From 3667868ace92480f85b715a09195a8f768160f0b Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Thu, 24 Aug 2023 12:06:02 -0400
Subject: [PATCH 01/25] added ani_threshold Float input to animummer task. also
 exposed cpus and memory. added logic for comparing ANI_HIGHEST_PERCENTAGE to
 ani_threshold and only outputting the name of the match if the threshold is
 surpassed. tested successfully in miniwdl.

---
 tasks/quality_control/task_mummer_ani.wdl | 55 +++++++++++++++--------
 1 file changed, 37 insertions(+), 18 deletions(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 088a985b5..086c15f64 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -6,12 +6,15 @@ task animummer {
     String samplename
     File? ref_genome
     Float mash_filter = 0.9
-    String docker="us-docker.pkg.dev/general-theiagen/staphb/mummer:4.0.0-rgdv2"
+    Float ani_threshold = 92.0
+    String docker= "us-docker.pkg.dev/general-theiagen/staphb/mummer:4.0.0-rgdv2"
+    Int cpus = 4
+    Int memory = 8
     Int disk_size = 100
   }
   command <<<
     # capture and version
-    mummer --version | tee MUMMER_VERSION
+    mummer --version | tee MUMMER_VERSION.txt
 
     # set the reference genome
     # if not defined by user, then use all 43 genomes in RGDv2
@@ -42,18 +45,21 @@ task animummer {
       echo "~{samplename} did not surpass the minimum mash genetic distance filter, thus ANI was not performed"
       echo "The output TSV only contains the header line"
       # set output variables as 0s or descriptive strings
-      echo "0.0" > ANI_HIGHEST_PERCENT_BASES_ALIGNED
-      echo "0.0" > ANI_HIGHEST_PERCENT
-      echo "ANI skipped due to high genetic divergence from reference genomes" > ANI_TOP_SPECIES_MATCH
+      echo "0.0" > ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt
+      echo "0.0" > ANI_HIGHEST_PERCENT.txt
+      echo "ANI skipped due to high genetic divergence from reference genomes" > ANI_TOP_SPECIES_MATCH.txt
     # if output TSV has greater than 1 lines, then parse for appropriate outputs
     else
       ## parse out highest percentBases aligned
-      cut -f 5 ~{samplename}.ani-mummer.out.tsv | sort -nr | head -n 1 | tee ANI_HIGHEST_PERCENT_BASES_ALIGNED
-      echo "highest percent bases aligned is: $(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED)"
+      cut -f 5 ~{samplename}.ani-mummer.out.tsv | sort -nr | head -n 1 | tee ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt
+      echo "highest percent bases aligned is: $(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)"
 
       ## parse out ANI value using highest percentBases aligned value
-      grep "$(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED)" ~{samplename}.ani-mummer.out.tsv | cut -f 3 | tee ANI_HIGHEST_PERCENT
-      echo "ANI value is: $(cat ANI_HIGHEST_PERCENT)"
+      grep "$(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)" ~{samplename}.ani-mummer.out.tsv | cut -f 3 | tee ANI_HIGHEST_PERCENT.txt
+      echo "Highest ANI value is: $(cat ANI_HIGHEST_PERCENT.txt)"
+      # set ANI_HIGHEST_PERCENT as a bash variable (float)
+      ANI_HIGHEST_PERCENT=$(cat ANI_HIGHEST_PERCENT.txt)
+
 
       # have to separate out results for ani_top_species match because user-defined reference genome FASTAs will not be named as they are in RGDv2
       if [[ -z "~{ref_genome}" ]]; then
@@ -63,27 +69,40 @@ task animummer {
         # cut on periods to pull out genus_species (in future this will inlcude lineages for Listeria and other sub-species designations)
         # have to create assembly_file_basename bash variable since output TSV does not include full path to assembly file, only filename
         assembly_file_basename=$(basename ~{assembly})
-        grep "$(cat ANI_HIGHEST_PERCENT)" ~{samplename}.ani-mummer.out.tsv | cut -f 1,2 | sed "s|${assembly_file_basename}||g" | xargs | cut -d '.' -f 3 | tee ANI_TOP_SPECIES_MATCH
-        echo "ANI top species match is: $(cat ANI_TOP_SPECIES_MATCH)"
+        grep "${ANI_HIGHEST_PERCENT}" ~{samplename}.ani-mummer.out.tsv | cut -f 1,2 | sed "s|${assembly_file_basename}||g" | xargs | cut -d '.' -f 3 | tee ANI_TOP_SPECIES_MATCH.txt
+        echo "ANI top species match is: $(cat ANI_TOP_SPECIES_MATCH.txt)"
+
+        # if ANI threshold is defined by user, compare to highest ANI value and only output if threshold is surpassed
+        if [[ -n "~{ani_threshold}" ]]; then
+          echo "Comparing user-defined ANI threshold to highest ANI value..."
+          # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
+          if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )"; then
+            echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
+            echo "The ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
+          else
+            echo "The highest ANI value $ANI_HIGHEST_PERCENT is greater than the user-defined threshold ~{ani_threshold}"
+          fi
+        fi
       else 
         # User specified a reference genome, use fasta filename as output string
-        basename "${REF_GENOME}" > ANI_TOP_SPECIES_MATCH
+        basename "${REF_GENOME}" > ANI_TOP_SPECIES_MATCH.txt
         echo "Reference genome used for ANI is: ${REF_GENOME}"
       fi
     fi
     
   >>>
   output {
-    Float ani_highest_percent = read_float("ANI_HIGHEST_PERCENT")
-    Float ani_highest_percent_bases_aligned = read_float("ANI_HIGHEST_PERCENT_BASES_ALIGNED")
+    Float ani_highest_percent = read_float("ANI_HIGHEST_PERCENT.txt")
+    Float ani_highest_percent_bases_aligned = read_float("ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt")
     File ani_output_tsv = "~{samplename}.ani-mummer.out.tsv"
-    String ani_top_species_match = read_string("ANI_TOP_SPECIES_MATCH")
-    String ani_mummer_version = read_string("MUMMER_VERSION")
+    String ani_top_species_match = read_string("ANI_TOP_SPECIES_MATCH.txt")
+    String ani_mummer_version = read_string("MUMMER_VERSION.txt")
+    String ani_docker = "~{docker}"
   }
   runtime {
     docker:  "~{docker}"
-    memory:  "8 GB"
-    cpu:   4
+    memory:  "~{memory} GB"
+    cpu:  cpus
     disks: "local-disk " + disk_size + " SSD"
     disk: disk_size + " GB"
     maxRetries: 3

From 697c28b619e53cf8a47dd8122f61738f5cda7970 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Thu, 24 Aug 2023 12:22:30 -0400
Subject: [PATCH 02/25] fixed awk syntax

---
 tasks/quality_control/task_mummer_ani.wdl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 086c15f64..3f2dedaf0 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -76,7 +76,7 @@ task animummer {
         if [[ -n "~{ani_threshold}" ]]; then
           echo "Comparing user-defined ANI threshold to highest ANI value..."
           # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
-          if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )"; then
+          if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )}"; then
             echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
             echo "The ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
           else

From 07a60a0e734674a20c452908246c993b8c9d830d Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Thu, 24 Aug 2023 15:17:56 -0400
Subject: [PATCH 03/25] added ani_docker to export_taxon_tables task and
 illumina pe workflow

