Wastewater Lineage De-convolution (#99)

* Add a titan wastewater variant calling workkflow

* Fix formatting

* Fix formatting (for real this time)

* adjust default minlength

* adjust trimmomatic params

* Set trimmmomatic minlen at workflow level

* Remove unused imports

* Add version capture task

* add meta block

* Update README.md

* Update README.md

* create titan_freyja workflow

* ensure executables are in path

* allow raw read inputs

* clean whitespace

* remove whitespace

* rename workflow and set header as samplename

* specify output file extension

* update versioning

* reset to jlevy docker for now

* small tweak

* update to staphb image

* allow user defined barcodes

* correct image tag

* revert to jlevy image

* add freyja plot workflow

* update to staphb freyja image

* add freyja plot workflow

* update freyja header for plot indexing

* adjust freyja commands

* fix interval flag

* fix plot options

* fix output variables

* adjust samplename array

* Update wf_freyja_plot.wdl

* add reference flag

* change default setting to plot summary

* Update wf_freyja_plot.wdl

* allow user-defined reference

* reference genome is optional

* close if statement

* fix task syntax

* allow user reference in ont workflows

* fix wdl syntax

* set user ref appropriately

* Output freyja demixed aggregate file

* fix collection array

* really fix collection array

* Update README.md

* Pull dockers from quay repo

* update summaries to include new outputs

* undo last commit -- wrong branch!

* allow for use of epi2me docker

* add medaka reference output

* fix typo

* modify bed dest path

* fix path

* Fix typo

* Update image diest

* comma

* capture reference from either artic or epi2me paths

* fix typo

* remove extra space

* add wildcard to ref path

* check for directory

* small tweak

* force zip

* small tweak 2

* add min/max length workflow vars

* Update test_clearlabs.yml

* Update test_ont.yml

* update to fastqscan

* Update test_ont.yml

* Update test_clearlabs.yml

* Update test_ont.yml

* check pass.vcf

* Default to freyja update_db

* Fix yml

* really fix yaml

.dockstore.yml

@@ -70,6 +70,21 @@ workflows:
    primaryDescriptorPath: /workflows/wf_titan_fasta.wdl
    testParameterFiles:
     - empty.json
+ - name: Titan_WWVC
+   subclass: WDL
+   primaryDescriptorPath: /workflows/wf_titan_wwvc.wdl
+   testParameterFiles:
+    - empty.json
+ - name: Freyja_FASTQ
+   subclass: WDL
+   primaryDescriptorPath: /workflows/wf_freyja_fastq.wdl
+   testParameterFiles:
+    - empty.json
+ - name: Freyja_Plot
+   subclass: WDL
+   primaryDescriptorPath: /workflows/wf_freyja_plot.wdl
+   testParameterFiles:
+    - empty.json
  - name: TheiaCoV_Validate
    subclass: WDL
    primaryDescriptorPath: /workflows/wf_theiacov_validate.wdl

README.md

@@ -4,4 +4,5 @@ Bioinformatics workflows for genomic characterization, submission preparation, a
 ### Contributors & Influence
 * Based on collaborative work with Andrew Lang, PhD & his [Genomic Analysis WDL workflows](https://github.com/AndrewLangvt/genomic_analyses)
 * Workflows and task development influenced by The Broad's [Viral Pipes](https://github.com/broadinstitute/viral-pipelines)
-* Titan Genomic Characterization workflows influenced by UPHL's [Cecret](https://github.com/UPHL-BioNGS/Cecret) & StaPH-B's [Monroe](https://staph-b.github.io/staphb_toolkit/workflow_docs/monroe/)
+* Titan workflows for genomic characterization influenced by UPHL's [Cecret](https://github.com/UPHL-BioNGS/Cecret) & StaPH-B's [Monroe](https://staph-b.github.io/staphb_toolkit/workflow_docs/monroe/)
+* The Titan workflow for waste water variant calling (Titan_WWVC) incorporates a modified version of the [CDPHE's WasteWaterVariantCalling WDL Worfklow](https://github.com/CDPHE/WasteWaterVariantCalling).

tasks/task_alignment.wdl

@@ -6,26 +6,34 @@ task bwa {
     File        read1
     File?        read2
     String      samplename
-    String?     reference_genome="/artic-ncov2019/primer_schemes/nCoV-2019/V3/nCoV-2019.reference.fasta"
+    File?     reference_genome
     Int?        cpus=6
   }
 
