coastguard.yml

settings:
  environ:
    working_directory: .
    temporary_directory: ''
    output_directory: ''
    custom:
      Baywatch:
        name: 'Pinfish'
        description: |
            Pipeline for annotating genomes using long read transcriptomics data with pinfish
        fields:
        - label: Input genome
          description: Input genome
          path: workflow.data.genome_fasta.path
          fieldType: file
        - label: cDNA or direct RNA reads in fastq format
          description: cDNA or direct RNA reads in fastq format
          path: workflow.data.reads_fastq.path
          fieldType: file
        - label: Extra option passed to minimap2 when generating index
          description: Extra option passed to minimap2 when generating index
          path: workflow.data.minimap_index_opts.path
          fieldType: string
          value: -k14
        - label: Extra options passed to minimap2
          description: Extra options passed to minimap2
          path: workflow.data.minimap2_opts.path
          fieldType: string
          value: -uf
        - label: Minmum mapping quality
          description: Minmum mapping quality
          path: workflow.data.minimum_mapping_quality.path
          fieldType: number
          value: 10
        - label: Options passed to spliced_bam2gff
          description: Options passed to spliced_bam2gff
          path: workflow.data.spliced_bam2gff_opts.path
          fieldType: string
          value: -s
        - label: Minimum cluster size
          description: Minimum cluster size
          path: workflow.data.minimum_cluster_size.path
          fieldType: number
          value: 10
        - label: Internal exon boundary tolerance
          description: Internal exon boundary tolerance
          path: workflow.data.exon_boundary_tolerance.path
          fieldType: number
          value: 10
        - label: Terminal exon boundary tolerance
          description: Terminal exon boundary tolerance
          path: workflow.data.terminal_exon_boundary_tolerance.path
          fieldType: number
          value: 30
        - label: Extra minimap2 options for mapping polished reads
          description: Extra minimap2 options for mapping polished reads
          path: workflow.data.minimap2_opts_polished.path
          fieldType: string
          value: -uf
        - label: Options passed to spliced_bam2gff when converting alignments of polished
            reads
          description: Options passed to spliced_bam2gff when converting alignments
            of polished reads
          path: workflow.data.spliced_bam2gff_opts_pol.path
          fieldType: string
          value: -s
        - label: Internal exon boundary tolerance for collapsing
          description: Internal exon boundary tolerance for collapsing
          path: workflow.data.collapse_internal_tol.path
          fieldType: number
          value: 5
        - label: Five prime boundary tolerance for collapsing
          description: Five prime boundary tolerance for collapsing
          path: workflow.data.collapse_five_tol.path
          fieldType: number
          value: 5000
        - label: Three prime boundary tolerance for collapsing
          description: Three prime boundary tolerance for collapsing
          path: workflow.data.collapse_three_tol.path
          fieldType: number
          value: 30
        - label: Threads
          description: Threads
          path: workflow.data.threads.path
          fieldType: number
          value: 50
workflow:
  data:
    genome_fasta:
      class: File
      path: ''
    reads_fastq:
      class: File
      path: ''
    minimap_index_opts: -k14
    minimap2_opts: -uf
    minimum_mapping_quality: 10
    spliced_bam2gff_opts: -s
    minimum_cluster_size: 10
    exon_boundary_tolerance: 10
    terminal_exon_boundary_tolerance: 30
    minimap2_opts_polished: -uf
    spliced_bam2gff_opts_pol: -s
    collapse_internal_tol: 5
    collapse_five_tol: 5000
    collapse_three_tol: 30
    threads: 50
  steps:
  - Process:
      name: 'Pinfish'
      HostSNK:
        target: all
        cores: 5
        location:
          class: Directory
          path: .
        config_template:
          class: File
          path: config.yml
        inputs:
        - path: data.genome_fasta
          confpath: genome_fasta
          remap: paths
        - path: data.reads_fastq
          confpath: reads_fastq
          remap: paths
        - path: data.minimap_index_opts
          confpath: minimap_index_opts
        - path: data.minimap2_opts
          confpath: minimap2_opts
        - path: data.minimum_mapping_quality
          confpath: minimum_mapping_quality
        - path: data.spliced_bam2gff_opts
          confpath: spliced_bam2gff_opts
        - path: data.minimum_cluster_size
          confpath: minimum_cluster_size
        - path: data.exon_boundary_tolerance
          confpath: exon_boundary_tolerance
        - path: data.terminal_exon_boundary_tolerance
          confpath: terminal_exon_boundary_tolerance
        - path: data.minimap2_opts_polished
          confpath: minimap2_opts_polished
        - path: data.spliced_bam2gff_opts_pol
          confpath: spliced_bam2gff_opts_pol
        - path: data.collapse_internal_tol
          confpath: collapse_internal_tol
        - path: data.collapse_five_tol
          confpath: collapse_five_tol
        - path: data.collapse_three_tol
          confpath: collapse_three_tol
        - path: data.threads
          confpath: threads