---
 tasks/utilities/task_broad_terra_tools.wdl       | 2 ++
 workflows/theiaprok/wf_theiaprok_illumina_pe.wdl | 4 +++-
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/tasks/utilities/task_broad_terra_tools.wdl b/tasks/utilities/task_broad_terra_tools.wdl
index 708d7bd92..491f5f18c 100644
--- a/tasks/utilities/task_broad_terra_tools.wdl
+++ b/tasks/utilities/task_broad_terra_tools.wdl
@@ -90,6 +90,7 @@ task export_taxon_tables {
     File? ani_output_tsv
     String? ani_top_species_match 
     String? ani_mummer_version
+    String? ani_docker
     File? amrfinderplus_all_report
     File? amrfinderplus_amr_report
     File? amrfinderplus_stress_report
@@ -583,6 +584,7 @@ task export_taxon_tables {
       "ani_output_tsv": "~{ani_output_tsv}",
       "ani_top_species_match": "~{ani_top_species_match}",
       "ani_mummer_version": "~{ani_mummer_version}",
+      "ani_docker": "~{ani_docker}",
       "resfinder_pheno_table": "~{resfinder_pheno_table}",
       "resfinder_pheno_table_species": "~{resfinder_pheno_table_species}",
       "resfinder_seqs": "~{resfinder_seqs}",
diff --git a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
index e960a7eb1..a74232c9f 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -211,7 +211,7 @@ workflow theiaprok_illumina_pe {
             quast_gc_percent = quast.gc_percent,
             busco_results = busco.busco_results,
             ani_highest_percent = ani.ani_highest_percent,
-            ani_highest_percent_bases_aligned = ani.ani_highest_percent_bases_aligned
+            ani_highest_percent_bases_aligned = ani.ani_highest_percent_bases_aligned,
         }
       }
       call merlin_magic_workflow.merlin_magic {
@@ -291,6 +291,7 @@ workflow theiaprok_illumina_pe {
             ani_output_tsv = ani.ani_output_tsv,
             ani_top_species_match = ani.ani_top_species_match,
             ani_mummer_version = ani.ani_mummer_version,
+            ani_docker = ani.ani_docker,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -605,6 +606,7 @@ workflow theiaprok_illumina_pe {
     File? ani_output_tsv = ani.ani_output_tsv
     String? ani_top_species_match = ani.ani_top_species_match
     String? ani_mummer_version = ani.ani_mummer_version
+    String? ani_mummer_docker = ani.ani_docker
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report

From 20cb4e389b8b8c6f2f8b25fcedc8607ee3e6e055 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Thu, 24 Aug 2023 15:32:40 -0400
Subject: [PATCH 04/25] remove "The"

---
 tasks/quality_control/task_mummer_ani.wdl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 3f2dedaf0..3b9a81295 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -78,7 +78,7 @@ task animummer {
           # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
           if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )}"; then
             echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
-            echo "The ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
+            echo "ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
           else
             echo "The highest ANI value $ANI_HIGHEST_PERCENT is greater than the user-defined threshold ~{ani_threshold}"
           fi

From d7de47e7a3e52912df6420835ff947413cc00223 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 25 Aug 2023 09:40:45 -0400
Subject: [PATCH 05/25] added ani_docker output to theiaprok_fasta wf

---
 workflows/theiaprok/wf_theiaprok_fasta.wdl | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/workflows/theiaprok/wf_theiaprok_fasta.wdl b/workflows/theiaprok/wf_theiaprok_fasta.wdl
index cf4f7df7b..7e776a825 100644
--- a/workflows/theiaprok/wf_theiaprok_fasta.wdl
+++ b/workflows/theiaprok/wf_theiaprok_fasta.wdl
@@ -166,6 +166,7 @@ workflow theiaprok_fasta {
         ani_output_tsv = ani.ani_output_tsv,
         ani_top_species_match = ani.ani_top_species_match,
         ani_mummer_version = ani.ani_mummer_version,
+        ani_docker = ani.ani_docker,
         amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
         amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
         amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -411,6 +412,7 @@ workflow theiaprok_fasta {
     File? ani_output_tsv = ani.ani_output_tsv
     String? ani_top_species_match = ani.ani_top_species_match
     String? ani_mummer_version = ani.ani_mummer_version
+    String? ani_mummer_docker = ani.ani_docker
     # NCBI-AMRFinderPlus Outputs
     File amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report
     File amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report

From ff175641c3ea7e113a7bd2e53dcf1e4e08bd87b3 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 25 Aug 2023 09:41:24 -0400
Subject: [PATCH 06/25] removed comma that I accidentally added to call block
 of qc_check_table in theiaprok illumina pe wf

---
 workflows/theiaprok/wf_theiaprok_illumina_pe.wdl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
index a74232c9f..c6af16285 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -211,7 +211,7 @@ workflow theiaprok_illumina_pe {
             quast_gc_percent = quast.gc_percent,
             busco_results = busco.busco_results,
             ani_highest_percent = ani.ani_highest_percent,
-            ani_highest_percent_bases_aligned = ani.ani_highest_percent_bases_aligned,
+            ani_highest_percent_bases_aligned = ani.ani_highest_percent_bases_aligned
         }
       }
       call merlin_magic_workflow.merlin_magic {

From 7e49db0ca523b66d1e0f1e7aab3b266515378ead Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 25 Aug 2023 09:57:12 -0400
Subject: [PATCH 07/25] added ani_docker output to theiaprok illumina se and
 ONT workflows

---
 workflows/theiaprok/wf_theiaprok_illumina_se.wdl | 2 ++
 workflows/theiaprok/wf_theiaprok_ont.wdl         | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
index 1d2c80fe5..7f23087ff 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
@@ -261,6 +261,7 @@ workflow theiaprok_illumina_se {
             ani_output_tsv = ani.ani_output_tsv,
             ani_top_species_match = ani.ani_top_species_match,
             ani_mummer_version = ani.ani_mummer_version,
+            ani_docker = ani.ani_docker,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -554,6 +555,7 @@ workflow theiaprok_illumina_se {
     File? ani_output_tsv = ani.ani_output_tsv
     String? ani_top_species_match = ani.ani_top_species_match
     String? ani_mummer_version = ani.ani_mummer_version
+    String? ani_mummer_docker = ani.ani_docker
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report
diff --git a/workflows/theiaprok/wf_theiaprok_ont.wdl b/workflows/theiaprok/wf_theiaprok_ont.wdl
index ded5d1807..5af688b51 100644
--- a/workflows/theiaprok/wf_theiaprok_ont.wdl
+++ b/workflows/theiaprok/wf_theiaprok_ont.wdl
@@ -258,6 +258,7 @@ workflow theiaprok_ont {
             ani_output_tsv = ani.ani_output_tsv,
             ani_top_species_match = ani.ani_top_species_match,
             ani_mummer_version = ani.ani_mummer_version,
+            ani_docker = ani.ani_docker,
             amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus.amrfinderplus_stress_report,
@@ -523,6 +524,7 @@ workflow theiaprok_ont {
     File? ani_output_tsv = ani.ani_output_tsv
     String? ani_top_species_match = ani.ani_top_species_match
     String? ani_mummer_version = ani.ani_mummer_version
+    String? ani_mummer_docker = ani.ani_docker
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report

From 5dc45e4fbc514f6ca983fafaeedf2d851be65851 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 25 Aug 2023 11:10:57 -0400
Subject: [PATCH 08/25] update CI