-  command {
+  command <<<
     # date and version control
     date | tee DATE
     echo "BWA $(bwa 2>&1 | grep Version )" | tee BWA_VERSION
     samtools --version | head -n1 | tee SAMTOOLS_VERSION
+
+    # set reference genome
+    if [[ ! -z "~{reference_genome}" ]]; then
+      echo "User reference identified; ~{reference_genome} will be utilized for alignement"
+      # move to primer_schemes dir; bwa fails if reference file not in this location
+      cp "~{reference_genome}" "/artic-ncov2019/primer_schemes/nCoV-2019/V3/nCoV-2019.reference.fasta"  
+    fi
 
     # Map with BWA MEM
+    echo "Running bwa mem -t ~{cpus} bwa_reference.bwa ~{read1} ~{read2} | samtools sort | samtools view -F 4 -o ~{samplename}.sorted.bam "
     bwa mem \
-    -t ${cpus} \
-    ${reference_genome} \
-    ${read1} ${read2} |\
-    samtools sort | samtools view -F 4 -o ${samplename}.sorted.bam
+    -t ~{cpus} \
+    "/artic-ncov2019/primer_schemes/nCoV-2019/V3/nCoV-2019.reference.fasta" \
+    ~{read1} ~{read2} |\
+    samtools sort | samtools view -F 4 -o ~{samplename}.sorted.bam
 
     # index BAMs
-    samtools index ${samplename}.sorted.bam
-  }
+    samtools index ~{samplename}.sorted.bam
+  >>>
 
   output {
     String     bwa_version = read_string("BWA_VERSION")
@@ -40,7 +48,6 @@ task bwa {
     cpu:          2
     disks:        "local-disk 100 SSD"
     preemptible:  0
-    maxRetries:   3
   }
 }
 

tasks/task_taxonID.wdl

@@ -309,3 +309,62 @@ task nextclade_output_parser_one_sample {
         String nextclade_aa_dels  = read_string("NEXTCLADE_AADELS")
     }
 }
+
+task freyja_one_sample {
+  input {
+    File primer_trimmed_bam
+    String samplename
+    File reference_genome
+    File? freyja_usher_barcodes
+    Boolean update_db = true
+    String docker = "staphb/freyja:1.2"
+  }
+  command <<<
+  # configure barcode settings and capture version  
+  #if [[ ! -z "~{freyja_usher_barcodes}" ]]; then
+  #  #capture database info 
+  #  azfreyja_usher_barcode_version=$(basename -- "~{freyja_usher_barcodes}")
+  #  echo "here"
+  #  #set environment with user-defined db
+  #  mv ~{freyja_usher_barcodes} /opt/conda/envs/freyja-env/lib/python3.7/site-packages/freyja/data/usher_barcodes.csv
+  #else
+  #  # update db if specified
+  #  if ~{update_db}; then 
+  #    freyja update 
+  #    freyja_usher_barcode_version="freyja update: $(date +"%Y-%m-%d")"
+  #  else 
+  #    freyja_usher_barcode_version="unmodified from freyja container: ~{docker}"
+  #  fi
+  #fi
+  
+  # always update freyja barcodes until v1.3.1 release (will allow user-defined ref files)
+  freyja update 
+  freyja_usher_barcode_version="freyja update: $(date +"%Y-%m-%d")"
+  
+  echo ${freyja_usher_barcode_version} | tee FREYJA_BARCODES
+  
+  # Call variants and capture sequencing depth information
+  freyja variants ~{primer_trimmed_bam} --variants ~{samplename}_freyja_variants.tsv --depths ~{samplename}_freyja_depths.tsv --ref ~{reference_genome}
+ 
+  # Demix variants 
+  freyja demix ~{samplename}_freyja_variants.tsv ~{samplename}_freyja_depths.tsv --output ~{samplename}_freyja_demixed.tmp
+  # Adjust output header
+  echo -e "\t/~{samplename}" > ~{samplename}_freyja_demixed.tsv
+  tail -n+2 ~{samplename}_freyja_demixed.tmp >> ~{samplename}_freyja_demixed.tsv
+
+
+  >>>
+  runtime {
+    memory: "4 GB"
+    cpu: 2
+    docker: "~{docker}"
+    disks: "local-disk 100 HDD"
+  }
+  output {
+    File freyja_variants = "~{samplename}_freyja_variants.tsv"
+    File freyja_depths = "~{samplename}_freyja_depths.tsv"
+    File freyja_demixed = "~{samplename}_freyja_demixed.tsv"
+    String freyja_barcode_version = read_string("FREYJA_BARCODES")
+  }
+
+}

tasks/task_versioning.wdl

@@ -8,7 +8,7 @@ task version_capture {
     volatile: true
   }
   command <<<
-    PHVG_Version="PHVG v1.5.3"
+    PHVG_Version="PHVG v1.6.0-dev"
     ~{default='' 'export TZ=' + timezone}
     date +"%Y-%m-%d" > TODAY
     echo $PHVG_Version > PHVG_VERSION