[New Workflow] Flye_denovo to replace DragonFlye

tasks/alignment/task_bwa.wdl

@@ -12,6 +12,7 @@ task bwa {
     String docker = "us-docker.pkg.dev/general-theiagen/staphb/ivar:1.3.1-titan"
   }
   command <<<
+    set -euo pipefail
     # date and version control
     date | tee DATE
     echo "BWA $(bwa 2>&1 | grep Version )" | tee BWA_VERSION
@@ -153,4 +154,49 @@ task bwa {
     preemptible: 0
     maxRetries: 3
   }
-}
+}
+
+task bwa_all {
+  input {
+    File draft_assembly_fasta
+    File read1
+    File read2
+    String samplename
+
+    Int cpu = 6
+    Int disk_size = 100
+    String docker = "us-docker.pkg.dev/general-theiagen/staphb/bwa:0.7.18"
+    Int memory = 16
+  }
+  command <<<
+    set -euo pipefail
+
+    bwa &> BWA_HELP
+    grep "Version" BWA_HELP | cut -d" " -f2 > BWA_VERSION
+
+    if [[ ! -f "~{draft_assembly_fasta}.bwt" ]]; then
+      echo "Indexing reference genome: ~{draft_assembly_fasta}"
+      bwa index ~{draft_assembly_fasta}
+    else
+      echo "Reference genome is already indexed: ~{draft_assembly_fasta}"
+    fi
+
+    bwa mem -t ~{cpu} -a ~{draft_assembly_fasta} ~{read1} > ~{samplename}_R1.sam
+    bwa mem -t ~{cpu} -a ~{draft_assembly_fasta} ~{read2} > ~{samplename}_R2.sam
+
+  >>>
+  output {
+    File read1_sam = "~{samplename}_R1.sam"
+    File read2_sam = "~{samplename}_R2.sam"
+    String bwa_version = read_string("BWA_VERSION")
+  }
+  runtime {
+    docker: "~{docker}"
+    memory: "~{memory} GB"
+    cpu: "~{cpu}"
+    disks:  "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
+    preemptible: 0
+  }
+}

tasks/assembly/task_bandage_plot.wdl

@@ -0,0 +1,30 @@
+version 1.0
+
+task bandage_plot {
+    input {
+      File assembly_graph_gfa
+      String samplename
+      Int cpu = 1
+      Int memory = 4
+      Int disk_size = 10
+      String docker = "us-docker.pkg.dev/general-theiagen/staphb/bandage:0.8.1"
+    }
+    command <<< 
+      set -euo pipefail
+      Bandage --version | tee VERSION
+      Bandage image ~{assembly_graph_gfa} ~{samplename}_bandage_plot.png
+    >>>
+    output {
+      File plot = "~{samplename}_bandage_plot.png"
+      String bandage_version = read_string("VERSION")
+    }
+    runtime {
+      docker: "~{docker}"
+      cpu: cpu
+      memory: "~{memory} GB"
+      disks: "local-disk " + disk_size + " HDD"
+      disk: disk_size + " GB"
+      maxRetries: 1
+      preemptible: 0
+  }
+}

tasks/assembly/task_dnaapler.wdl

@@ -0,0 +1,63 @@
+version 1.0
+
+task dnaapler {
+  input {
+    File input_fasta
+    String samplename
+    String dnaapler_mode = "all" # The mode of reorientation to execute (default: 'all')
+    Int cpu = 4
+    Int disk_size = 100
+    Int memory = 16
+    String docker = "us-docker.pkg.dev/general-theiagen/staphb/dnaapler:1.0.1"
+  }
+  command <<< 
+    set -euo pipefail
+
+    # dnaapler version
+    dnaapler --version | tee VERSION
+
+    # Create a subdirectory for dnaapler outputs
+    output_dir="dnaapler_output"
+    mkdir -p "$output_dir"
+    echo "Output directory created: $output_dir"
+
+    # Run dnaapler subcommand
+    echo "Running dnaapler..."
+    dnaapler ~{dnaapler_mode} \
+      -i ~{input_fasta} \
+      -o "$output_dir" \
+      -p ~{samplename} \
+      -t ~{cpu} \
+      -f || {
+        echo "ERROR: dnaapler command failed. Check logs for details." >&2
+        exit 1
+      }
+
+    echo "dnaapler command completed successfully."
+
+    # Check if output FASTA file exists
+    if [[ ! -f "$output_dir"/~{samplename}_reoriented.fasta ]]; then
+      echo "ERROR: Expected output file not found: $output_dir/~{samplename}_reoriented.fasta" >&2
+      exit 1
+    fi
+
+    # Move the final reoriented FASTA file to the task's working directory
+    echo "Moving output FASTA file to working directory..."
+    mv "$output_dir"/~{samplename}_reoriented.fasta .
+
+    echo "dnaapler task completed successfully for sample: ~{samplename}"
+  >>>
+  output {
+    File reoriented_fasta = "~{samplename}_reoriented.fasta"
+    String dnaapler_version = read_string("VERSION")
+  }
+  runtime {
+    docker: "~{docker}"
+    cpu: cpu
+    memory: "~{memory} GB"
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
+    preemptible: 0
+  }
+}