---
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml | 6 +++---
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml | 6 +++---
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
index 3ba26335f..1a763b493 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -574,7 +574,7 @@
     - path: miniwdl_run/wdl/tasks/quality_control/task_fastq_scan.wdl
       md5sum: 65bd7b4ef51b04df67aaf9276ff0918a
     - path: miniwdl_run/wdl/tasks/quality_control/task_mummer_ani.wdl
-      md5sum: 21e8b1cbe4c8be26d1ac8b8013970166
+      md5sum: 33a111c260b2fb0b6db6d9ed77ec8e87
     - path: miniwdl_run/wdl/tasks/quality_control/task_ncbi_scrub.wdl
       md5sum: f41beadaf9d24ae77b05ac3b962127a8
     - path: miniwdl_run/wdl/tasks/quality_control/task_qc_check_phb.wdl
@@ -632,9 +632,9 @@
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl
       md5sum: 024971d1439dff7d59c0a26a824bd2c6
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
-      md5sum: 499eb797a809306edf7e97005f67f8f2
+      md5sum: 227ee827aca5f05b23eaa608119ddc9a
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
-      md5sum: a76d59109075ce8b861e63ffe70d7c77
+      md5sum: 084648227cdc63e06e10ec27e728bdc6
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: 53555c2f3e144e55f362080c5e75e434
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_pe.wdl
diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
index 6d2a6ccbb..a2cf78fc3 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -544,7 +544,7 @@
     - path: miniwdl_run/wdl/tasks/quality_control/task_fastq_scan.wdl
       md5sum: 65bd7b4ef51b04df67aaf9276ff0918a
     - path: miniwdl_run/wdl/tasks/quality_control/task_mummer_ani.wdl
-      md5sum: 21e8b1cbe4c8be26d1ac8b8013970166
+      md5sum: 33a111c260b2fb0b6db6d9ed77ec8e87
     - path: miniwdl_run/wdl/tasks/quality_control/task_qc_check_phb.wdl
       contains: ["version", "qc_check_table", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_quast.wdl
@@ -600,9 +600,9 @@
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl
       md5sum: 024971d1439dff7d59c0a26a824bd2c6
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
-      md5sum: 499eb797a809306edf7e97005f67f8f2
+      md5sum: 227ee827aca5f05b23eaa608119ddc9a
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
-      md5sum: c6ab2e13cf4802664b1c93e40a1cd5ba
+      md5sum: 6c6ccc821ae81ef22529f10398c60a06
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: 53555c2f3e144e55f362080c5e75e434
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl

From 3cc9766bcaf246c64a798d55f7dd16ca372d980d Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 1 Sep 2023 15:43:10 -0400
Subject: [PATCH 09/25] update CI

---
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml | 2 +-
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
index 582cfc2c5..4b15211a2 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -632,7 +632,7 @@
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl
       md5sum: 024971d1439dff7d59c0a26a824bd2c6
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
-      md5sum: 227ee827aca5f05b23eaa608119ddc9a
+      md5sum: 1d7365ed0f29a7632568a37bfce0f1b6
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
       md5sum: 084648227cdc63e06e10ec27e728bdc6
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
index f1b5c6104..6ca48b0ed 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -600,9 +600,9 @@
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl
       md5sum: 024971d1439dff7d59c0a26a824bd2c6
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
-      md5sum: 227ee827aca5f05b23eaa608119ddc9a
+      md5sum: 1d7365ed0f29a7632568a37bfce0f1b6
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
-      md5sum: 6c6ccc821ae81ef22529f10398c60a06
+      md5sum: d144cb871c175029a34107a369e70c5d
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: 53555c2f3e144e55f362080c5e75e434
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl

From a5566753be57cb7915a69adde607f098d8b96f38 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 6 Oct 2023 16:56:03 -0400
Subject: [PATCH 10/25] mummer_ani task: added percent_base_aligned_threshold
 with 70 as default. default ani_threshold is now 85. Also added logic to only
 output ani_top_species_match if both thresholds are surpassed.

---
 tasks/quality_control/task_mummer_ani.wdl | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 3b9a81295..2c8d87d11 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -6,7 +6,9 @@ task animummer {
     String samplename
     File? ref_genome
     Float mash_filter = 0.9
-    Float ani_threshold = 92.0
+    # these 2 thresholds were set as they are used by CDC enterics lab/PulseNet for ANI thresholds
+    Float ani_threshold = 85.0
+    Float percent_bases_aligned_threshold = 70.0
     String docker= "us-docker.pkg.dev/general-theiagen/staphb/mummer:4.0.0-rgdv2"
     Int cpus = 4
     Int memory = 8
@@ -53,6 +55,7 @@ task animummer {
       ## parse out highest percentBases aligned
       cut -f 5 ~{samplename}.ani-mummer.out.tsv | sort -nr | head -n 1 | tee ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt
       echo "highest percent bases aligned is: $(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)"
+      ANI_HIGHEST_PERCENT_BASES_ALIGNED=$(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)
 
       ## parse out ANI value using highest percentBases aligned value
       grep "$(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)" ~{samplename}.ani-mummer.out.tsv | cut -f 3 | tee ANI_HIGHEST_PERCENT.txt
@@ -72,15 +75,21 @@ task animummer {
         grep "${ANI_HIGHEST_PERCENT}" ~{samplename}.ani-mummer.out.tsv | cut -f 1,2 | sed "s|${assembly_file_basename}||g" | xargs | cut -d '.' -f 3 | tee ANI_TOP_SPECIES_MATCH.txt
         echo "ANI top species match is: $(cat ANI_TOP_SPECIES_MATCH.txt)"
 
-        # if ANI threshold is defined by user, compare to highest ANI value and only output if threshold is surpassed
-        if [[ -n "~{ani_threshold}" ]]; then
+        # if ANI threshold or percent_bases_aligned_threshold is defined by user (they both are by default), compare to highest ANI value and corresponding percent_bases_aligned value and only output ANI_top_species_match if both thresholds are surpassed
+        if [[ -n "~{ani_threshold}" || -n "~{percent_bases_aligned_threshold}" ]]; then
           echo "Comparing user-defined ANI threshold to highest ANI value..."
           # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
           if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )}"; then
             echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
             echo "ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
+          # else if: compare percent_bases_aligned_threshold to ANI_HIGHEST_PERCENT_BASES_ALIGNED using awk
+          elif ! awk "BEGIN{ exit (${ANI_HIGHEST_PERCENT_BASES_ALIGNED} < ~{percent_bases_aligned_threshold} )}"; then
+            echo "The highest ANI percent bases aligned value ${ANI_HIGHEST_PERCENT_BASES_ALIGNED} is less than the user-defined threshold of ~{percent_bases_aligned_threshold}"
+            # overwrite ANI_TOP_SPECIES_MATCH.txt when percent_bases_aligned threshold is not surpassed
+            echo "ANI percent bases aligned did not surpass the user-defined threshold of ~{percent_bases_aligned_threshold}" > ANI_TOP_SPECIES_MATCH.txt 
           else
-            echo "The highest ANI value $ANI_HIGHEST_PERCENT is greater than the user-defined threshold ~{ani_threshold}"
+            echo "The highest ANI value ${ANI_HIGHEST_PERCENT} is greater than the user-defined threshold ~{ani_threshold}"
+            echo "The highest percent bases aligned value ${ANI_HIGHEST_PERCENT_BASES_ALIGNED} is greater than the user-defined threshold ~{percent_bases_aligned_threshold}"
           fi
         fi
       else 

