From 76cf2d50be841a4afab8d8cd7b5693265fe3c5f6 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Mon, 11 Sep 2023 16:55:44 +0000
Subject: [PATCH 01/24] skeleton on kmerfinder task

---
 tasks/taxon_id/task_kmerfinder.wdl | 38 ++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
 create mode 100644 tasks/taxon_id/task_kmerfinder.wdl

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
new file mode 100644
index 000000000..6cb07e0c2
--- /dev/null
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -0,0 +1,38 @@
+version 1.0
+
+task kmerfinder {
+  input {
+    File assembly
+    File kmerfinder_db
+    String kmerfinder_db_name # allowed: archea, bacteria, fungi, protozoa, virus, typestrain
+    String samplename
+    String docker = "us-docker.pkg.dev/general-theiagen/biocontainers/kmerfinder:3.0.2--hdfd78af_0"
+    Int memory = 32
+    Int cpu = 4
+    String kmerfinder_args = ""
+  }
+  command <<<
+    # Decompress the kmerfinder database
+    mkdir db
+    tar -C ./db/ -xzvf ~{kmerfinder_db}  
+
+    # Run kmerfinder
+    kmerfinder.py \
+        -db ./db/~{kmerfinder_db_name} \
+        -i ~{assembly} \
+        -o ~{samplename} \
+        ~{kmerfinder_args} \
+        -x
+
+  >>>
+  output {
+    String kmerfinder_docker = docker
+  }
+  runtime {
+    docker: docker
+    memory: "~{memory} GB"
+    cpu: cpu
+    disks: "local-disk 100 SSD"
+    preemptible: 0
+  }
+}
\ No newline at end of file

From 9762160bb4f482c43eaa8fce3f5c8c69e5c51696 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 12 Sep 2023 09:16:46 +0000
Subject: [PATCH 02/24] remove kmerfinder_db_name

---
 tasks/taxon_id/task_kmerfinder.wdl | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 6cb07e0c2..8bbcd97a2 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -4,7 +4,6 @@ task kmerfinder {
   input {
     File assembly
     File kmerfinder_db
-    String kmerfinder_db_name # allowed: archea, bacteria, fungi, protozoa, virus, typestrain
     String samplename
     String docker = "us-docker.pkg.dev/general-theiagen/biocontainers/kmerfinder:3.0.2--hdfd78af_0"
     Int memory = 32
@@ -18,7 +17,7 @@ task kmerfinder {
 
     # Run kmerfinder
     kmerfinder.py \
-        -db ./db/~{kmerfinder_db_name} \
+        -db ./db/ \
         -i ~{assembly} \
         -o ~{samplename} \
         ~{kmerfinder_args} \

From 9457c555dae71dc670f695292df15a349733023f Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Wed, 13 Sep 2023 15:25:15 +0000
Subject: [PATCH 03/24] still trying to get kmerfinder to run

---
 tasks/taxon_id/task_kmerfinder.wdl | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 8bbcd97a2..518ab51a9 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -17,7 +17,8 @@ task kmerfinder {
 
     # Run kmerfinder
     kmerfinder.py \
-        -db ./db/ \
+        -db ./db/* \
+        -tax ./db/*/*.tax \
         -i ~{assembly} \
         -o ~{samplename} \
         ~{kmerfinder_args} \

From 52f173b6f753561f3cefe44844de1d2971e06d6d Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Wed, 13 Sep 2023 16:16:27 +0000
Subject: [PATCH 04/24] is this working now?!?

---
 tasks/taxon_id/task_kmerfinder.wdl | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 518ab51a9..198f568eb 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -1,9 +1,9 @@
 version 1.0
 
-task kmerfinder {
+task kmerfinder_bacteria {
   input {
     File assembly
-    File kmerfinder_db
+    File kmerfinder_db = "gs://theiagen-public-files-rp/terra/theiaprok-files/kmerfinder_bacteria.tar.gz"
     String samplename
     String docker = "us-docker.pkg.dev/general-theiagen/biocontainers/kmerfinder:3.0.2--hdfd78af_0"
     Int memory = 32
@@ -17,8 +17,8 @@ task kmerfinder {
 
     # Run kmerfinder
     kmerfinder.py \
-        -db ./db/* \
-        -tax ./db/*/*.tax \
+        -db ./db/bacteria/bacteria.ATG \
+        -tax ./db/bacteria/bacteria.tax \
         -i ~{assembly} \
         -o ~{samplename} \
         ~{kmerfinder_args} \

From 9fca4710335cbbaa0494e02c7cc901ad2763af6f Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Thu, 14 Sep 2023 10:04:41 +0000
Subject: [PATCH 05/24] first working version of kmerfinder task

---
 tasks/taxon_id/task_kmerfinder.wdl | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 198f568eb..0ffc972f9 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -11,7 +11,7 @@ task kmerfinder_bacteria {
     String kmerfinder_args = ""
   }
   command <<<
-    # Decompress the kmerfinder database
+    # Decompress the kmerfinder bacterial qdatabase
     mkdir db
     tar -C ./db/ -xzvf ~{kmerfinder_db}  
 
@@ -21,12 +21,15 @@ task kmerfinder_bacteria {
         -tax ./db/bacteria/bacteria.tax \
         -i ~{assembly} \
         -o ~{samplename} \
-        ~{kmerfinder_args} \
-        -x
+        ~{kmerfinder_args} 
 
+    mv ~{samplename}/results.txt ~{samplename}_kmerfinder.txt
+
+    # extract the top result
   >>>
   output {
     String kmerfinder_docker = docker
+    File kmerfinder_txt = "~{samplename}_kmerfinder.txt"
   }
   runtime {
     docker: docker

From 36b43ddc1ee7914687c0749bfac6a5ec87e2b3ab Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Thu, 14 Sep 2023 10:36:02 +0000
Subject: [PATCH 06/24] parse top hit

---
 tasks/taxon_id/task_kmerfinder.wdl | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 0ffc972f9..7a66fb708 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -23,6 +23,17 @@ task kmerfinder_bacteria {
         -o ~{samplename} \
         ~{kmerfinder_args} 
 
+    # parse outputs
+    if [ ! -f ~{samplename}/results.txt ]; then
+      PF="No hit detected in database"
+    else
+      PF="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 19)"
+        if [ "$PF" == "" ]; then
+          PF="No hit detected in database"
+        fi  
+    fi
+    echo $PF | tee TOP_HIT
+    
     mv ~{samplename}/results.txt ~{samplename}_kmerfinder.txt
 
     # extract the top result
@@ -30,6 +41,7 @@ task kmerfinder_bacteria {
   output {
     String kmerfinder_docker = docker
     File kmerfinder_txt = "~{samplename}_kmerfinder.txt"
+    String kmerfinder_top_hit = read_string("TOP_HIT")
   }
   runtime {
     docker: docker

