY   Minimum percentage identity to consider a broken gene. [default: 0.80]
-    #--threads THREADS                             Threads to use for BLAST+. [default: 1]
-    #--log_fp LOG_FP                               Record logging messages to file
-    #--debug                                       Print debug messages
   }
   command <<<
     echo $(hicap --version 2>&1) | sed 's/^hicap //' | tee VERSION
@@ -36,8 +23,13 @@ task hicap {
 
     hicap \
       -q ~{assembly} \
-      -o output_dir
-
+      -o output_dir \
+      --gene_coverage ~{gene_coverage} \
+      --gene_identity ~{gene_identity} \
+      --broken_gene_length ~{broken_gene_length} \
+      --broken_gene_identity ~{broken_gene_identity} \
+      --threads ~{cpu}
+
     filename=$(basename ~{assembly})
 
     # if there are no hits for a cap locus, no file is produced

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -516,7 +516,7 @@
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
       md5sum: 9b8e2da62c8572a369c786a9bbc3a36e
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: 3547638e39dcf408234a6c1df57d198d
+      md5sum: 1bcb64a5e28d26e3603f456e99bfa7b6
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_pe.wdl
       contains: ["version", "QC", "output"]
     - path: miniwdl_run/workflow.log

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -487,7 +487,7 @@
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
       md5sum: 02dc0075bf28d557d7b81aa2dc61feab
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: 3547638e39dcf408234a6c1df57d198d
+      md5sum: 1bcb64a5e28d26e3603f456e99bfa7b6
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl
       md5sum: 09d9f68b9ca8bf94b6145ff9bed2edd1
     - path: miniwdl_run/workflow.log

diff --git a/workflows/utilities/wf_merlin_magic.wdl b/workflows/utilities/wf_merlin_magic.wdl
@@ -131,6 +131,11 @@ workflow merlin_magic {
     Int? emmtyper_min_perfect
     Int? emmtyper_min_good
     Int? emmtyper_max_size
+    #hicap options
+    Float? hicap_gene_coverage
+    Float? hicap_gene_identity
+    Float? hicap_broken_gene_identity
+    Int? hicap_broken_gene_length
     # kaptive options
     Int? kaptive_start_end_margin
     Float? kaptive_min_identity
@@ -600,7 +605,11 @@ workflow merlin_magic {
       input:
         assembly = assembly,
         samplename = samplename,
-        docker = hicap_docker_image
+        docker = hicap_docker_image,
+        gene_coverage = hicap_gene_coverage,
+        gene_identity = hicap_gene_identity,
+        broken_gene_identity = hicap_broken_gene_identity,
+        broken_gene_length = hicap_broken_gene_length
     }
   }
   if (merlin_tag == "Vibrio" || merlin_tag == "Vibrio cholerae") {