From aea02358c350e65f50ffaaaa406608df9954e836 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 6 Oct 2023 17:52:48 -0400
Subject: [PATCH 11/25] clarified messages when 2 thresholds are not passed

---
 tasks/quality_control/task_mummer_ani.wdl | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 2c8d87d11..630ff9d10 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -80,13 +80,13 @@ task animummer {
           echo "Comparing user-defined ANI threshold to highest ANI value..."
           # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
           if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )}"; then
-            echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
-            echo "ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
+            echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined ANI threshold of ~{ani_threshold}"
+            echo "ANI top species match did not surpass the user-defined ANI threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
           # else if: compare percent_bases_aligned_threshold to ANI_HIGHEST_PERCENT_BASES_ALIGNED using awk
           elif ! awk "BEGIN{ exit (${ANI_HIGHEST_PERCENT_BASES_ALIGNED} < ~{percent_bases_aligned_threshold} )}"; then
             echo "The highest ANI percent bases aligned value ${ANI_HIGHEST_PERCENT_BASES_ALIGNED} is less than the user-defined threshold of ~{percent_bases_aligned_threshold}"
             # overwrite ANI_TOP_SPECIES_MATCH.txt when percent_bases_aligned threshold is not surpassed
-            echo "ANI percent bases aligned did not surpass the user-defined threshold of ~{percent_bases_aligned_threshold}" > ANI_TOP_SPECIES_MATCH.txt 
+            echo "ANI top species match did not surpass the user-defined percent bases aligned threshold of ~{percent_bases_aligned_threshold}" > ANI_TOP_SPECIES_MATCH.txt 
           else
             echo "The highest ANI value ${ANI_HIGHEST_PERCENT} is greater than the user-defined threshold ~{ani_threshold}"
             echo "The highest percent bases aligned value ${ANI_HIGHEST_PERCENT_BASES_ALIGNED} is greater than the user-defined threshold ~{percent_bases_aligned_threshold}"

From 818065a610319892f3a320dc91b0e9409c360abf Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Thu, 24 Aug 2023 12:06:02 -0400
Subject: [PATCH 12/25] added ani_threshold Float input to animummer task. also
 exposed cpus and memory. added logic for comparing ANI_HIGHEST_PERCENTAGE to
 ani_threshold and only outputting the name of the match if the threshold is
 surpassed. tested successfully in miniwdl.

---
 tasks/quality_control/task_mummer_ani.wdl | 55 +++++++++++++++--------
 1 file changed, 37 insertions(+), 18 deletions(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 088a985b5..086c15f64 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -6,12 +6,15 @@ task animummer {
     String samplename
     File? ref_genome
     Float mash_filter = 0.9
-    String docker="us-docker.pkg.dev/general-theiagen/staphb/mummer:4.0.0-rgdv2"
+    Float ani_threshold = 92.0
+    String docker= "us-docker.pkg.dev/general-theiagen/staphb/mummer:4.0.0-rgdv2"
+    Int cpus = 4
+    Int memory = 8
     Int disk_size = 100
   }
   command <<<
     # capture and version
-    mummer --version | tee MUMMER_VERSION
+    mummer --version | tee MUMMER_VERSION.txt
 
     # set the reference genome
     # if not defined by user, then use all 43 genomes in RGDv2
@@ -42,18 +45,21 @@ task animummer {
       echo "~{samplename} did not surpass the minimum mash genetic distance filter, thus ANI was not performed"
       echo "The output TSV only contains the header line"
       # set output variables as 0s or descriptive strings
-      echo "0.0" > ANI_HIGHEST_PERCENT_BASES_ALIGNED
-      echo "0.0" > ANI_HIGHEST_PERCENT
-      echo "ANI skipped due to high genetic divergence from reference genomes" > ANI_TOP_SPECIES_MATCH
+      echo "0.0" > ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt
+      echo "0.0" > ANI_HIGHEST_PERCENT.txt
+      echo "ANI skipped due to high genetic divergence from reference genomes" > ANI_TOP_SPECIES_MATCH.txt
     # if output TSV has greater than 1 lines, then parse for appropriate outputs
     else
       ## parse out highest percentBases aligned
-      cut -f 5 ~{samplename}.ani-mummer.out.tsv | sort -nr | head -n 1 | tee ANI_HIGHEST_PERCENT_BASES_ALIGNED
-      echo "highest percent bases aligned is: $(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED)"
+      cut -f 5 ~{samplename}.ani-mummer.out.tsv | sort -nr | head -n 1 | tee ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt
+      echo "highest percent bases aligned is: $(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)"
 
       ## parse out ANI value using highest percentBases aligned value
-      grep "$(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED)" ~{samplename}.ani-mummer.out.tsv | cut -f 3 | tee ANI_HIGHEST_PERCENT
-      echo "ANI value is: $(cat ANI_HIGHEST_PERCENT)"
+      grep "$(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)" ~{samplename}.ani-mummer.out.tsv | cut -f 3 | tee ANI_HIGHEST_PERCENT.txt
+      echo "Highest ANI value is: $(cat ANI_HIGHEST_PERCENT.txt)"
+      # set ANI_HIGHEST_PERCENT as a bash variable (float)
+      ANI_HIGHEST_PERCENT=$(cat ANI_HIGHEST_PERCENT.txt)
+
 
       # have to separate out results for ani_top_species match because user-defined reference genome FASTAs will not be named as they are in RGDv2
       if [[ -z "~{ref_genome}" ]]; then
@@ -63,27 +69,40 @@ task animummer {
         # cut on periods to pull out genus_species (in future this will inlcude lineages for Listeria and other sub-species designations)
         # have to create assembly_file_basename bash variable since output TSV does not include full path to assembly file, only filename
         assembly_file_basename=$(basename ~{assembly})
-        grep "$(cat ANI_HIGHEST_PERCENT)" ~{samplename}.ani-mummer.out.tsv | cut -f 1,2 | sed "s|${assembly_file_basename}||g" | xargs | cut -d '.' -f 3 | tee ANI_TOP_SPECIES_MATCH
-        echo "ANI top species match is: $(cat ANI_TOP_SPECIES_MATCH)"
+        grep "${ANI_HIGHEST_PERCENT}" ~{samplename}.ani-mummer.out.tsv | cut -f 1,2 | sed "s|${assembly_file_basename}||g" | xargs | cut -d '.' -f 3 | tee ANI_TOP_SPECIES_MATCH.txt
+        echo "ANI top species match is: $(cat ANI_TOP_SPECIES_MATCH.txt)"
+
+        # if ANI threshold is defined by user, compare to highest ANI value and only output if threshold is surpassed
+        if [[ -n "~{ani_threshold}" ]]; then
+          echo "Comparing user-defined ANI threshold to highest ANI value..."
+          # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
+          if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )"; then
+            echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
+            echo "The ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
+          else
+            echo "The highest ANI value $ANI_HIGHEST_PERCENT is greater than the user-defined threshold ~{ani_threshold}"
+          fi
+        fi
       else 
         # User specified a reference genome, use fasta filename as output string
-        basename "${REF_GENOME}" > ANI_TOP_SPECIES_MATCH
+        basename "${REF_GENOME}" > ANI_TOP_SPECIES_MATCH.txt
         echo "Reference genome used for ANI is: ${REF_GENOME}"
       fi
     fi
     
