Merge pull request #2 from Kincekara/dev

v 1.1.0
Kincekara · Sep 7, 2023 · ff4155d · ff4155d
2 parents 511ccb6 + 2955502
commit ff4155d
Show file tree

Hide file tree

Showing 10 changed files with 141 additions and 62 deletions.
diff --git a/README.md b/README.md
@@ -38,17 +38,17 @@ The following programs and tools are used in the C-BIRD pipeline.
 
 | Tools | Version | Comments |
 | --- | --- | --- |
-| [FastP](https://github.com/OpenGene/fastp) | 0.23.2 | QC, adapter removal, quality filtering and trimming |
-| [BBTools](https://jgi.doe.gov/data-and-tools/software-tools/bbtools/) | 38.98 | phiX removal & optional normalization |
-| [Kraken2](https://github.com/DerrickWood/kraken2) | 2.1.2 | Taxonomic profiling & contamination check |
+| [FastP](https://github.com/OpenGene/fastp) | 0.23.4 | QC, adapter removal, quality filtering and trimming |
+| [BBTools](https://jgi.doe.gov/data-and-tools/software-tools/bbtools/) | 39.01 | phiX removal & optional normalization |
+| [Kraken2](https://github.com/DerrickWood/kraken2) | 2.1.3 | Taxonomic profiling & contamination check |
 | [Bracken](https://github.com/jenniferlu717/Bracken) | 2.8 | Abundance estimation |
 | [SPAdes](https://github.com/ablab/spades) | 3.15.5 | *De novo* assembly |
 | [Mash](https://github.com/marbl/Mash) | 2.3 | Bacterial identification |
-| [QUAST](https://github.com/ablab/quast) | 5.0.2 | Genome assembly evaluation |
-| [BUSCO](https://gitlab.com/ezlab/busco/-/tree/master) | 5.4.3 | Genomic data quality assesment |
+| [QUAST](https://github.com/ablab/quast) | 5.2.0 | Genome assembly evaluation |
+| [BUSCO](https://gitlab.com/ezlab/busco/-/tree/master) | 5.4.7 | Genomic data quality assesment |
 | [mlst](https://github.com/tseemann/mlst) | 2.22.0 | MLST typing |
-| [AMRFinderPlus](https://github.com/ncbi/amr) | 3.10.40 | AMR gene identification |
-| [BLAST+](https://blast.ncbi.nlm.nih.gov/doc/blast-help/downloadblastdata.html)| 2.13.0 | Target gene search |
+| [AMRFinderPlus](https://github.com/ncbi/amr) | 3.11.18 | AMR gene identification |
+| [BLAST+](https://blast.ncbi.nlm.nih.gov/doc/blast-help/downloadblastdata.html)| 2.14.0 | Target gene search |
 | [PlasmidFinder](https://bitbucket.org/genomicepidemiology/plasmidfinder/src/master/) | 2.1.6 | Plasmid detection |
 | Cbird-Util | 0.8 | Individual summary report generation |
 

diff --git a/tasks/task_amrfinderplus.wdl b/tasks/task_amrfinderplus.wdl
@@ -4,15 +4,13 @@ task amrfinderplus_nuc {
   input {
     File assembly
     String samplename
-    File amr_db
-    # Parameters 
-    # --indent_min Minimum DNA %identity [0-1]; default is 0.9 (90%) or curated threshold if it exists
-    # --mincov Minimum DNA %coverage [0-1]; default is 0.5 (50%)
-    String? organism # make optional?
+    File amr_db     
+    String bracken_organism
+    String? mash_organism
     Float? minid
     Float? mincov
     Int cpu = 4
-    String docker = "kincekara/amrfinder:3.10.40"
+    String docker = "kincekara/amrfinder:3.11.18"
   }
   command <<<
     # logging info
@@ -21,49 +19,129 @@ task amrfinderplus_nuc {
 