From 9f18b7d895f3268d6a61e792eeff39bce0ab85dd Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Thu, 14 Sep 2023 10:45:42 +0000
Subject: [PATCH 07/24] make results output file optional

---
 tasks/taxon_id/task_kmerfinder.wdl | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 7a66fb708..bf74bbcb9 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -30,17 +30,14 @@ task kmerfinder_bacteria {
       PF="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 19)"
         if [ "$PF" == "" ]; then
           PF="No hit detected in database"
-        fi  
+        fi
+      mv ~{samplename}/results.txt ~{samplename}_kmerfinder.txt  
     fi
     echo $PF | tee TOP_HIT
-    
-    mv ~{samplename}/results.txt ~{samplename}_kmerfinder.txt
-
-    # extract the top result
   >>>
   output {
     String kmerfinder_docker = docker
-    File kmerfinder_txt = "~{samplename}_kmerfinder.txt"
+    File? kmerfinder_txt = "~{samplename}_kmerfinder.txt"
     String kmerfinder_top_hit = read_string("TOP_HIT")
   }
   runtime {

From 98ba2c628cfc3ab2f729739eb68ba072cd593f60 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Thu, 14 Sep 2023 13:53:35 +0000
Subject: [PATCH 08/24] final version of task (for now..)

---
 tasks/taxon_id/task_kmerfinder.wdl | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index bf74bbcb9..332ef8edd 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -3,11 +3,12 @@ version 1.0
 task kmerfinder_bacteria {
   input {
     File assembly
-    File kmerfinder_db = "gs://theiagen-public-files-rp/terra/theiaprok-files/kmerfinder_bacteria.tar.gz"
     String samplename
+    File kmerfinder_db = "gs://theiagen-public-files-rp/terra/theiaprok-files/kmerfinder_bacteria.tar.gz"
     String docker = "us-docker.pkg.dev/general-theiagen/biocontainers/kmerfinder:3.0.2--hdfd78af_0"
     Int memory = 32
     Int cpu = 4
+    Int disk_size = 100
     String kmerfinder_args = ""
   }
   command <<<
@@ -31,20 +32,20 @@ task kmerfinder_bacteria {
         if [ "$PF" == "" ]; then
           PF="No hit detected in database"
         fi
-      mv ~{samplename}/results.txt ~{samplename}_kmerfinder.txt  
+      mv -v ~{samplename}/results.txt ~{samplename}_kmerfinder.tsv
     fi
     echo $PF | tee TOP_HIT
   >>>
   output {
     String kmerfinder_docker = docker
-    File? kmerfinder_txt = "~{samplename}_kmerfinder.txt"
+    File? kmerfinder_results_tsv = "~{samplename}_kmerfinder.tsv"
     String kmerfinder_top_hit = read_string("TOP_HIT")
   }
   runtime {
     docker: docker
     memory: "~{memory} GB"
     cpu: cpu
-    disks: "local-disk 100 SSD"
+    disks: "local-disk ~{disk_size} SSD"
     preemptible: 0
   }
 }
\ No newline at end of file

From d14549a92b005c13a64c2a959d797d4020b63f3e Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Thu, 14 Sep 2023 14:10:48 +0000
Subject: [PATCH 09/24] add kmerfinder_bacteria to theiaprok!

---
 tasks/utilities/task_broad_terra_tools.wdl       |  6 ++++++
 workflows/theiaprok/wf_theiaprok_fasta.wdl       | 16 ++++++++++++++++
 workflows/theiaprok/wf_theiaprok_illumina_pe.wdl | 16 ++++++++++++++++
 workflows/theiaprok/wf_theiaprok_illumina_se.wdl | 16 ++++++++++++++++
 workflows/theiaprok/wf_theiaprok_ont.wdl         | 16 ++++++++++++++++
 5 files changed, 70 insertions(+)