   >>>
   output {
-    Float ani_highest_percent = read_float("ANI_HIGHEST_PERCENT")
-    Float ani_highest_percent_bases_aligned = read_float("ANI_HIGHEST_PERCENT_BASES_ALIGNED")
+    Float ani_highest_percent = read_float("ANI_HIGHEST_PERCENT.txt")
+    Float ani_highest_percent_bases_aligned = read_float("ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt")
     File ani_output_tsv = "~{samplename}.ani-mummer.out.tsv"
-    String ani_top_species_match = read_string("ANI_TOP_SPECIES_MATCH")
-    String ani_mummer_version = read_string("MUMMER_VERSION")
+    String ani_top_species_match = read_string("ANI_TOP_SPECIES_MATCH.txt")
+    String ani_mummer_version = read_string("MUMMER_VERSION.txt")
+    String ani_docker = "~{docker}"
   }
   runtime {
     docker:  "~{docker}"
-    memory:  "8 GB"
-    cpu:   4
+    memory:  "~{memory} GB"
+    cpu:  cpus
     disks: "local-disk " + disk_size + " SSD"
     disk: disk_size + " GB"
     maxRetries: 3

From 3c49ebc0b80592ef86c8ec6146147359ef07a9df Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Thu, 24 Aug 2023 12:22:30 -0400
Subject: [PATCH 13/25] fixed awk syntax

---
 tasks/quality_control/task_mummer_ani.wdl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 086c15f64..3f2dedaf0 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -76,7 +76,7 @@ task animummer {
         if [[ -n "~{ani_threshold}" ]]; then
           echo "Comparing user-defined ANI threshold to highest ANI value..."
           # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
-          if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )"; then
+          if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )}"; then
             echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
             echo "The ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
           else

From dbbb92c744dfc5d7ca2c84989286c12bcb7b7870 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Thu, 24 Aug 2023 15:17:56 -0400
Subject: [PATCH 14/25] added ani_docker to export_taxon_tables task and
 illumina pe workflow

---
 tasks/utilities/task_broad_terra_tools.wdl       | 2 ++
 workflows/theiaprok/wf_theiaprok_illumina_pe.wdl | 4 +++-
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/tasks/utilities/task_broad_terra_tools.wdl b/tasks/utilities/task_broad_terra_tools.wdl
index 2ae737a5c..eb819b302 100644
--- a/tasks/utilities/task_broad_terra_tools.wdl
+++ b/tasks/utilities/task_broad_terra_tools.wdl
@@ -92,6 +92,7 @@ task export_taxon_tables {
     File? ani_output_tsv
     String? ani_top_species_match 
     String? ani_mummer_version
+    String? ani_docker
     String? kmerfinder_docker
     File? kmerfinder_results_tsv
     String? kmerfinder_top_hit
@@ -595,6 +596,7 @@ task export_taxon_tables {
       "ani_output_tsv": "~{ani_output_tsv}",
       "ani_top_species_match": "~{ani_top_species_match}",
       "ani_mummer_version": "~{ani_mummer_version}",
+      "ani_docker": "~{ani_docker}",
       "kmerfinder_docker": "~{kmerfinder_docker}",
       "kmerfinder_results_tsv": "~{kmerfinder_results_tsv}",
       "kmerfinder_top_hit": "~{kmerfinder_top_hit}",
diff --git a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
index feae677c0..af6f9389e 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -220,7 +220,7 @@ workflow theiaprok_illumina_pe {
             quast_gc_percent = quast.gc_percent,
             busco_results = busco.busco_results,
             ani_highest_percent = ani.ani_highest_percent,
-            ani_highest_percent_bases_aligned = ani.ani_highest_percent_bases_aligned
+            ani_highest_percent_bases_aligned = ani.ani_highest_percent_bases_aligned,
         }
       }
       call merlin_magic_workflow.merlin_magic {
@@ -300,6 +300,7 @@ workflow theiaprok_illumina_pe {
             ani_output_tsv = ani.ani_output_tsv,
             ani_top_species_match = ani.ani_top_species_match,
             ani_mummer_version = ani.ani_mummer_version,
+            ani_docker = ani.ani_docker,
             kmerfinder_docker = kmerfinder.kmerfinder_docker,
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
@@ -621,6 +622,7 @@ workflow theiaprok_illumina_pe {
     File? ani_output_tsv = ani.ani_output_tsv
     String? ani_top_species_match = ani.ani_top_species_match
     String? ani_mummer_version = ani.ani_mummer_version
+    String? ani_mummer_docker = ani.ani_docker
     # kmerfinder outputs
     String? kmerfinder_docker = kmerfinder.kmerfinder_docker
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv

From 55dfc2fd09b8cfd3cc2593b3b996299855aab96c Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Thu, 24 Aug 2023 15:32:40 -0400
Subject: [PATCH 15/25] remove "The"

---
 tasks/quality_control/task_mummer_ani.wdl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 3f2dedaf0..3b9a81295 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -78,7 +78,7 @@ task animummer {
           # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
           if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )}"; then
             echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
-            echo "The ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
+            echo "ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
           else
             echo "The highest ANI value $ANI_HIGHEST_PERCENT is greater than the user-defined threshold ~{ani_threshold}"
           fi

From c62449e9cc95c69b05674bc072519f69fa5c799c Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 25 Aug 2023 09:40:45 -0400
Subject: [PATCH 16/25] added ani_docker output to theiaprok_fasta wf

---
 workflows/theiaprok/wf_theiaprok_fasta.wdl | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/workflows/theiaprok/wf_theiaprok_fasta.wdl b/workflows/theiaprok/wf_theiaprok_fasta.wdl
index 2c4ce9f15..c5fe43576 100644
--- a/workflows/theiaprok/wf_theiaprok_fasta.wdl
+++ b/workflows/theiaprok/wf_theiaprok_fasta.wdl
@@ -175,6 +175,7 @@ workflow theiaprok_fasta {
         ani_output_tsv = ani.ani_output_tsv,
         ani_top_species_match = ani.ani_top_species_match,
         ani_mummer_version = ani.ani_mummer_version,
+        ani_docker = ani.ani_docker,
         kmerfinder_docker = kmerfinder.kmerfinder_docker,
         kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
         kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
@@ -427,6 +428,7 @@ workflow theiaprok_fasta {
     File? ani_output_tsv = ani.ani_output_tsv
     String? ani_top_species_match = ani.ani_top_species_match
     String? ani_mummer_version = ani.ani_mummer_version
+    String? ani_mummer_docker = ani.ani_docker
     # kmerfinder outputs
     String? kmerfinder_docker = kmerfinder.kmerfinder_docker
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv

From c9340fef93207ca78b2fd1cf0219c3f70df4e81b Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 25 Aug 2023 09:41:24 -0400
Subject: [PATCH 17/25] removed comma that I accidentally added to call block
 of qc_check_table in theiaprok illumina pe wf

---
 workflows/theiaprok/wf_theiaprok_illumina_pe.wdl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
index af6f9389e..4ffbf5ee9 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -220,7 +220,7 @@ workflow theiaprok_illumina_pe {
             quast_gc_percent = quast.gc_percent,
             busco_results = busco.busco_results,
             ani_highest_percent = ani.ani_highest_percent,
-            ani_highest_percent_bases_aligned = ani.ani_highest_percent_bases_aligned,
+            ani_highest_percent_bases_aligned = ani.ani_highest_percent_bases_aligned
         }
       }
       call merlin_magic_workflow.merlin_magic {

