pipeline_definition.yaml

---
NAME: metasub_assembly_cap
VERSION: 0.1
SNAKEMAKE:
  DIR: snakemake_files
  CONF: pipeline_config.py
REQUIREMENTS: {}
SAMPLE_TYPES: []
PACKAGE_MEGA:
  RECIPE_DIR: recipes

ORIGINS:
- - raw_short_read_dna

FILE_TYPES:
- name: gz_fastq
  ext: fastq.gz
- name: minsketch
  ext: msh
- fastqc
- bam
- sam
- txt
- m8
- name: gz_m8
  ext: m8.gz
- name: mpa
  ext: mpa.tsv
- fasta
- fastg
- vcf
- tsv
- csv
- name: gz_tsv
  ext: tsv.gz
- json
- name: jellyfish
  ext: jf


RESULT_TYPES:
- NAME: raw_short_read_dna
  FILES:
    read1: gz_fastq
    read2: gz_fastq

- NAME: megahit_assembly
  DEPENDENCIES:
  - raw_short_read_dna
  FILES:
    contigs: fasta

- NAME: concat_contigs
  LEVEL: GROUP
  DEPENDENCIES:
  - megahit_assembly
  FILES:
    contigs: fasta

- NAME: self_align
  LEVEL: GROUP
  DEPENDENCIES:
  - concat_contigs
  FILES:
    alignment: m8

- NAME: find_homologous_contigs
  LEVEL: GROUP
  DEPENDENCIES:
  - self_align
  FILES:
    strain_groups: csv
    strain_contigs: fasta
    homology_groups: csv
    homology_contigs: fasta
    
- NAME: align_to_nt
  LEVEL: GROUP
  DEPENDENCIES:
  - find_homologous_contigs
  FILES:
    alignment: m8

- NAME: group_assignments
  LEVEL: GROUP
  DEPENDENCIES:
  - align_to_nt
  FILES:
    taxa_ids: csv