diff --git a/tasks/utilities/task_broad_terra_tools.wdl b/tasks/utilities/task_broad_terra_tools.wdl
index b585d73f2..f8715322b 100644
--- a/tasks/utilities/task_broad_terra_tools.wdl
+++ b/tasks/utilities/task_broad_terra_tools.wdl
@@ -92,6 +92,9 @@ task export_taxon_tables {
     File? ani_output_tsv
     String? ani_top_species_match 
     String? ani_mummer_version
+    String? kmerfinder_docker
+    File? kmerfinder_results_tsv
+    String? kmerfinder_top_hit
     File? amrfinderplus_all_report
     File? amrfinderplus_amr_report
     File? amrfinderplus_stress_report
@@ -587,6 +590,9 @@ task export_taxon_tables {
       "ani_output_tsv": "~{ani_output_tsv}",
       "ani_top_species_match": "~{ani_top_species_match}",
       "ani_mummer_version": "~{ani_mummer_version}",
+      "kmerfinder_docker": "~{kmerfinder_docker}",
+      "kmerfinder_results_tsv": "~{kmerfinder_results_tsv}",
+      "kmerfinder_top_hit": "~{kmerfinder_top_hit}",
       "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 cf4f7df7b..82f2d0f8a 100644
--- a/workflows/theiaprok/wf_theiaprok_fasta.wdl
+++ b/workflows/theiaprok/wf_theiaprok_fasta.wdl
@@ -5,6 +5,7 @@ import "../../tasks/quality_control/task_quast.wdl" as quast_task
 import "../../tasks/quality_control/task_busco.wdl" as busco_task
 import "../../tasks/taxon_id/task_gambit.wdl" as gambit_task
 import "../../tasks/quality_control/task_mummer_ani.wdl" as ani_task
+import "../../tasks/taxon_id/task_kmerfinder.wdl" as kmerfinder_task
 import "../../tasks/gene_typing/task_amrfinderplus.wdl" as amrfinderplus
 import "../../tasks/gene_typing/task_resfinder.wdl" as resfinder
 import "../../tasks/species_typing/task_ts_mlst.wdl" as ts_mlst_task
@@ -35,6 +36,7 @@ workflow theiaprok_fasta {
     String terra_workspace="NA"
     # module options
     Boolean call_ani = false # by default do not call ANI task, but user has ability to enable this task if working with enteric pathogens or supply their own high-quality reference genome
+    Boolean call_kmerfinder = false
     Boolean call_resfinder = false
     String genome_annotation = "prokka" # options: "prokka" or "bakta"
     String? expected_taxon # allow user to provide organism (e.g. "Clostridioides_difficile") string to amrfinder. Useful when gambit does not predict the correct species
@@ -66,6 +68,13 @@ workflow theiaprok_fasta {
         samplename = samplename
     }
   }
+  if (call_kmerfinder) {
+    call kmerfinder_task.kmerfinder_bacteria as kmerfinder {
+      input:
+        assembly = assembly_fasta,
+        samplename = samplename
+    }
+  }
   call amrfinderplus.amrfinderplus_nuc as amrfinderplus_task {
     input:
       assembly = assembly_fasta,
@@ -166,6 +175,9 @@ 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,
+        kmerfinder_docker = kmerfinder.kmerfinder_docker,
+        kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
+        kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
         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 +423,10 @@ 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
+    # kmerfinder outputs
+    String? kmerfinder_docker = kmerfinder.kmerfinder_docker
+    File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
+    String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     # 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 e960a7eb1..51af2f0e8 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -9,6 +9,7 @@ import "../../tasks/quality_control/task_screen.wdl" as screen
 import "../../tasks/quality_control/task_busco.wdl" as busco_task
 import "../../tasks/taxon_id/task_gambit.wdl" as gambit_task
 import "../../tasks/quality_control/task_mummer_ani.wdl" as ani_task
+import "../../tasks/taxon_id/task_kmerfinder.wdl" as kmerfinder_task
 import "../../tasks/gene_typing/task_amrfinderplus.wdl" as amrfinderplus
 import "../../tasks/gene_typing/task_resfinder.wdl" as resfinder
 import "../../tasks/species_typing/task_ts_mlst.wdl" as ts_mlst_task
@@ -53,6 +54,7 @@ workflow theiaprok_illumina_pe {
     Int trim_window_size = 10
     # module options
     Boolean call_ani = false # by default do not call ANI task, but user has ability to enable this task if working with enteric pathogens or supply their own high-quality reference genome
+    Boolean call_kmerfinder = false 
     Boolean call_resfinder = false
     String genome_annotation = "prokka" # options: "prokka" or "bakta"
     String? expected_taxon  # allow user to provide organism (e.g. "Clostridioides_difficile") string to amrfinder. Useful when gambit does not predict the correct species    # qc check parameters
@@ -142,6 +144,13 @@ workflow theiaprok_illumina_pe {
             samplename = samplename
         }
       }
+      if (call_kmerfinder) {
+        call kmerfinder_task.kmerfinder_bacteria as kmerfinder {
+          input:
+            assembly = shovill_pe.assembly_fasta,
+            samplename = samplename
+        }
+      }
       call amrfinderplus.amrfinderplus_nuc as amrfinderplus_task {
         input:
           assembly = shovill_pe.assembly_fasta,
@@ -291,6 +300,9 @@ 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,
+            kmerfinder_docker = kmerfinder.kmerfinder_docker,
+            kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
+            kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             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 +617,10 @@ 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
+    # kmerfinder outputs
+    String? kmerfinder_docker = kmerfinder.kmerfinder_docker
+    File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
+    String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     # 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 4f2025bd6..cf93aec8d 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
@@ -9,6 +9,7 @@ import "../../tasks/quality_control/task_screen.wdl" as screen
 import "../../tasks/quality_control/task_busco.wdl" as busco_task
 import "../../tasks/taxon_id/task_gambit.wdl" as gambit_task
 import "../../tasks/quality_control/task_mummer_ani.wdl" as ani_task
+import "../../tasks/taxon_id/task_kmerfinder.wdl" as kmerfinder_task
 import "../../tasks/gene_typing/task_amrfinderplus.wdl" as amrfinderplus
 import "../../tasks/gene_typing/task_resfinder.wdl" as resfinder
 import "../../tasks/species_typing/task_ts_mlst.wdl" as ts_mlst_task
@@ -52,6 +53,7 @@ workflow theiaprok_illumina_se {
     Int trim_window_size = 4
     # module options
     Boolean call_ani = false # by default do not call ANI task, but user has ability to enable this task if working with enteric pathogens or supply their own high-quality reference genome
+    Boolean call_kmerfinder = false 
     Boolean call_resfinder = false
     String genome_annotation = "prokka" # options: "prokka" or "bakta"
     String? expected_taxon # allow user to provide organism (e.g. "Clostridioides_difficile") string to amrfinder. Useful when gambit does not predict the correct species
@@ -135,6 +137,13 @@ workflow theiaprok_illumina_se {
             samplename = samplename
         }
       }
+      if (call_kmerfinder) {
+        call kmerfinder_task.kmerfinder_bacteria as kmerfinder {
+          input:
+            assembly = shovill_se.assembly_fasta,
+            samplename = samplename
+        }
+      }
       call amrfinderplus.amrfinderplus_nuc as amrfinderplus_task {
         input:
           assembly = shovill_se.assembly_fasta,
@@ -264,6 +273,9 @@ 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,
+            kmerfinder_docker = kmerfinder.kmerfinder_docker,
+            kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
+            kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -557,6 +569,10 @@ 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
+    # kmerfinder outputs
+    String? kmerfinder_docker = kmerfinder.kmerfinder_docker
+    File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
+    String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     # 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 06628be0a..e8421283c 100644
--- a/workflows/theiaprok/wf_theiaprok_ont.wdl
+++ b/workflows/theiaprok/wf_theiaprok_ont.wdl
@@ -9,6 +9,7 @@ import "../../tasks/quality_control/task_screen.wdl" as screen_task
 import "../../tasks/quality_control/task_busco.wdl" as busco_task
 import "../../tasks/taxon_id/task_gambit.wdl" as gambit_task
 import "../../tasks/quality_control/task_mummer_ani.wdl" as ani_task
+import "../../tasks/taxon_id/task_kmerfinder.wdl" as kmerfinder_task
 import "../../tasks/gene_typing/task_amrfinderplus.wdl" as amrfinderplus_task
 import "../../tasks/gene_typing/task_resfinder.wdl" as resfinder_task
 import "../../tasks/species_typing/task_ts_mlst.wdl" as ts_mlst_task
@@ -48,6 +49,7 @@ workflow theiaprok_ont {
     Int min_coverage = 5 # reduced from 10 because some institutions sequence at lower depth because of longer read length
     # module options
     Boolean call_ani = false # by default do not call ANI task, but user has ability to enable this task if working with enteric pathogens or supply their own high-quality reference genome
+    Boolean call_kmerfinder = false
     Boolean call_resfinder = false
     String genome_annotation = "prokka" # options: "prokka" or "bakta"
     String? expected_taxon # allow user to provide organism (e.g. "Clostridioides_difficile") string to amrfinder. Useful when gambit does not predict the correct species
@@ -131,6 +133,13 @@ workflow theiaprok_ont {
             samplename = samplename
         }
       }
+      if (call_kmerfinder) {
+        call kmerfinder_task.kmerfinder_bacteria as kmerfinder {
+          input:
+            assembly = dragonflye.assembly_fasta,
+            samplename = samplename
+        }
+      }
       call amrfinderplus_task.amrfinderplus_nuc as amrfinderplus {
         input:
           assembly = dragonflye.assembly_fasta,
@@ -262,6 +271,9 @@ 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,
+            kmerfinder_docker = kmerfinder.kmerfinder_docker,
+            kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
+            kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus.amrfinderplus_stress_report,
@@ -531,6 +543,10 @@ 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
+    # kmerfinder outputs
+    String? kmerfinder_docker = kmerfinder.kmerfinder_docker
+    File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
+    String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report