From cb94465b1b3bf9eee92f46929eaf4cb24b34ad37 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 25 Aug 2023 09:57:12 -0400
Subject: [PATCH 18/25] added ani_docker output to theiaprok illumina se and
 ONT workflows

---
 workflows/theiaprok/wf_theiaprok_illumina_se.wdl | 2 ++
 workflows/theiaprok/wf_theiaprok_ont.wdl         | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
index 8a269be90..365b1ed2f 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
@@ -273,6 +273,7 @@ workflow theiaprok_illumina_se {
             ani_output_tsv = ani.ani_output_tsv,
             ani_top_species_match = ani.ani_top_species_match,
             ani_mummer_version = ani.ani_mummer_version,
+            ani_docker = ani.ani_docker,
             kmerfinder_docker = kmerfinder.kmerfinder_docker,
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
@@ -573,6 +574,7 @@ workflow theiaprok_illumina_se {
     File? ani_output_tsv = ani.ani_output_tsv
     String? ani_top_species_match = ani.ani_top_species_match
     String? ani_mummer_version = ani.ani_mummer_version
+    String? ani_mummer_docker = ani.ani_docker
     # kmerfinder outputs
     String? kmerfinder_docker = kmerfinder.kmerfinder_docker
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
diff --git a/workflows/theiaprok/wf_theiaprok_ont.wdl b/workflows/theiaprok/wf_theiaprok_ont.wdl
index 5254f8640..593f19723 100644
--- a/workflows/theiaprok/wf_theiaprok_ont.wdl
+++ b/workflows/theiaprok/wf_theiaprok_ont.wdl
@@ -271,6 +271,7 @@ workflow theiaprok_ont {
             ani_output_tsv = ani.ani_output_tsv,
             ani_top_species_match = ani.ani_top_species_match,
             ani_mummer_version = ani.ani_mummer_version,
+            ani_docker = ani.ani_docker,
             kmerfinder_docker = kmerfinder.kmerfinder_docker,
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
@@ -547,6 +548,7 @@ workflow theiaprok_ont {
     File? ani_output_tsv = ani.ani_output_tsv
     String? ani_top_species_match = ani.ani_top_species_match
     String? ani_mummer_version = ani.ani_mummer_version
+    String? ani_mummer_docker = ani.ani_docker
     # kmerfinder outputs
     String? kmerfinder_docker = kmerfinder.kmerfinder_docker
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv

From f08ee35133cac13c423561c4923b239b7cf7e887 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 25 Aug 2023 11:10:57 -0400
Subject: [PATCH 19/25] update CI

---
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml | 2 +-
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
index ee07b678f..9e4362064 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -574,7 +574,7 @@
     - path: miniwdl_run/wdl/tasks/quality_control/task_fastq_scan.wdl
       contains: ["version", "fastq_scan", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_mummer_ani.wdl
-      md5sum: 21e8b1cbe4c8be26d1ac8b8013970166
+      md5sum: 33a111c260b2fb0b6db6d9ed77ec8e87
     - path: miniwdl_run/wdl/tasks/quality_control/task_ncbi_scrub.wdl
       contains: ["version", "scrub", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_qc_check_phb.wdl
diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
index 5d4873a19..b46305071 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -544,7 +544,7 @@
     - path: miniwdl_run/wdl/tasks/quality_control/task_fastq_scan.wdl
       contains: ["version", "fastq_scan", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_mummer_ani.wdl
-      md5sum: 21e8b1cbe4c8be26d1ac8b8013970166
+      md5sum: 33a111c260b2fb0b6db6d9ed77ec8e87
     - path: miniwdl_run/wdl/tasks/quality_control/task_qc_check_phb.wdl
       contains: ["version", "qc_check_table", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_quast.wdl

From be3a734533c240e56390708914722240dc9c5935 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 6 Oct 2023 16:56:03 -0400
Subject: [PATCH 20/25] mummer_ani task: added percent_base_aligned_threshold
 with 70 as default. default ani_threshold is now 85. Also added logic to only
 output ani_top_species_match if both thresholds are surpassed.

---
 tasks/quality_control/task_mummer_ani.wdl | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 3b9a81295..2c8d87d11 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -6,7 +6,9 @@ task animummer {
     String samplename
     File? ref_genome
     Float mash_filter = 0.9
-    Float ani_threshold = 92.0
+    # these 2 thresholds were set as they are used by CDC enterics lab/PulseNet for ANI thresholds
+    Float ani_threshold = 85.0
+    Float percent_bases_aligned_threshold = 70.0
     String docker= "us-docker.pkg.dev/general-theiagen/staphb/mummer:4.0.0-rgdv2"
     Int cpus = 4
     Int memory = 8
@@ -53,6 +55,7 @@ task animummer {
       ## parse out highest percentBases aligned
       cut -f 5 ~{samplename}.ani-mummer.out.tsv | sort -nr | head -n 1 | tee ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt
       echo "highest percent bases aligned is: $(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)"
+      ANI_HIGHEST_PERCENT_BASES_ALIGNED=$(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)
 
       ## parse out ANI value using highest percentBases aligned value
       grep "$(cat ANI_HIGHEST_PERCENT_BASES_ALIGNED.txt)" ~{samplename}.ani-mummer.out.tsv | cut -f 3 | tee ANI_HIGHEST_PERCENT.txt
@@ -72,15 +75,21 @@ task animummer {
         grep "${ANI_HIGHEST_PERCENT}" ~{samplename}.ani-mummer.out.tsv | cut -f 1,2 | sed "s|${assembly_file_basename}||g" | xargs | cut -d '.' -f 3 | tee ANI_TOP_SPECIES_MATCH.txt
         echo "ANI top species match is: $(cat ANI_TOP_SPECIES_MATCH.txt)"
 
-        # if ANI threshold is defined by user, compare to highest ANI value and only output if threshold is surpassed
-        if [[ -n "~{ani_threshold}" ]]; then
+        # if ANI threshold or percent_bases_aligned_threshold is defined by user (they both are by default), compare to highest ANI value and corresponding percent_bases_aligned value and only output ANI_top_species_match if both thresholds are surpassed
+        if [[ -n "~{ani_threshold}" || -n "~{percent_bases_aligned_threshold}" ]]; then
           echo "Comparing user-defined ANI threshold to highest ANI value..."
           # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
           if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )}"; then
             echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
             echo "ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
+          # else if: compare percent_bases_aligned_threshold to ANI_HIGHEST_PERCENT_BASES_ALIGNED using awk
+          elif ! awk "BEGIN{ exit (${ANI_HIGHEST_PERCENT_BASES_ALIGNED} < ~{percent_bases_aligned_threshold} )}"; then
+            echo "The highest ANI percent bases aligned value ${ANI_HIGHEST_PERCENT_BASES_ALIGNED} is less than the user-defined threshold of ~{percent_bases_aligned_threshold}"
+            # overwrite ANI_TOP_SPECIES_MATCH.txt when percent_bases_aligned threshold is not surpassed
+            echo "ANI percent bases aligned did not surpass the user-defined threshold of ~{percent_bases_aligned_threshold}" > ANI_TOP_SPECIES_MATCH.txt 
           else
-            echo "The highest ANI value $ANI_HIGHEST_PERCENT is greater than the user-defined threshold ~{ani_threshold}"
+            echo "The highest ANI value ${ANI_HIGHEST_PERCENT} is greater than the user-defined threshold ~{ani_threshold}"
+            echo "The highest percent bases aligned value ${ANI_HIGHEST_PERCENT_BASES_ALIGNED} is greater than the user-defined threshold ~{percent_bases_aligned_threshold}"
           fi
         fi
       else 