tasks/assembly/task_flye.wdl

@@ -0,0 +1,68 @@
+version 1.0
+
+task flye {
+  input {
+    File read1
+    String samplename
+    String read_type = "--nano-hq" # Default read type
+    Int? genome_length # requires `asm_coverage`
+    Int? asm_coverage # reduced coverage for initial disjointig assembly
+
+    Int flye_polishing_iterations = 1
+    Int? minimum_overlap
+
+    Float? read_error_rate
+    Boolean uneven_coverage_mode = false
+    Boolean keep_haplotypes = false
+    Boolean no_alt_contigs = false
+    Boolean scaffold = false
+
+    String? additional_parameters # Any extra Flye-specific parameters
+
+    Int cpu = 4
+    Int disk_size = 100
+    String docker = "us-docker.pkg.dev/general-theiagen/staphb/flye:2.9.4"
+    Int memory = 32
+  }
+  command <<<
+    set -euo pipefail
+    flye --version | tee VERSION
+
+    # genome size parameter requires asm_coverage
+    flye \
+      "~{read_type}" "~{read1}" \
+      --iterations ~{flye_polishing_iterations} \
+      ~{"--min-overlap" + minimum_overlap} \
+      ~{if defined(asm_coverage) then "--genome-size " + genome_length else ""} \
+      ~{"--asm-coverage " + asm_coverage} \
+      ~{"--read-error " + read_error_rate} \
+      ~{true="--meta" false="" uneven_coverage_mode} \
+      ~{true="--keep-haplotypes" false="" keep_haplotypes} \
+      ~{true="--no-alt-contigs" false="" no_alt_contigs} \
+      ~{true="--scaffold" false="" scaffold} \
+      ~{"--extra-params " + additional_parameters } \
+      --threads ~{cpu} \
+      --out-dir .
+
+    mv assembly.fasta ~{samplename}.assembly.fasta
+    mv assembly_info.txt ~{samplename}.assembly_info.txt
+    mv assembly_graph.gfa ~{samplename}.assembly_graph.gfa
+
+  >>>
+  output {
+    File assembly_fasta = "~{samplename}.assembly.fasta"
+    File assembly_graph_gfa = "~{samplename}.assembly_graph.gfa" 
+    File assembly_info = "~{samplename}.assembly_info.txt" 
+    String flye_version = read_string("VERSION")
+    String flye_docker = "~{docker}"
+  }
+  runtime {
+    docker: "~{docker}"
+    cpu: cpu
+    memory: "~{memory} GB"
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
+    preemptible: 0
+  }
+}

tasks/polishing/task_medaka.wdl

@@ -0,0 +1,55 @@
+version 1.0
+
+task medaka {
+  input {
+    File unpolished_fasta
+    String samplename
+    File read1
+    Boolean auto_model = true # Enable automatic Medaka model selection
+    String medaka_model = "r1041_e82_400bps_sup_v5.0.0" # Default model if auto_model is disabled or no model is resolved
+    Int cpu = 4
+    Int memory = 16
+    Int disk_size = 100
+    String docker = "us-docker.pkg.dev/general-theiagen/staphb/medaka:2.0.1"
+  }
+  command <<<    
+    set -euo pipefail
+
+    medaka --version | tee MEDAKA_VERSION
+
+    # Attempt automatic model resolution if enabled
+    if [[ "~{auto_model}" == "true" ]]; then
+      echo "Attempting automatic model selection..."
+      medaka tools resolve_model --auto_model consensus ~{read1} > auto_model.txt || true
+      resolved_model=$(cat auto_model.txt || echo "")
+      medaka_model="${resolved_model:-~{medaka_model}}"
+    fi
+
+    echo "Using Medaka model for polishing: $medaka_model"
+    echo "$medaka_model" > MEDAKA_MODEL
+
+    # Perform Medaka polishing
+    medaka_consensus \
+      -i "~{read1}" \
+      -d "~{unpolished_fasta}" \
+      -o . \
+      -m "$medaka_model" \
+      -t "~{cpu}"
+
+    mv consensus.fasta ~{samplename}.polished.fasta
+  >>>
+  output {
+    File medaka_fasta = "~{samplename}.polished.fasta"
+    String medaka_version = read_string("MEDAKA_VERSION")
+    String resolved_medaka_model = read_string("MEDAKA_MODEL")
+  }
+  runtime {
+    docker: "~{docker}"
+    cpu: cpu
+    memory: "~{memory} GB"
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
+    preemptible: 0
+  }
+}