From c55ca1012252cd68870033b3d46f018776905d7c Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Mon, 18 Sep 2023 15:56:39 +0000
Subject: [PATCH 10/24] fix typo

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

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 332ef8edd..5e9c14f0e 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -12,7 +12,7 @@ task kmerfinder_bacteria {
     String kmerfinder_args = ""
   }
   command <<<
-    # Decompress the kmerfinder bacterial qdatabase
+    # Decompress the kmerfinder bacterial database
     mkdir db
     tar -C ./db/ -xzvf ~{kmerfinder_db}  
 

From c834b8985dc61bf5e1c8977c04a18a58b5a50c84 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 08:15:26 +0000
Subject: [PATCH 11/24] pass empty file for tests - kmerfinder_db

---
 tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json | 4 +++-
 tests/inputs/theiaprok/wf_theiaprok_illumina_se.json | 4 +++-
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
index c5e3444a7..143c2737a 100644
--- a/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
+++ b/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
@@ -3,6 +3,7 @@
     "theiaprok_illumina_pe.read1_raw": "bactopia-tests/data/species/portiera/illumina/SRR2838702_R1.fastq.gz",
     "theiaprok_illumina_pe.read2_raw": "bactopia-tests/data/species/portiera/illumina/SRR2838702_R2.fastq.gz",
     "theiaprok_illumina_pe.skip_screen": true,
+    "theiaprok_illumina_pe.call_kmerfinder": false,
     "theiaprok_illumina_pe.read_QC_trim.read_processing": "trimmomatic",
     "theiaprok_illumina_pe.read_QC_trim.call_midas": false,
     "theiaprok_illumina_pe.read_QC_trim.midas.midas_db" : "./tests/inputs/empty-for-test.txt",
@@ -36,5 +37,6 @@
     "theiaprok_illumina_pe.merlin_magic.cladetyper.ref_clade5_annotated" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_pe.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_pe.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
-    "theiaprok_illumina_pe.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt"
+    "theiaprok_illumina_pe.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt",
+    "theiaprok_illumina_pe.gambit.kmerfinder_db" : "./tests/inputs/completely-empty-for-test.txt"
 }
diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
index 126792e6d..c0aacc2bd 100644
--- a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
+++ b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
@@ -2,6 +2,7 @@
     "theiaprok_illumina_se.samplename": "test",
     "theiaprok_illumina_se.read1_raw": "bactopia-tests/data/species/portiera/illumina/SRR2838702_R1.fastq.gz",
     "theiaprok_illumina_se.skip_screen": true,
+    "theiaprok_illumina_pe.call_kmerfinder": false,
     "theiaprok_illumina_se.read_QC_trim.read_processing": "trimmomatic",
     "theiaprok_illumina_se.read_QC_trim.call_midas": false,
     "theiaprok_illumina_se.read_QC_trim.midas.midas_db" : "./tests/inputs/empty-for-test.txt",
@@ -35,5 +36,6 @@
     "theiaprok_illumina_se.merlin_magic.cladetyper.ref_clade5_annotated" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_se.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_se.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
-    "theiaprok_illumina_se.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt"
+    "theiaprok_illumina_se.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt",
+    "theiaprok_illumina_pe.gambit.kmerfinder_db" : "./tests/inputs/completely-empty-for-test.txt"
 }

From 2bdc5eea36a18013b00600e4a1c9607478ae8472 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 08:31:20 +0000
Subject: [PATCH 12/24] fix input json

---
 tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json | 2 +-
 tests/inputs/theiaprok/wf_theiaprok_illumina_se.json | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
index 143c2737a..e66701e50 100644
--- a/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
+++ b/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
@@ -38,5 +38,5 @@
     "theiaprok_illumina_pe.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_pe.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
     "theiaprok_illumina_pe.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt",
-    "theiaprok_illumina_pe.gambit.kmerfinder_db" : "./tests/inputs/completely-empty-for-test.txt"
+    "theiaprok_illumina_pe.kmerfinder.kmerfinder_db" : "./tests/inputs/completely-empty-for-test.txt"
 }
diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
index c0aacc2bd..fbaface2e 100644
--- a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
+++ b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
@@ -37,5 +37,5 @@
     "theiaprok_illumina_se.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_se.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
     "theiaprok_illumina_se.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt",
-    "theiaprok_illumina_pe.gambit.kmerfinder_db" : "./tests/inputs/completely-empty-for-test.txt"
+    "theiaprok_illumina_pe.kmerfinder.kmerfinder_db" : "./tests/inputs/completely-empty-for-test.txt"
 }

From 97842168fe5f5c21011f819434d59dd4776ebd06 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 08:37:40 +0000
Subject: [PATCH 13/24] add kmerfinder query coverage to theiaprok SE, PE,
 FASTA and ONT

---
 tasks/taxon_id/task_kmerfinder.wdl               | 5 +++++
 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 ++
 6 files changed, 15 insertions(+)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 5e9c14f0e..1b0265d37 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -27,19 +27,24 @@ task kmerfinder_bacteria {
     # parse outputs
     if [ ! -f ~{samplename}/results.txt ]; then
       PF="No hit detected in database"
+      QC="No hit detected in database"
     else
       PF="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 19)"
+      QC="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 6)"
         if [ "$PF" == "" ]; then
           PF="No hit detected in database"
+          QC="No hit detected in database"
         fi
       mv -v ~{samplename}/results.txt ~{samplename}_kmerfinder.tsv
     fi
     echo $PF | tee TOP_HIT
+    echo $QC | tee QC_METRIC
   >>>
   output {
     String kmerfinder_docker = docker
     File? kmerfinder_results_tsv = "~{samplename}_kmerfinder.tsv"
     String kmerfinder_top_hit = read_string("TOP_HIT")
+    String kmerfinder_query_coverage = read_string("QC_METRIC")
   }
   runtime {
     docker: docker
diff --git a/tasks/utilities/task_broad_terra_tools.wdl b/tasks/utilities/task_broad_terra_tools.wdl
index f8715322b..711ebcdf7 100644
--- a/tasks/utilities/task_broad_terra_tools.wdl
+++ b/tasks/utilities/task_broad_terra_tools.wdl
@@ -95,6 +95,7 @@ task export_taxon_tables {
     String? kmerfinder_docker
     File? kmerfinder_results_tsv
     String? kmerfinder_top_hit
+    String? kmerfinder_query_coverage
     File? amrfinderplus_all_report
     File? amrfinderplus_amr_report
     File? amrfinderplus_stress_report
@@ -593,6 +594,7 @@ task export_taxon_tables {
       "kmerfinder_docker": "~{kmerfinder_docker}",
       "kmerfinder_results_tsv": "~{kmerfinder_results_tsv}",
       "kmerfinder_top_hit": "~{kmerfinder_top_hit}",
+      "kmerfinder_query_coverage": "~{kmerfinder_query_coverage}",
       "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 82f2d0f8a..db3682394 100644
--- a/workflows/theiaprok/wf_theiaprok_fasta.wdl
+++ b/workflows/theiaprok/wf_theiaprok_fasta.wdl
@@ -178,6 +178,7 @@ workflow theiaprok_fasta {
         kmerfinder_docker = kmerfinder.kmerfinder_docker,
         kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
         kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
+        kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
         amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
         amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
         amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -427,6 +428,7 @@ workflow theiaprok_fasta {
     String? kmerfinder_docker = kmerfinder.kmerfinder_docker
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
+    String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     # 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 51af2f0e8..4f56f085a 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -303,6 +303,7 @@ workflow theiaprok_illumina_pe {
             kmerfinder_docker = kmerfinder.kmerfinder_docker,
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
+            kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -621,6 +622,7 @@ workflow theiaprok_illumina_pe {
     String? kmerfinder_docker = kmerfinder.kmerfinder_docker
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
+    String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     # 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 cf93aec8d..599b902c5 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
@@ -276,6 +276,7 @@ workflow theiaprok_illumina_se {
             kmerfinder_docker = kmerfinder.kmerfinder_docker,
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
+            kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -573,6 +574,7 @@ workflow theiaprok_illumina_se {
     String? kmerfinder_docker = kmerfinder.kmerfinder_docker
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
+    String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     # 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 e8421283c..ed96dcedc 100644
--- a/workflows/theiaprok/wf_theiaprok_ont.wdl
+++ b/workflows/theiaprok/wf_theiaprok_ont.wdl
@@ -274,6 +274,7 @@ workflow theiaprok_ont {
             kmerfinder_docker = kmerfinder.kmerfinder_docker,
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
+            kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus.amrfinderplus_stress_report,
@@ -547,6 +548,7 @@ workflow theiaprok_ont {
     String? kmerfinder_docker = kmerfinder.kmerfinder_docker
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
+    String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report

From 3381183b8ef87f02009523646ff3fffd1bcb8e23 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 08:38:30 +0000
Subject: [PATCH 14/24] remove undeeded skip on kmerfinder (set to false by
 default)

---
 tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json | 1 -
 tests/inputs/theiaprok/wf_theiaprok_illumina_se.json | 1 -
 2 files changed, 2 deletions(-)

diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
index e66701e50..3dd186232 100644
--- a/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
+++ b/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
@@ -3,7 +3,6 @@
     "theiaprok_illumina_pe.read1_raw": "bactopia-tests/data/species/portiera/illumina/SRR2838702_R1.fastq.gz",
     "theiaprok_illumina_pe.read2_raw": "bactopia-tests/data/species/portiera/illumina/SRR2838702_R2.fastq.gz",
     "theiaprok_illumina_pe.skip_screen": true,
-    "theiaprok_illumina_pe.call_kmerfinder": false,
     "theiaprok_illumina_pe.read_QC_trim.read_processing": "trimmomatic",
     "theiaprok_illumina_pe.read_QC_trim.call_midas": false,
     "theiaprok_illumina_pe.read_QC_trim.midas.midas_db" : "./tests/inputs/empty-for-test.txt",
diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
index fbaface2e..1eb3f13d6 100644
--- a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
+++ b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
@@ -2,7 +2,6 @@
     "theiaprok_illumina_se.samplename": "test",
     "theiaprok_illumina_se.read1_raw": "bactopia-tests/data/species/portiera/illumina/SRR2838702_R1.fastq.gz",
     "theiaprok_illumina_se.skip_screen": true,
-    "theiaprok_illumina_pe.call_kmerfinder": false,
     "theiaprok_illumina_se.read_QC_trim.read_processing": "trimmomatic",
     "theiaprok_illumina_se.read_QC_trim.call_midas": false,
     "theiaprok_illumina_se.read_QC_trim.midas.midas_db" : "./tests/inputs/empty-for-test.txt",

From e2dfb57dc527cb81f7af6f16d4b0145422328734 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 08:45:25 +0000
Subject: [PATCH 15/24] fix ops

---
 tests/inputs/theiaprok/wf_theiaprok_illumina_se.json | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
index 1eb3f13d6..9e0c42ed7 100644
--- a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
+++ b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
@@ -36,5 +36,5 @@
     "theiaprok_illumina_se.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_se.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
     "theiaprok_illumina_se.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt",
-    "theiaprok_illumina_pe.kmerfinder.kmerfinder_db" : "./tests/inputs/completely-empty-for-test.txt"
+    "theiaprok_illumina_se.kmerfinder.kmerfinder_db" : "./tests/inputs/completely-empty-for-test.txt"
 }

From b6dd70969fff98cbd15a53be76b328c6a8e920b0 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 09:01:45 +0000
Subject: [PATCH 16/24] update md5sum (part1)

---
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml | 4 ++--
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml | 2 +-
 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 5fa38e056..ae62062bb 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -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: 0236363c7f0694cd3f96416aa43e2f91
+      md5sum: 3c45d7016af73f22e16984c0e6693662
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
-      md5sum: a76d59109075ce8b861e63ffe70d7c77
+      md5sum: 43b2512c04abd5bd2907c2e9f59666dd
     - 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 dcb4661b5..15a6a8ef3 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -600,7 +600,7 @@
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl
       md5sum: 024971d1439dff7d59c0a26a824bd2c6
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
-      md5sum: 0236363c7f0694cd3f96416aa43e2f91
+      md5sum: 3c45d7016af73f22e16984c0e6693662
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
       md5sum: 858d33eb64b9bda618a47a999f370df5
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl

From be9a3e0828b5a49b2a097f98f81416fde85f8791 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 09:17:07 +0000
Subject: [PATCH 17/24] update md5sum (part2)

---
 tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
index 15a6a8ef3..145eb3705 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -602,7 +602,7 @@
     - path: miniwdl_run/wdl/tasks/utilities/task_broad_terra_tools.wdl
       md5sum: 3c45d7016af73f22e16984c0e6693662
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
-      md5sum: 858d33eb64b9bda618a47a999f370df5
+      md5sum: 97d84dae28752f6ff49307ae6962b7f2
     - 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 38fbef075097dc4eb408bf515e676447bcee72b8 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 17:03:21 +0000
Subject: [PATCH 18/24] fiz sister typo