     mkdir db
     tar -C ./db/ -xzvf ~{amr_db}
-
-    ### set $amrfinder_organism BASH variable based on gambit_predicted_taxon or user-defined input string
-    ### final variable has strict syntax/spelling based on list from amrfinder --list_organisms
-    # there may be other Acinetobacter species to add later, like those in the A. baumannii-calcoaceticus species complex
-    if [[ "~{organism}" == *"Acinetobacter"*"baumannii"* ]]; then
-      amrfinder_organism="Acinetobacter_baumannii"
-    elif [[ "~{organism}" == *"Campylobacter"*"coli"* ]] || [[ "~{organism}" == *"Campylobacter"*"jejuni"* ]]; then
-      amrfinder_organism="Campylobacter"
-    elif [[ "~{organism}" == *"Clostridioides"*"difficile"* ]]; then
-      amrfinder_organism="Clostridioides_difficile"
-    elif [[ "~{organism}" == *"Enterococcus"*"faecalis"* ]]; then 
-      amrfinder_organism="Enterococcus_faecalis"
-    elif [[ "~{organism}" == *"Enterococcus"*"faecium"* ]] || [[ "~{organism}" == *"Enterococcus"*"hirae"* ]]; then 
-      amrfinder_organism="Enterococcus_faecium"
-    # should capture all Shigella and Escherichia species
-    elif [[ "~{organism}" == *"Escherichia"* ]] || [[ "~{organism}" == *"Shigella"* ]]; then 
+
+    # select mash organism if avalible
+    if [[ "~{mash_organism}" != "" ]]; then
+      organism="~{mash_organism}"
+    else
+      organism="~{bracken_organism}"
+    fi
+    echo "organism is set to:" $organism
+
+    ## curated organisms ##
+    # A. baumannii-calcoaceticus species complex
+    declare -a abcc=(
+      "Acinetobacter baumannii"
+      "Acinetobacter calcoaceticus"
+      "Acinetobacter lactucae"
+      "Acinetobacter nosocomialis"
+      "Acinetobacter pittii"
+      "Acinetobacter seifertii"
+    )
+    # Burkholderia cepacia species complex
+    declare -a bcc=(
+      "Burkholderia aenigmatica"
+      "Burkholderia ambifaria"   
+      "Burkholderia anthina"   
+      "Burkholderia arboris"   
+      "Burkholderia catarinensis"   
+      "Burkholderia cenocepacia"   
+      "Burkholderia cepacia" 
+      "Burkholderia cf. cepacia"  
+      "Burkholderia contaminans"   
+      "Burkholderia diffusa"   
+      "Burkholderia dolosa"   
+      "Burkholderia lata"   
+      "Burkholderia latens"
+      "Burkholderia metallica"  
+      "Burkholderia multivorans"   
+      "Burkholderia orbicola"   
+      "Burkholderia paludis"   
+      "Burkholderia pseudomultivorans"   
+      "Burkholderia puraquae"   
+      "Burkholderia pyrrocinia"   
+      "Burkholderia semiarida"   
+      "Burkholderia seminalis"   
+      "Burkholderia sola"   
+      "Burkholderia stabilis"   
+      "Burkholderia stagnalis"   
+      "Burkholderia territorii"   
+      "Burkholderia ubonensis"   
+      "Burkholderia vietnamiensis" 
+    )
+    # Burkholderia pseudomallei species complex
+    declare -a bpc=(
+      "Burkholderia humptydooensis"   
+      "Burkholderia mallei"   
+      "Burkholderia mayonis"   
+      "Burkholderia oklahomensis"   
+      "Burkholderia pseudomallei"   
+      "Burkholderia savannae"   
+      "Burkholderia singularis"   
+      "Burkholderia thailandensis"   
+    )
+    # other species
+    declare -a taxa=(   
+      "Citrobacter freundii"
+      "Clostridioides difficile"
+      "Enterobacter asburiae"
+      "Enterobacter cloacae"
+      "Enterococcus faecalis"    
+      "Klebsiella oxytoca"
+      "Neisseria meningitidis"
+      "Neisseria gonorrhoeae"
+      "Pseudomonas aeruginosa" 
+      "Serratia marcescens"  
+      "Staphylococcus aureus"
+      "Staphylococcus pseudintermedius"
