[TheiaProk wfs] upgrade StxTyper version and OPERON outputs

docs/workflows/genomic_characterization/theiaprok.md

@@ -2055,8 +2055,10 @@ The TheiaProk workflows automatically activate taxa-specific sub-workflows after
 | staphopiasccmec_types_and_mecA_presence | String | staphopia-sccmec Hamming distance file | FASTA, ONT, PE, SE |
 | staphopiasccmec_version | String | staphopia-sccmec presence and absence TSV file | FASTA, ONT, PE, SE |
 | stxtyper_all_hits | String | Comma-separated list of matches of all types. Includes complete, partial, frameshift, internal stop, and novel hits. List is de-duplicated so multiple identical hits are only listed once. For example if 5 partial stx2 hits are detected in the genome, only 1 "stx2" will be listed in this field. To view the potential subtype for each partial hit, the user will need to view the stxtyper_report TSV file. | FASTA, ONT, PE, SE |
+| stxtyper_ambiguous_hits | String | Comma-separated list of matches that have the OPERON output of "AMBIGUOUS". Ambiguous bases found in the query sequence (e.g., N) | FASTA, ONT, PE, SE |
 | stxtyper_complete_operons | String | Comma-separated list of all COMPLETE operons detected by StxTyper. Show multiple hits if present in results. | FASTA, ONT, PE, SE |
 | stxtyper_docker | String | Name of docker image used by the stxtyper task. | FASTA, ONT, PE, SE |
+| stxtyper_extended_operons | String | Comma-separated list of all EXTENDED operons detected by StxTyper if coding sequence extends beyond the reference stop codon for one or both of the reference proteins. | FASTA, ONT, PE, SE |
 | stxtyper_novel_hits | String | Comma-separated list of matches that have the OPERON output of "COMPLETE_NOVEL". Possible outputs "stx1", "stx2", or "stx1,stx2" | FASTA, ONT, PE, SE |
 | stxtyper_num_hits | Int | Number of "hits" or rows present in the `stxtyper_report` TSV file | FASTA, ONT, PE, SE |
 | stxtyper_partial_hits | String | Possible outputs "stx1", "stx2", or "stx1,stx2". Tells the user that there was a partial hit to either the A or B subunit, but does not describe which subunit, only the possible types from the PARTIAL matches. | FASTA, ONT, PE, SE |

tasks/species_typing/escherichia_shigella/task_stxtyper.wdl

@@ -5,7 +5,7 @@ task stxtyper {
     File assembly
     String samplename
     Boolean enable_debugging = false # Additional messages are printed and files in $TMPDIR are not removed after running
-    String docker = "us-docker.pkg.dev/general-theiagen/staphb/stxtyper:1.0.24"
+    String docker = "us-docker.pkg.dev/general-theiagen/staphb/stxtyper:1.0.40"
     Int disk_size = 50
     Int cpu = 1
     Int memory = 4
@@ -27,6 +27,7 @@ task stxtyper {
       --nucleotide ~{assembly} \
       --name ~{samplename} \
       --output ~{samplename}_stxtyper.tsv \
+      --threads ~{cpu} \
       ~{true='--debug' false='' enable_debugging} \
       --log ~{samplename}_stxtyper.log
 
@@ -40,9 +41,9 @@ task stxtyper {
       echo "DEBUG: No hits found in StxTyper output TSV. Exiting task with exit code 0 now."
 
       # create empty output files
-      touch stxtyper_all_hits.txt stxtyper_complete_operons.txt stxtyper_partial_hits.txt stxtyper_stx_frameshifts_or_internal_stop_hits.txt  stx_novel_hits.txt
+      touch stxtyper_all_hits.txt stxtyper_complete_operons.txt stxtyper_partial_hits.txt stxtyper_stx_frameshifts_or_internal_stop_hits.txt  stx_novel_hits.txt stxtyper_extended_operons.txt stxtyper_ambiguous_hits.txt
       # put "none" into all of them so task does not fail
-      echo "None" | tee stxtyper_all_hits.txt stxtyper_complete_operons.txt stxtyper_partial_hits.txt stxtyper_stx_frameshifts_or_internal_stop_hits.txt stx_novel_hits.txt 
+      echo "None" | tee stxtyper_all_hits.txt stxtyper_complete_operons.txt stxtyper_partial_hits.txt stxtyper_stx_frameshifts_or_internal_stop_hits.txt stx_novel_hits.txt stxtyper_extended_operons.txt stxtyper_ambiguous_hits.txt
       exit 0
     fi
 