From 3e23ba7b6f0802498adb2062e7de9657b57b9591 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 6 Oct 2023 17:52:48 -0400
Subject: [PATCH 21/25] clarified messages when 2 thresholds are not passed

---
 tasks/quality_control/task_mummer_ani.wdl | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 2c8d87d11..630ff9d10 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -80,13 +80,13 @@ task animummer {
           echo "Comparing user-defined ANI threshold to highest ANI value..."
           # compare ANI_HIGHEST_PERCENT to ani_threshold using awk
           if ! awk "BEGIN{ exit ($ANI_HIGHEST_PERCENT < ~{ani_threshold} )}"; then
-            echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined threshold of ~{ani_threshold}"
-            echo "ANI top species match did not surpass the user-defined threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
+            echo "The highest ANI value $ANI_HIGHEST_PERCENT is less than the user-defined ANI threshold of ~{ani_threshold}"
+            echo "ANI top species match did not surpass the user-defined ANI threshold of ~{ani_threshold}" > ANI_TOP_SPECIES_MATCH.txt
           # else if: compare percent_bases_aligned_threshold to ANI_HIGHEST_PERCENT_BASES_ALIGNED using awk
           elif ! awk "BEGIN{ exit (${ANI_HIGHEST_PERCENT_BASES_ALIGNED} < ~{percent_bases_aligned_threshold} )}"; then
             echo "The highest ANI percent bases aligned value ${ANI_HIGHEST_PERCENT_BASES_ALIGNED} is less than the user-defined threshold of ~{percent_bases_aligned_threshold}"
             # overwrite ANI_TOP_SPECIES_MATCH.txt when percent_bases_aligned threshold is not surpassed
-            echo "ANI percent bases aligned did not surpass the user-defined threshold of ~{percent_bases_aligned_threshold}" > ANI_TOP_SPECIES_MATCH.txt 
+            echo "ANI top species match did not surpass the user-defined percent bases aligned threshold of ~{percent_bases_aligned_threshold}" > ANI_TOP_SPECIES_MATCH.txt 
           else
             echo "The highest ANI value ${ANI_HIGHEST_PERCENT} is greater than the user-defined threshold ~{ani_threshold}"
             echo "The highest percent bases aligned value ${ANI_HIGHEST_PERCENT_BASES_ALIGNED} is greater than the user-defined threshold ~{percent_bases_aligned_threshold}"

From 1c86b20ce0e63dfbb2d936d2567b9c018ed0f5b1 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Tue, 10 Oct 2023 12:16:41 -0400
Subject: [PATCH 22/25] removed accidental duplication in code introduced
 during merge conflict resolution (my bad!)

---
 tasks/utilities/task_broad_terra_tools.wdl       | 2 --
 workflows/theiaprok/wf_theiaprok_fasta.wdl       | 2 --
 workflows/theiaprok/wf_theiaprok_illumina_pe.wdl | 2 --
 workflows/theiaprok/wf_theiaprok_illumina_se.wdl | 2 --
 workflows/theiaprok/wf_theiaprok_ont.wdl         | 2 --
 5 files changed, 10 deletions(-)

diff --git a/tasks/utilities/task_broad_terra_tools.wdl b/tasks/utilities/task_broad_terra_tools.wdl
index 8ee0f3613..eb819b302 100644
--- a/tasks/utilities/task_broad_terra_tools.wdl
+++ b/tasks/utilities/task_broad_terra_tools.wdl
@@ -99,7 +99,6 @@ task export_taxon_tables {
     String? kmerfinder_query_coverage
     String? kmerfinder_template_coverage
     String? kmerfinder_database
-    String? ani_docker
     File? amrfinderplus_all_report
     File? amrfinderplus_amr_report
     File? amrfinderplus_stress_report
@@ -604,7 +603,6 @@ task export_taxon_tables {
       "kmerfinder_query_coverage": "~{kmerfinder_query_coverage}",
       "kmerfinder_template_coverage": "~{kmerfinder_template_coverage}",
       "kmerfinder_database": "~{kmerfinder_database}",
-      "ani_docker": "~{ani_docker}",
       "resfinder_pheno_table": "~{resfinder_pheno_table}",
       "resfinder_pheno_table_species": "~{resfinder_pheno_table_species}",
       "resfinder_seqs": "~{resfinder_seqs}",
diff --git a/workflows/theiaprok/wf_theiaprok_fasta.wdl b/workflows/theiaprok/wf_theiaprok_fasta.wdl
index 65f67fdc2..c5fe43576 100644
--- a/workflows/theiaprok/wf_theiaprok_fasta.wdl
+++ b/workflows/theiaprok/wf_theiaprok_fasta.wdl
@@ -182,7 +182,6 @@ workflow theiaprok_fasta {
         kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
         kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
         kmerfinder_database = kmerfinder.kmerfinder_database,
-        ani_docker = ani.ani_docker,
         amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
         amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
         amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -437,7 +436,6 @@ workflow theiaprok_fasta {
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
     String? kmerfinder_database = kmerfinder.kmerfinder_database
-    String? ani_mummer_docker = ani.ani_docker
     # NCBI-AMRFinderPlus Outputs
     File amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report
     File amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report
diff --git a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
index 15e807e8e..4ffbf5ee9 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -307,7 +307,6 @@ workflow theiaprok_illumina_pe {
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
             kmerfinder_database = kmerfinder.kmerfinder_database,
-            ani_docker = ani.ani_docker,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -631,7 +630,6 @@ workflow theiaprok_illumina_pe {
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
     String? kmerfinder_database = kmerfinder.kmerfinder_database
-    String? ani_mummer_docker = ani.ani_docker
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report
diff --git a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
index 9f27ce061..365b1ed2f 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
@@ -280,7 +280,6 @@ workflow theiaprok_illumina_se {
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
             kmerfinder_database = kmerfinder.kmerfinder_database,
-            ani_docker = ani.ani_docker,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -583,7 +582,6 @@ workflow theiaprok_illumina_se {
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
     String? kmerfinder_database = kmerfinder.kmerfinder_database
-    String? ani_mummer_docker = ani.ani_docker
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report
diff --git a/workflows/theiaprok/wf_theiaprok_ont.wdl b/workflows/theiaprok/wf_theiaprok_ont.wdl
index 3acc68b30..593f19723 100644
--- a/workflows/theiaprok/wf_theiaprok_ont.wdl
+++ b/workflows/theiaprok/wf_theiaprok_ont.wdl
@@ -278,7 +278,6 @@ workflow theiaprok_ont {
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
             kmerfinder_database = kmerfinder.kmerfinder_database,
-            ani_docker = ani.ani_docker,
             amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus.amrfinderplus_stress_report,
@@ -557,7 +556,6 @@ workflow theiaprok_ont {
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
     String? kmerfinder_database = kmerfinder.kmerfinder_database
-    String? ani_mummer_docker = ani.ani_docker
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report