+      "Streptococcus agalactiae"
+      "Streptococcus pyogenes"
+      "Vibrio cholerae"
+    )
+
+    # check organism in curated organism list
+    genus=$(echo $organism | cut -d " " -f1)
+    taxon=$(echo $organism | cut -d " " -f1,2)
+
+    if [[ "$genus" == "Acinetobacter" ]]; then
+      for i in "${abcc[@]}"; do
+        if [[ "$taxon" == "$i" ]]; then
+          amrfinder_organism="Acinetobacter_baumannii"
+          break
+        fi
+      done
+    elif [[ "$genus" == "Burkholderia" ]]; then
+      for i in "${bcc[@]}"; do
+        if [[ "$taxon" == "$i" ]]; then
+          amrfinder_organism="Burkholderia_cepacia"
+          break
+        fi
+      done
+      for i in "${bpc[@]}"; do
+        if [[ "$taxon" == "$i" ]]; then
+          amrfinder_organism="Burkholderia_pseudomallei"
+          break
+        fi
+      done
+    elif [[ "$genus" == "Shigella" ]] || [[ "$genus" == "Escherichia" ]]; then
       amrfinder_organism="Escherichia"
-    # add other Klebsiella species later? Cannot use K. oxytoca as per amrfinderplus wiki
-    elif [[ "~{organism}" == *"Klebsiella"*"aerogenes"* ]] || [[ "~{organism}" == *"Klebsiella"*"pneumoniae"* ]]; then 
-      amrfinder_organism="Klebsiella"
-    # because some people spell the species 'gonorrhea' differently
-    elif [[ "~{organism}" == *"Neisseria"*"gonorrhea"* ]] || [[ "~{organism}" == *"Neisseria"*"gonorrhoeae"* ]] || [[ "~{organism}" == *"Neisseria"*"meningitidis"* ]]; then 
-      amrfinder_organism="Neisseria"
-    elif [[ "~{organism}" == *"Pseudomonas"*"aeruginosa"* ]]; then 
-      amrfinder_organism="Pseudomonas_aeruginosa"
-    # pretty broad, could work on Salmonella bongori and other species
-    elif [[ "~{organism}" == *"Salmonella"* ]]; then 
+    elif [[ "$genus" == "Salmonella" ]]; then
       amrfinder_organism="Salmonella"
-    elif [[ "~{organism}" == *"Staphylococcus"*"aureus"* ]]; then 
-      amrfinder_organism="Staphylococcus_aureus"
-    elif [[ "~{organism}" == *"Staphylococcus"*"pseudintermedius"* ]]; then 
-      amrfinder_organism="Staphylococcus_pseudintermedius"
-    elif [[ "~{organism}" == *"Streptococcus"*"agalactiae"* ]]; then 
-      amrfinder_organism="Streptococcus_agalactiae"
-    elif [[ "~{organism}" == *"Streptococcus"*"pneumoniae"* ]] || [[ "~{organism}" == *"Streptococcus"*"mitis"* ]]; then 
+    elif [[ "$taxon" == "Campylobacter coli" ]] || [[ "$taxon" == "Campylobacter jejuni" ]]; then
+      amrfinder_organism="Campylobacter"
+    elif [[ "$taxon" == "Enterococcus faecium" ]] || [[ "$taxon" == "Enterococcus hirae" ]]; then
+      amrfinder_organism="Enterococcus_faecium"
+    elif [[ "$taxon" == "Klebsiella pneumoniae" ]] || [[ "$taxon" == "Klebsiella aerogenes" ]]; then
+      amrfinder_organism="Klebsiella_pneumoniae"
+    elif [[ "$taxon" == "Streptococcus pneumoniae" ]] || [[ "$taxon" == "Streptococcus mitis" ]]; then
       amrfinder_organism="Streptococcus_pneumoniae"
-    elif [[ "~{organism}" == *"Streptococcus"*"pyogenes"* ]]; then 
-      amrfinder_organism="Streptococcus_pyogenes"
-    elif [[ "~{organism}" == *"Vibrio"*"cholerae"* ]]; then 
-      amrfinder_organism="Vibrio_cholerae"
-    else 
-      echo "Either Bracken predicted taxon is not supported by NCBI-AMRFinderPlus or the user did not supply an organism as input."
-      echo "Skipping the use of amrfinder --organism optional parameter."
+    else    
+      for i in "${taxa[@]}"; do
+        if [[ "$taxon" == "$i" ]]; then
+          amrfinder_organism=${taxon// /_}
+          break
+        fi
+      done
     fi
 