---
 tasks/utilities/task_broad_terra_tools.wdl       | 4 ++--
 workflows/theiaprok/wf_theiaprok_fasta.wdl       | 4 ++--
 workflows/theiaprok/wf_theiaprok_illumina_pe.wdl | 4 ++--
 workflows/theiaprok/wf_theiaprok_ont.wdl         | 4 ++--
 4 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/tasks/utilities/task_broad_terra_tools.wdl b/tasks/utilities/task_broad_terra_tools.wdl
index 711ebcdf7..01bd5095c 100644
--- a/tasks/utilities/task_broad_terra_tools.wdl
+++ b/tasks/utilities/task_broad_terra_tools.wdl
@@ -172,7 +172,7 @@ task export_taxon_tables {
     String? lissero_serotype
     File? sistr_results
     File? sistr_allele_json
-    File? sister_allele_fasta
+    File? sistr_allele_fasta
     File? sistr_cgmlst
     String? sistr_version
     String? sistr_predicted_serotype
@@ -497,7 +497,7 @@ task export_taxon_tables {
       "lissero_serotype": "~{lissero_serotype}",
       "sistr_results": "~{sistr_results}",
       "sistr_allele_json": "~{sistr_allele_json}",
-      "sister_allele_fasta": "~{sister_allele_fasta}",
+      "sistr_allele_fasta": "~{sistr_allele_fasta}",
       "sistr_cgmlst": "~{sistr_cgmlst}",
       "sistr_version": "~{sistr_version}",
       "sistr_predicted_serotype": "~{sistr_predicted_serotype}",
diff --git a/workflows/theiaprok/wf_theiaprok_fasta.wdl b/workflows/theiaprok/wf_theiaprok_fasta.wdl
index db3682394..d4ddd54af 100644
--- a/workflows/theiaprok/wf_theiaprok_fasta.wdl
+++ b/workflows/theiaprok/wf_theiaprok_fasta.wdl
@@ -255,7 +255,7 @@ workflow theiaprok_fasta {
         lissero_serotype = merlin_magic.lissero_serotype,
         sistr_results = merlin_magic.sistr_results,
         sistr_allele_json = merlin_magic.sistr_allele_json,
-        sister_allele_fasta = merlin_magic.sistr_allele_fasta,
+        sistr_allele_fasta = merlin_magic.sistr_allele_fasta,
         sistr_cgmlst = merlin_magic.sistr_cgmlst,
         sistr_version = merlin_magic.sistr_version,
         sistr_predicted_serotype = merlin_magic.sistr_predicted_serotype,
@@ -515,7 +515,7 @@ workflow theiaprok_fasta {
     # Salmonella Typing
     File? sistr_results = merlin_magic.sistr_results
     File? sistr_allele_json = merlin_magic.sistr_allele_json
-    File? sister_allele_fasta = merlin_magic.sistr_allele_fasta
+    File? sistr_allele_fasta = merlin_magic.sistr_allele_fasta
     File? sistr_cgmlst = merlin_magic.sistr_cgmlst
     String? sistr_version = merlin_magic.sistr_version
     String? sistr_predicted_serotype = merlin_magic.sistr_predicted_serotype
diff --git a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
index 4f56f085a..c917d6ed2 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -380,7 +380,7 @@ workflow theiaprok_illumina_pe {
             lissero_serotype = merlin_magic.lissero_serotype,
             sistr_results = merlin_magic.sistr_results,
             sistr_allele_json = merlin_magic.sistr_allele_json,
-            sister_allele_fasta = merlin_magic.sistr_allele_fasta,
+            sistr_allele_fasta = merlin_magic.sistr_allele_fasta,
             sistr_cgmlst = merlin_magic.sistr_cgmlst,
             sistr_version = merlin_magic.sistr_version,
             sistr_predicted_serotype = merlin_magic.sistr_predicted_serotype,
@@ -737,7 +737,7 @@ workflow theiaprok_illumina_pe {
     # Salmonella Typing
     File? sistr_results = merlin_magic.sistr_results
     File? sistr_allele_json = merlin_magic.sistr_allele_json
-    File? sister_allele_fasta = merlin_magic.sistr_allele_fasta
+    File? sistr_allele_fasta = merlin_magic.sistr_allele_fasta
     File? sistr_cgmlst = merlin_magic.sistr_cgmlst
     String? sistr_version = merlin_magic.sistr_version
     String? sistr_predicted_serotype = merlin_magic.sistr_predicted_serotype
diff --git a/workflows/theiaprok/wf_theiaprok_ont.wdl b/workflows/theiaprok/wf_theiaprok_ont.wdl
index ed96dcedc..043d6a98e 100644
--- a/workflows/theiaprok/wf_theiaprok_ont.wdl
+++ b/workflows/theiaprok/wf_theiaprok_ont.wdl
@@ -341,7 +341,7 @@ workflow theiaprok_ont {
             lissero_serotype = merlin_magic.lissero_serotype,
             sistr_results = merlin_magic.sistr_results,
             sistr_allele_json = merlin_magic.sistr_allele_json,
-            sister_allele_fasta = merlin_magic.sistr_allele_fasta,
+            sistr_allele_fasta = merlin_magic.sistr_allele_fasta,
             sistr_cgmlst = merlin_magic.sistr_cgmlst,
             sistr_version = merlin_magic.sistr_version,
             sistr_predicted_serotype = merlin_magic.sistr_predicted_serotype,
@@ -651,7 +651,7 @@ workflow theiaprok_ont {
     # Salmonella Typing
     File? sistr_results = merlin_magic.sistr_results
     File? sistr_allele_json = merlin_magic.sistr_allele_json
-    File? sister_allele_fasta = merlin_magic.sistr_allele_fasta
+    File? sistr_allele_fasta = merlin_magic.sistr_allele_fasta
     File? sistr_cgmlst = merlin_magic.sistr_cgmlst
     String? sistr_version = merlin_magic.sistr_version
     String? sistr_predicted_serotype = merlin_magic.sistr_predicted_serotype

From 5c7b44e4af4d89862d8e6617cf6c77ecb4a6b3cb Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 17:06:52 +0000
Subject: [PATCH 19/24] missed a file