@@ -88,11 +89,25 @@ task stxtyper {
       if [ "$(grep --silent -E 'FRAMESHIFT|INTERNAL_STOP' ~{samplename}_stxtyper.tsv; echo $?)" -gt 0 ]; then
         echo "None" > stxtyper_stx_frameshifts_or_internal_stop_hits.txt
       fi
+
+      ### extended operons
+      echo "DEBUG: Parsing extended operons..."
+      awk -F'\t' -v OFS=, '$4 == "EXTENDED" {print $3}' ~{samplename}_stxtyper.tsv | paste -sd, - | tee stxtyper_extended_operons.txt
+      if [ "$(grep --silent 'EXTENDED' ~{samplename}_stxtyper.tsv; echo $?)" -gt 0 ]; then
+        echo "None" > stxtyper_extended_operons.txt
+      fi
+
+      ### ambiguous hits
+      echo "DEBUG: Parsing ambiguous hits..."
+      awk -F'\t' -v OFS=, '$4 == "AMBIGUOUS" {print $3}' ~{samplename}_stxtyper.tsv | paste -sd, - | tee stxtyper_ambiguous_hits.txt
+      if [ "$(grep --silent 'AMBIGUOUS' ~{samplename}_stxtyper.tsv; echo $?)" -gt 0 ]; then
+        echo "None" > stxtyper_ambiguous_hits.txt
+      fi
 
       echo "DEBUG: generating stx_type_all string output now..."
       # sort and uniq so there are no duplicates; then paste into a single comma-separated line with commas
       # sed is to remove any instances of "None" from the output
-      cat stxtyper_complete_operons.txt stxtyper_partial_hits.txt stxtyper_stx_frameshifts_or_internal_stop_hits.txt stx_novel_hits.txt | sed '/None/d' | sort | uniq | paste -sd, - > stxtyper_all_hits.txt
+      cat stxtyper_complete_operons.txt stxtyper_partial_hits.txt stxtyper_stx_frameshifts_or_internal_stop_hits.txt stx_novel_hits.txt stxtyper_extended_operons.txt stxtyper_ambiguous_hits.txt | sed '/None/d' | sort | uniq | paste -sd, - > stxtyper_all_hits.txt
 
     fi
     echo "DEBUG: Finished parsing StxTyper output TSV."
@@ -109,6 +124,8 @@ task stxtyper {
     String stxtyper_partial_hits = read_string("stxtyper_partial_hits.txt")
     String stxtyper_frameshifts_or_internal_stop_hits =  read_string("stxtyper_stx_frameshifts_or_internal_stop_hits.txt")
     String stxtyper_novel_hits = read_string("stx_novel_hits.txt")
+    String stxtyper_extended_operons = read_string("stxtyper_extended_operons.txt")
+    String stxtyper_ambiguous_hits = read_string("stxtyper_ambiguous_hits.txt")
   }
   runtime {
     docker: "~{docker}"

tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml

@@ -514,9 +514,9 @@
     - path: miniwdl_run/wdl/tasks/utilities/data_export/task_broad_terra_tools.wdl
       md5sum: 59e18911ba07c16e01df38abe0e70477
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
-      md5sum: a6130333e29e84f7fc5488fd717cee95
+      md5sum: 7215782a19b40e3414a622b1db758c4f
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: adf5789053a7f720f4555be4472270d0
+      md5sum: 98d1cf63c946ec480fdab9b8a0e1b4dd
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_pe.wdl
       contains: ["version", "QC", "output"]
     - path: miniwdl_run/workflow.log

tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml

@@ -485,9 +485,9 @@
     - path: miniwdl_run/wdl/tasks/utilities/data_export/task_broad_terra_tools.wdl
       md5sum: 59e18911ba07c16e01df38abe0e70477
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
-      md5sum: c019ebbd2d88234d580a83d776de0d4a
+      md5sum: 2ef93d54c79615b57df794b91965d7e4
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
-      md5sum: adf5789053a7f720f4555be4472270d0
+      md5sum: 98d1cf63c946ec480fdab9b8a0e1b4dd
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl
       md5sum: 09d9f68b9ca8bf94b6145ff9bed2edd1
     - path: miniwdl_run/workflow.log

