Merge pull request #102 from theiagen/smw-patch-dev

Snippy Gene Query fixes

tasks/gene_typing/task_snippy_gene_query.wdl

@@ -6,7 +6,7 @@ task snippy_gene_query {
     File snippy_variants_results
     String? query_gene
     File? reference
-    String docker = "quay.io/theiagen/utility:1.2"
+    String docker = "quay.io/theiagen/terra-tools:2023-06-21"
     Int cpus = 8
     Int memory = 32
   }
@@ -15,7 +15,7 @@ task snippy_gene_query {
     if [ -z "~{query_gene}" ]; then 
         no_hit="NA: No query gene was provided"
     else 
-        no_hit="No variants identified in queried genes (~{query_gene})" 
+        no_hit="No variants identified in queried genes (~{query_gene}) relative to the reference genome " 
     fi
 
     # if provided, check that query gene strings are present in reference genome
@@ -43,7 +43,8 @@ task snippy_gene_query {
       if grep -q  "${qgene}" ~{snippy_variants_results}; then 
         grep "${qgene}" ~{snippy_variants_results} | awk '{print "'~{samplename}'," $0}' >> ./gene_query.csv
         # curate relevant columns of queried hits to single output
-        grep "${qgene}" ./gene_query.csv | awk -F"," '{print "'${qgene}': "$15" ("$12"; "$7")"}' >> snippy_variant_hits_tmp
+        # awk syntax: see https://stackoverflow.com/questions/29642102/how-to-make-awk-ignore-the-field-delimiter-inside-double-quotes?noredirect=1&lq=1
+        grep "${qgene}" ./gene_query.csv | awk -vFPAT='([^,]*)|("[^"]+")' -vOFS=, '{print "'${qgene}': "$15" ("$12"; "$7")"}' >> snippy_variant_hits_tmp
       fi
     done
 

tasks/gene_typing/task_snippy_variants.wdl

@@ -54,6 +54,7 @@ task snippy_variants {
   output {
     String snippy_variants_version = read_string("VERSION")
     String snippy_variants_docker = docker
+    File snippy_variants_reference_genome = "~{reference_genome_file}"
     File snippy_variants_outdir_tarball = "./~{samplename}_snippy_variants_outdir.tar"
     Array[File] snippy_variants_outputs = glob("~{samplename}/~{samplename}*")
     File snippy_variants_results = "~{samplename}/~{samplename}.csv"

tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml

@@ -563,7 +563,7 @@
     - path: miniwdl_run/wdl/tasks/gene_typing/task_resfinder.wdl
       md5sum: b50d1c42e23fe6643e127c539d03eda3
     - path: miniwdl_run/wdl/tasks/gene_typing/task_snippy_variants.wdl
-      md5sum: 138cac57599c2fff7c6646d60033d2ec
+      md5sum: 72185fe726e83be5242a7e11401823a9
     - path: miniwdl_run/wdl/tasks/quality_control/task_bbduk.wdl
       md5sum: 501ac92b8a4a605ec178d879e40e9b00
     - path: miniwdl_run/wdl/tasks/quality_control/task_busco.wdl
@@ -638,7 +638,7 @@
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
       md5sum: e98bdaa0c00d01b5e77859713ab91689
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: 01f665f21ec65aa77a43a68694c4f1f3
+      md5sum: 33f917eae3094f0c74ba895b3b1d4c9d
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_pe.wdl
       md5sum: 40d4e09a82030c8219b37f883cddaca4
     - path: miniwdl_run/workflow.log

tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml

@@ -533,7 +533,7 @@
     - path: miniwdl_run/wdl/tasks/gene_typing/task_resfinder.wdl
       md5sum: b50d1c42e23fe6643e127c539d03eda3
     - path: miniwdl_run/wdl/tasks/gene_typing/task_snippy_variants.wdl
-      md5sum: 138cac57599c2fff7c6646d60033d2ec
+      md5sum: 72185fe726e83be5242a7e11401823a9
     - path: miniwdl_run/wdl/tasks/quality_control/task_bbduk.wdl
       md5sum: 501ac92b8a4a605ec178d879e40e9b00
     - path: miniwdl_run/wdl/tasks/quality_control/task_busco.wdl
@@ -606,7 +606,7 @@
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
       md5sum: 17be58829cc4e6a1780cf227a2ec99ac
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: 01f665f21ec65aa77a43a68694c4f1f3
+      md5sum: 33f917eae3094f0c74ba895b3b1d4c9d
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl
       md5sum: 53d322d895837c0bcb049786572e944d
     - path: miniwdl_run/workflow.log

workflows/theiaeuk/wf_theiaeuk_illumina_pe.wdl

@@ -276,6 +276,7 @@ workflow theiaeuk_illumina_pe {
     String? theiaeuk_snippy_variants_query = merlin_magic.snippy_variants_query
     String? theiaeuk_snippy_variants_query_check = merlin_magic.snippy_variants_query_check
     String? theiaeuk_snippy_variants_hits = merlin_magic.snippy_variants_hits
+    File? theiaeuk_snippy_variants_reference_genome = merlin_magic.snippy_variants_reference_genome
     File? theiaeuk_snippy_variants_gene_query_results = merlin_magic.snippy_variants_gene_query_results
     # Array[File]? snippy_outputs = merlin_magic.snippy_outputs
     File? theiaeuk_snippy_variants_results = merlin_magic.snippy_variants_results

workflows/utilities/wf_merlin_magic.wdl

@@ -639,6 +639,7 @@ workflow merlin_magic {
     String? cladetyper_docker_image = cladetyper.gambit_cladetyper_docker_image
     String? cladetype_annotated_ref = cladetyper.clade_spec_ref
     # snippy variants
+    String snippy_variants_reference_genome = select_first([snippy_cauris.snippy_variants_reference_genome, snippy_afumigatus.snippy_variants_reference_genome, snippy_crypto.snippy_variants_reference_genome, "No reference genome detected"])
     String snippy_variants_version = select_first([snippy_cauris.snippy_variants_version, snippy_afumigatus.snippy_variants_version, snippy_crypto.snippy_variants_version, "No matching taxon detected"])
     String snippy_variants_query = select_first([snippy_gene_query_cauris.snippy_variants_query, snippy_gene_query_afumigatus.snippy_variants_query, snippy_gene_query_crypto.snippy_variants_query, "No matching taxon detected"])
     String snippy_variants_query_check = select_first([snippy_gene_query_cauris.snippy_variants_query_check, snippy_gene_query_afumigatus.snippy_variants_query_check, snippy_gene_query_crypto.snippy_variants_query_check, "No matching taxon detected"])