     # checking bash variable
@@ -75,7 +153,7 @@ task amrfinderplus_nuc {
       # send STDOUT/ERR to log file for capturing database version
       amrfinder --plus \
         -d ./db/ \
-        --organism ${amrfinder_organism} \
+        --organism "${amrfinder_organism}" \
         ~{'--name ' + samplename} \
         ~{'--nucleotide ' + assembly} \
         ~{'-o ' + samplename + '_amrfinder_all.tsv'} \

diff --git a/tasks/task_bbtools.wdl b/tasks/task_bbtools.wdl
@@ -11,7 +11,7 @@ task assembly_prep {
     Int min_depth = 5
     Int read_threshold = 8000000
     Int memory = 8
-    String docker = "kincekara/bbduk:38.98"
+    String docker = "kincekara/bbtools:39.01"
   }
 
   command <<<

diff --git a/tasks/task_blast.wdl b/tasks/task_blast.wdl
@@ -8,7 +8,7 @@ task tblastn {
     Float evalue = 0.001
     Int max_hsps = 1
     Int percent_identity = 90
-    String docker = "staphb/blast:2.13.0"
+    String docker = "staphb/blast:2.14.0"
   }
 
   command <<<

diff --git a/tasks/task_busco.wdl b/tasks/task_busco.wdl
@@ -5,7 +5,7 @@ task busco {
     File assembly   
     File busco_db
     String samplename
-    String docker = "kincekara/busco:5.4.3"
+    String docker = "kincekara/busco:5.4.7"
     Int? memory = 16
     Int? cpu = 4
   }

diff --git a/tasks/task_fastp.wdl b/tasks/task_fastp.wdl
@@ -5,7 +5,7 @@ task fastp_pe {
     File read1
     File read2
     File? adapters
-    String docker = "kincekara/fastp:0.23.2"
+    String docker = "kincekara/fastp:0.23.4"
     String samplename
     Int? leading = 1
     Int? front_mean_quality = 10

diff --git a/tasks/task_quast.wdl b/tasks/task_quast.wdl
@@ -4,7 +4,7 @@ task quast {
   input {
     File assembly
     String samplename
-    String docker= "staphb/quast:5.0.2"
+    String docker= "kincekara/quast:5.2.0-lite"
   }
   command <<<
     # version

diff --git a/tasks/task_taxonomy.wdl b/tasks/task_taxonomy.wdl
@@ -6,7 +6,7 @@ task profile {
     File read2
     File kraken2_db
     String samplename
-    String docker = "kincekara/kraken-bracken:k2.1.2-b2.8"
+    String docker = "kincekara/kraken-bracken:k2.1.3-b2.8"
     Int? bracken_read_len = 100
     Int? bracken_threshold = 10
     String? min_hit_groups = 3

diff --git a/tasks/task_version.wdl b/tasks/task_version.wdl
@@ -8,7 +8,7 @@ task version_capture {
     volatile: true
   }
   command {
-    cbird_version="C-BIRD v1.0.0"
+    cbird_version="C-BIRD v1.1.0"
     ~{default='' 'export TZ=' + timezone}
     date +"%Y-%m-%d" > TODAY
     echo "$cbird_version" > CBIRD_VERSION

diff --git a/workflows/wf_c-bird.wdl b/workflows/wf_c-bird.wdl
@@ -119,7 +119,8 @@ workflow cbird_workflow {
       samplename = samplename,
       assembly = assembly.scaffolds_trim,
       amr_db = amrfinder_database,
-      organism = profile.bracken_taxon
+      bracken_organism = profile.bracken_taxon,
+      mash_organism = predict_taxon.taxon
     }
 
     call plasmid.plasmidfinder {