---
 workflows/theiaprok/wf_theiaprok_illumina_se.wdl | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
index 599b902c5..b4822d18d 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
@@ -353,7 +353,7 @@ workflow theiaprok_illumina_se {
             lissero_serotype = merlin_magic.lissero_serotype,
             sistr_results = merlin_magic.sistr_results,
             sistr_allele_json = merlin_magic.sistr_allele_json,
-            sister_allele_fasta = merlin_magic.sistr_allele_fasta,
+            sistr_allele_fasta = merlin_magic.sistr_allele_fasta,
             sistr_cgmlst = merlin_magic.sistr_cgmlst,
             sistr_version = merlin_magic.sistr_version,
             sistr_predicted_serotype = merlin_magic.sistr_predicted_serotype,
@@ -689,7 +689,7 @@ workflow theiaprok_illumina_se {
     # Salmonella Typing
     File? sistr_results = merlin_magic.sistr_results
     File? sistr_allele_json = merlin_magic.sistr_allele_json
-    File? sister_allele_fasta = merlin_magic.sistr_allele_fasta
+    File? sistr_allele_fasta = merlin_magic.sistr_allele_fasta
     File? sistr_cgmlst = merlin_magic.sistr_cgmlst
     String? sistr_version = merlin_magic.sistr_version
     String? sistr_predicted_serotype = merlin_magic.sistr_predicted_serotype

From 53d0683248ee382cb82d33644fb10f0d8178cbeb Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Tue, 19 Sep 2023 17:26:25 +0000
Subject: [PATCH 20/24] update CI

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

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
index ae62062bb..0a21761c7 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -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: 3c45d7016af73f22e16984c0e6693662
+      md5sum: 49b238cfa9148f6b3e05b11287489277
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
-      md5sum: 43b2512c04abd5bd2907c2e9f59666dd
+      md5sum: 1e243db7cc93ab641e46c369f9b512c4
     - 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 145eb3705..8fe6c9926 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: 3c45d7016af73f22e16984c0e6693662
+      md5sum: 49b238cfa9148f6b3e05b11287489277
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
-      md5sum: 97d84dae28752f6ff49307ae6962b7f2
+      md5sum: 764fa10ded5db9358d683d513390c934
     - 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 05c2f960e7d9b68a622c2fabb8aef0a565c9b69e Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Fri, 22 Sep 2023 11:39:16 +0000
Subject: [PATCH 21/24] add kmerfinder_template_coverage to theiaprok output

---
 tasks/taxon_id/task_kmerfinder.wdl               | 5 +++++
 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 ++
 6 files changed, 15 insertions(+)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 1b0265d37..827025b49 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -28,23 +28,28 @@ task kmerfinder_bacteria {
     if [ ! -f ~{samplename}/results.txt ]; then
       PF="No hit detected in database"
       QC="No hit detected in database"
+      TC="No hit detected in database"
     else
       PF="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 19)"
       QC="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 6)"
+      TC="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 7)"
         if [ "$PF" == "" ]; then
           PF="No hit detected in database"
           QC="No hit detected in database"
+          TC="No hit detected in database"
         fi
       mv -v ~{samplename}/results.txt ~{samplename}_kmerfinder.tsv
     fi
     echo $PF | tee TOP_HIT
     echo $QC | tee QC_METRIC
+    echo $TC | tee TEMPLATE_COVERAGE
   >>>
   output {
     String kmerfinder_docker = docker
     File? kmerfinder_results_tsv = "~{samplename}_kmerfinder.tsv"
     String kmerfinder_top_hit = read_string("TOP_HIT")
     String kmerfinder_query_coverage = read_string("QC_METRIC")
+    String kmerfinder_template_coverage = read_string("TEMPLATE_COVERAGE")
   }
   runtime {
     docker: docker
diff --git a/tasks/utilities/task_broad_terra_tools.wdl b/tasks/utilities/task_broad_terra_tools.wdl
index 01bd5095c..50e3c23e7 100644
--- a/tasks/utilities/task_broad_terra_tools.wdl
+++ b/tasks/utilities/task_broad_terra_tools.wdl
@@ -96,6 +96,7 @@ task export_taxon_tables {
     File? kmerfinder_results_tsv
     String? kmerfinder_top_hit
     String? kmerfinder_query_coverage
+    String? kmerfinder_template_coverage
     File? amrfinderplus_all_report
     File? amrfinderplus_amr_report
     File? amrfinderplus_stress_report
@@ -595,6 +596,7 @@ task export_taxon_tables {
       "kmerfinder_results_tsv": "~{kmerfinder_results_tsv}",
       "kmerfinder_top_hit": "~{kmerfinder_top_hit}",
       "kmerfinder_query_coverage": "~{kmerfinder_query_coverage}",
+      "kmerfinder_template_coverage": "~{kmerfinder_template_coverage}",
       "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 d4ddd54af..185e158b7 100644
--- a/workflows/theiaprok/wf_theiaprok_fasta.wdl
+++ b/workflows/theiaprok/wf_theiaprok_fasta.wdl
@@ -179,6 +179,7 @@ workflow theiaprok_fasta {
         kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
         kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
         kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
+        kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
         amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
         amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
         amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -429,6 +430,7 @@ workflow theiaprok_fasta {
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
+    String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
     # 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 c917d6ed2..e347ea403 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -304,6 +304,7 @@ workflow theiaprok_illumina_pe {
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
+            kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -623,6 +624,7 @@ workflow theiaprok_illumina_pe {
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
+    String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
     # 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 b4822d18d..de1f82abe 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
@@ -277,6 +277,7 @@ workflow theiaprok_illumina_se {
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
+            kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -575,6 +576,7 @@ workflow theiaprok_illumina_se {
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
+    String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
     # 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 043d6a98e..d9c5527d3 100644
--- a/workflows/theiaprok/wf_theiaprok_ont.wdl
+++ b/workflows/theiaprok/wf_theiaprok_ont.wdl
@@ -275,6 +275,7 @@ workflow theiaprok_ont {
             kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv,
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
+            kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
             amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus.amrfinderplus_stress_report,
@@ -549,6 +550,7 @@ workflow theiaprok_ont {
     File? kmerfinder_results_tsv = kmerfinder.kmerfinder_results_tsv
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
+    String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report

From 65ed7cfd23c8fca487e0b2db19a0afbb31f61d7d Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Thu, 28 Sep 2023 10:01:54 +0000
Subject: [PATCH 22/24] expose database name on kmerfinder outputs

---
 tasks/taxon_id/task_kmerfinder.wdl               | 7 ++++++-
 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 ++
 6 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 827025b49..75df08df0 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -4,7 +4,7 @@ task kmerfinder_bacteria {
   input {
     File assembly
     String samplename
-    File kmerfinder_db = "gs://theiagen-public-files-rp/terra/theiaprok-files/kmerfinder_bacteria.tar.gz"
+    File kmerfinder_db = "gs://theiagen-public-files-rp/terra/theiaprok-files/kmerfinder_bacteria_20230911.tar.gz"
     String docker = "us-docker.pkg.dev/general-theiagen/biocontainers/kmerfinder:3.0.2--hdfd78af_0"
     Int memory = 32
     Int cpu = 4
@@ -43,6 +43,10 @@ task kmerfinder_bacteria {
     echo $PF | tee TOP_HIT
     echo $QC | tee QC_METRIC
     echo $TC | tee TEMPLATE_COVERAGE
+
+    # extract database name
+    DB=$(basename ~{kmerfinder_db} | sed 's/\.tar\.gz$//')
+    echo $DB | tee DATABASE
   >>>
   output {
     String kmerfinder_docker = docker
@@ -50,6 +54,7 @@ task kmerfinder_bacteria {
     String kmerfinder_top_hit = read_string("TOP_HIT")
     String kmerfinder_query_coverage = read_string("QC_METRIC")
     String kmerfinder_template_coverage = read_string("TEMPLATE_COVERAGE")
+    String kmerfinder_database = read_string("DATABASE")
   }
   runtime {
     docker: docker
diff --git a/tasks/utilities/task_broad_terra_tools.wdl b/tasks/utilities/task_broad_terra_tools.wdl
index 50e3c23e7..0f430a08e 100644
--- a/tasks/utilities/task_broad_terra_tools.wdl
+++ b/tasks/utilities/task_broad_terra_tools.wdl
@@ -97,6 +97,7 @@ task export_taxon_tables {
     String? kmerfinder_top_hit
     String? kmerfinder_query_coverage
     String? kmerfinder_template_coverage
+    String? kmerfinder_database
     File? amrfinderplus_all_report
     File? amrfinderplus_amr_report
     File? amrfinderplus_stress_report
@@ -597,6 +598,7 @@ task export_taxon_tables {
       "kmerfinder_top_hit": "~{kmerfinder_top_hit}",
       "kmerfinder_query_coverage": "~{kmerfinder_query_coverage}",
       "kmerfinder_template_coverage": "~{kmerfinder_template_coverage}",
+      "kmerfinder_database": "~{kmerfinder_database}",
       "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 185e158b7..e26894185 100644
--- a/workflows/theiaprok/wf_theiaprok_fasta.wdl
+++ b/workflows/theiaprok/wf_theiaprok_fasta.wdl
@@ -180,6 +180,7 @@ workflow theiaprok_fasta {
         kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
         kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
         kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
+        kmerfinder_database = kmerfinder.kmerfinder_database,
         amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
         amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
         amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -431,6 +432,7 @@ workflow theiaprok_fasta {
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
+    String? kmerfinder_database = kmerfinder.kmerfinder_database
     # 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 e347ea403..9565bf6e3 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
@@ -305,6 +305,7 @@ workflow theiaprok_illumina_pe {
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
+            kmerfinder_database = kmerfinder.kmerfinder_database,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -625,6 +626,7 @@ workflow theiaprok_illumina_pe {
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
+    String? kmerfinder_database = kmerfinder.kmerfinder_database
     # 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 de1f82abe..d806c6ea4 100644
--- a/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
+++ b/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
@@ -278,6 +278,7 @@ workflow theiaprok_illumina_se {
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
+            kmerfinder_database = kmerfinder.kmerfinder_database,
             amrfinderplus_all_report = amrfinderplus_task.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus_task.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus_task.amrfinderplus_stress_report,
@@ -577,6 +578,7 @@ workflow theiaprok_illumina_se {
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
+    String? kmerfinder_database = kmerfinder.kmerfinder_database
     # 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 d9c5527d3..b81560080 100644
--- a/workflows/theiaprok/wf_theiaprok_ont.wdl
+++ b/workflows/theiaprok/wf_theiaprok_ont.wdl
@@ -276,6 +276,7 @@ workflow theiaprok_ont {
             kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit,
             kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage,
             kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage,
+            kmerfinder_database = kmerfinder.kmerfinder_database,
             amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report,
             amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report,
             amrfinderplus_stress_report = amrfinderplus.amrfinderplus_stress_report,
@@ -551,6 +552,7 @@ workflow theiaprok_ont {
     String? kmerfinder_top_hit = kmerfinder.kmerfinder_top_hit
     String? kmerfinder_query_coverage = kmerfinder.kmerfinder_query_coverage
     String? kmerfinder_template_coverage = kmerfinder.kmerfinder_template_coverage
+    String? kmerfinder_database = kmerfinder.kmerfinder_database
     # NCBI-AMRFinderPlus Outputs
     File? amrfinderplus_all_report = amrfinderplus.amrfinderplus_all_report
     File? amrfinderplus_amr_report = amrfinderplus.amrfinderplus_amr_report

From 3c4bbfabd561f85409f78271a06d225fae9897d5 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Thu, 28 Sep 2023 10:23:46 +0000
Subject: [PATCH 23/24] update md5sum

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

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
index 7883ece6c..198209aa4 100644
--- a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
+++ b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -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: 49b238cfa9148f6b3e05b11287489277
+      md5sum: 43ef050bde1fb8755f38e697a1794918
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_pe.wdl
-      md5sum: 1e243db7cc93ab641e46c369f9b512c4
+      md5sum: 6dc6c393281a19e8dcbcc15964b8e08a
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: c4861a59d49b13b67706631a0e1246c4
     - 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 31e42ae47..9aebb81ba 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: 49b238cfa9148f6b3e05b11287489277
+      md5sum: 43ef050bde1fb8755f38e697a1794918
     - path: miniwdl_run/wdl/workflows/theiaprok/wf_theiaprok_illumina_se.wdl
-      md5sum: 764fa10ded5db9358d683d513390c934
+      md5sum: 24bfd35867f4ae864364e24195bf7f6f
     - path: miniwdl_run/wdl/workflows/utilities/wf_merlin_magic.wdl
       md5sum: c4861a59d49b13b67706631a0e1246c4
     - path: miniwdl_run/wdl/workflows/utilities/wf_read_QC_trim_se.wdl

From da98fa63652c59e1f23a0d00ecee6df57f586f84 Mon Sep 17 00:00:00 2001
From: cimendes <iines.mendes@gmail.com>
Date: Mon, 2 Oct 2023 11:10:32 +0000
Subject: [PATCH 24/24] fix bug - headers being outputed to datatable

---
 tasks/taxon_id/task_kmerfinder.wdl | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tasks/taxon_id/task_kmerfinder.wdl b/tasks/taxon_id/task_kmerfinder.wdl
index 75df08df0..61b3bed78 100644
--- a/tasks/taxon_id/task_kmerfinder.wdl
+++ b/tasks/taxon_id/task_kmerfinder.wdl
@@ -33,7 +33,8 @@ task kmerfinder_bacteria {
       PF="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 19)"
       QC="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 6)"
       TC="$(cat ~{samplename}/results.txt | head -n 2 | tail -n 1 | cut -f 7)"
-        if [ "$PF" == "" ]; then
+        # String is empty or just contains the header
+        if [ "$PF" == "" ] || [ "$PF" == "Species" ]; then
           PF="No hit detected in database"
           QC="No hit detected in database"
           TC="No hit detected in database"