workflows/theiaprok/wf_theiaprok_fasta.wdl

@@ -586,6 +586,8 @@ workflow theiaprok_fasta {
     String? stxtyper_partial_hits = merlin_magic.stxtyper_partial_hits
     String? stxtyper_stx_frameshifts_or_internal_stop_hits =  merlin_magic.stxtyper_stx_frameshifts_or_internal_stop_hits
     String? stxtyper_novel_hits = merlin_magic.stxtyper_novel_hits
+    String? stxtyper_extended_operons = merlin_magic.stxtyper_extended_operons
+    String? stxtyper_ambiguous_hits = merlin_magic.stxtyper_ambiguous_hits
     # Listeria Typing
     File? lissero_results = merlin_magic.lissero_results
     String? lissero_version = merlin_magic.lissero_version

workflows/theiaprok/wf_theiaprok_illumina_pe.wdl

@@ -867,6 +867,8 @@ workflow theiaprok_illumina_pe {
     String? stxtyper_partial_hits = merlin_magic.stxtyper_partial_hits
     String? stxtyper_stx_frameshifts_or_internal_stop_hits =  merlin_magic.stxtyper_stx_frameshifts_or_internal_stop_hits
     String? stxtyper_novel_hits = merlin_magic.stxtyper_novel_hits
+    String? stxtyper_extended_operons = merlin_magic.stxtyper_extended_operons
+    String? stxtyper_ambiguous_hits = merlin_magic.stxtyper_ambiguous_hits
     # Shigella sonnei Typing
     File? sonneityping_mykrobe_report_csv = merlin_magic.sonneityping_mykrobe_report_csv
     File? sonneityping_mykrobe_report_json = merlin_magic.sonneityping_mykrobe_report_json

workflows/theiaprok/wf_theiaprok_illumina_se.wdl

@@ -795,6 +795,8 @@ workflow theiaprok_illumina_se {
     String? stxtyper_partial_hits = merlin_magic.stxtyper_partial_hits
     String? stxtyper_stx_frameshifts_or_internal_stop_hits =  merlin_magic.stxtyper_stx_frameshifts_or_internal_stop_hits
     String? stxtyper_novel_hits = merlin_magic.stxtyper_novel_hits
+    String? stxtyper_extended_operons = merlin_magic.stxtyper_extended_operons
+    String? stxtyper_ambiguous_hits = merlin_magic.stxtyper_ambiguous_hits
     # Shigella sonnei Typing
     File? sonneityping_mykrobe_report_csv = merlin_magic.sonneityping_mykrobe_report_csv
     File? sonneityping_mykrobe_report_json = merlin_magic.sonneityping_mykrobe_report_json

workflows/theiaprok/wf_theiaprok_ont.wdl

@@ -737,6 +737,8 @@ workflow theiaprok_ont {
     String? stxtyper_partial_hits = merlin_magic.stxtyper_partial_hits
     String? stxtyper_stx_frameshifts_or_internal_stop_hits =  merlin_magic.stxtyper_stx_frameshifts_or_internal_stop_hits
     String? stxtyper_novel_hits = merlin_magic.stxtyper_novel_hits
+    String? stxtyper_extended_operons = merlin_magic.stxtyper_extended_operons
+    String? stxtyper_ambiguous_hits = merlin_magic.stxtyper_ambiguous_hits
     # Shigella sonnei Typing
     File? sonneityping_mykrobe_report_csv = merlin_magic.sonneityping_mykrobe_report_csv
     File? sonneityping_mykrobe_report_json = merlin_magic.sonneityping_mykrobe_report_json

workflows/utilities/wf_merlin_magic.wdl

@@ -811,6 +811,8 @@ workflow merlin_magic {
     String? stxtyper_partial_hits = stxtyper.stxtyper_partial_hits
     String? stxtyper_stx_frameshifts_or_internal_stop_hits =  stxtyper.stxtyper_frameshifts_or_internal_stop_hits
     String? stxtyper_novel_hits = stxtyper.stxtyper_novel_hits
+    String? stxtyper_extended_operons = stxtyper.stxtyper_extended_operons
+    String? stxtyper_ambiguous_hits = stxtyper.stxtyper_ambiguous_hits
     # Shigella sonnei Typing
     File? sonneityping_mykrobe_report_csv = sonneityping.sonneityping_mykrobe_report_csv
     File? sonneityping_mykrobe_report_json = sonneityping.sonneityping_mykrobe_report_json