From d25e9565d8c5ff613490b38fcdbac931721b78b4 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Tue, 10 Oct 2023 12:36:33 -0400
Subject: [PATCH 23/25] update CI for theiaprok illumina pe and se

---
 .../workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml  | 7 +++----
 .../workflows/theiaprok/test_wf_theiaprok_illumina_se.yml  | 7 +++----
 2 files changed, 6 insertions(+), 8 deletions(-)

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
index 52c7979fc..19a51c2ae 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -574,8 +574,7 @@
     - path: miniwdl_run/wdl/tasks/quality_control/task_fastq_scan.wdl
       contains: ["version", "fastq_scan", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_mummer_ani.wdl
-      md5sum: 33a111c260b2fb0b6db6d9ed77ec8e87
-      md5sum: 33a111c260b2fb0b6db6d9ed77ec8e87
+      md5sum: b578e00dc2dd8a978718e0b67ee9330a
     - path: miniwdl_run/wdl/tasks/quality_control/task_ncbi_scrub.wdl
       contains: ["version", "scrub", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_qc_check_phb.wdl
@@ -633,9 +632,9 @@
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl
       md5sum: 024971d1439dff7d59c0a26a824bd2c6
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
-      md5sum: 43ef050bde1fb8755f38e697a1794918
+      md5sum: 26ac9a20c8043a28d373bfe0ca361cc6
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
-      md5sum: 6dc6c393281a19e8dcbcc15964b8e08a
+      md5sum: ee5c79cbc195aad75c0a39c3c2acd28e
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: 2614fac622fa2035b80a7b220b1aed86
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_pe.wdl
diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
index 013ad6f4b..da8b5054f 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -544,8 +544,7 @@
     - path: miniwdl_run/wdl/tasks/quality_control/task_fastq_scan.wdl
       contains: ["version", "fastq_scan", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_mummer_ani.wdl
-      md5sum: 33a111c260b2fb0b6db6d9ed77ec8e87
-      md5sum: 33a111c260b2fb0b6db6d9ed77ec8e87
+      md5sum: b578e00dc2dd8a978718e0b67ee9330a
     - path: miniwdl_run/wdl/tasks/quality_control/task_qc_check_phb.wdl
       contains: ["version", "qc_check_table", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_quast.wdl
@@ -601,9 +600,9 @@
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl
       md5sum: 024971d1439dff7d59c0a26a824bd2c6
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
-      md5sum: 1d7365ed0f29a7632568a37bfce0f1b6
+      md5sum: 26ac9a20c8043a28d373bfe0ca361cc6
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
-      md5sum: d144cb871c175029a34107a369e70c5d
+      md5sum: 68a10e1e76ba0d7ec4c119e4f5fc3ecd
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: 2614fac622fa2035b80a7b220b1aed86
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl

From b500a8ac5df0443ef5f5022068779dd3358b0d91 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 20 Oct 2023 12:22:30 -0400
Subject: [PATCH 24/25] lower default ani_threshold to 80 (CDC standard) from
 85

---
 tasks/quality_control/task_mummer_ani.wdl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tasks/quality_control/task_mummer_ani.wdl b/tasks/quality_control/task_mummer_ani.wdl
index 630ff9d10..fc048bfb1 100644
--- a/tasks/quality_control/task_mummer_ani.wdl
+++ b/tasks/quality_control/task_mummer_ani.wdl
@@ -7,7 +7,7 @@ task animummer {
     File? ref_genome
     Float mash_filter = 0.9
     # these 2 thresholds were set as they are used by CDC enterics lab/PulseNet for ANI thresholds
-    Float ani_threshold = 85.0
+    Float ani_threshold = 80.0
     Float percent_bases_aligned_threshold = 70.0
     String docker= "us-docker.pkg.dev/general-theiagen/staphb/mummer:4.0.0-rgdv2"
     Int cpus = 4

From 5a540523e046319c981062a79b9031a4c2ab09d3 Mon Sep 17 00:00:00 2001
From: kapsakcj <kapsakcj@gmail.com>
Date: Fri, 20 Oct 2023 12:39:56 -0400
Subject: [PATCH 25/25] update CI

---
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml | 6 +++---
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml | 6 +++---
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
index 20d79901a..8a6a1e96a 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -574,7 +574,7 @@
     - path: miniwdl_run/wdl/tasks/quality_control/task_fastq_scan.wdl
       contains: ["version", "fastq_scan", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_mummer_ani.wdl
-      md5sum: b578e00dc2dd8a978718e0b67ee9330a
+      md5sum: d50a40d533d51834d0000971dc2c2014
     - path: miniwdl_run/wdl/tasks/quality_control/task_ncbi_scrub.wdl
       contains: ["version", "scrub", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_qc_check_phb.wdl
@@ -624,7 +624,7 @@
     - path: miniwdl_run/wdl/tasks/species_typing/task_ts_mlst.wdl
       md5sum: d49ae0b02e798af0636eb2721bb434b4
     - path: miniwdl_run/wdl/tasks/task_versioning.wdl
-      md5sum: 9c02b95d23a602d9842fdeac883037b1
+      md5sum: 3469cf6787f279ffa81fa50f6257fcd7
     - path: miniwdl_run/wdl/tasks/taxon_id/task_gambit.wdl
       md5sum: 08987d952c67c6ff6debf6898af15f9a
     - path: miniwdl_run/wdl/tasks/taxon_id/task_kraken2.wdl
@@ -634,7 +634,7 @@
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
       md5sum: 26ac9a20c8043a28d373bfe0ca361cc6
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
-      md5sum: 0c87a7279c4870a821c3dc1db9a6a94b
+      md5sum: ac4971ad992c3e8abee5d3817928a8f2
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: 00bd2489b2a7aa5b88340a940961a857
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_pe.wdl
diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
index 35fafa245..714fb937e 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -544,7 +544,7 @@
     - path: miniwdl_run/wdl/tasks/quality_control/task_fastq_scan.wdl
       contains: ["version", "fastq_scan", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_mummer_ani.wdl
-      md5sum: b578e00dc2dd8a978718e0b67ee9330a
+      md5sum: d50a40d533d51834d0000971dc2c2014
     - path: miniwdl_run/wdl/tasks/quality_control/task_qc_check_phb.wdl
       contains: ["version", "qc_check_table", "output"]
     - path: miniwdl_run/wdl/tasks/quality_control/task_quast.wdl
@@ -592,7 +592,7 @@
     - path: miniwdl_run/wdl/tasks/species_typing/task_ts_mlst.wdl
       md5sum: d49ae0b02e798af0636eb2721bb434b4
     - path: miniwdl_run/wdl/tasks/task_versioning.wdl
-      md5sum: 9c02b95d23a602d9842fdeac883037b1
+      md5sum: 3469cf6787f279ffa81fa50f6257fcd7
     - path: miniwdl_run/wdl/tasks/taxon_id/task_gambit.wdl
       md5sum: 08987d952c67c6ff6debf6898af15f9a
     - path: miniwdl_run/wdl/tasks/taxon_id/task_kraken2.wdl
@@ -602,7 +602,7 @@
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
       md5sum: 26ac9a20c8043a28d373bfe0ca361cc6
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
-      md5sum: e1d9e75dae5176ceeb95b88a5d3bbba7
+      md5sum: 7303247badeb82119ccef528f7367f89
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: 00bd2489b2a7aa5b88340a940961a857
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl