Snakefile.template

# This is a template file that needs to be moved to the working directory of the pipeline. It contains the target files of the pipeline. Please customize for your dataset.
import os

# Include config file and rule definitions
configfile: "config.yaml"

wildcard_constraints:
    genome = "GRCh..",
    patient = "[0-9\-]+",
    sample = "[0-9m_]+"

ALIGNERS = ["minimap2"]
PATIENTS = ["sample1", "sample2"]

for patient in PATIENTS:
    os.makedirs("logs/pbmm2_align/%s/" % (patient), exist_ok=True)
    os.makedirs("logs/ngmlr_align/%s/" % (patient), exist_ok=True)

include: "align.smk"
include: "call.smk"
include: "filter_and_merge.smk"

rule all:
    input:
        expand("aligned_bam/{patient}/pooled.{aligner}.{genome}.bam", patient=PATIENTS,
               aligner=ALIGNERS,
               genome=["GRCh37", "GRCh38"]),
        expand("alignment_stats/{patient}/{genome}/{patient}.{genome}.nb.html", genome=["GRCh37", "GRCh38"],
               patient=PATIENTS),
        expand("calls/svim/{aligner}/{patient}_{genome}/variants.vcf", aligner=ALIGNERS,
               patient=PATIENTS,
               genome=["GRCh37", "GRCh38"]),
        expand("calls/pbsv/{aligner}/{patient}_{genome}/variants.vcf", aligner=ALIGNERS,
               patient=PATIENTS,
               genome=["GRCh37", "GRCh38"]),
        expand("calls/sniffles/{aligner}/{patient}_{genome}/variants.vcf", aligner=ALIGNERS,
               patient=PATIENTS,
               genome=["GRCh37", "GRCh38"]),
        expand("calls/{caller}/{aligner}/{patient}_{genome}/variants.tsv", caller=["svim", "pbsv", "sniffles"],
               aligner=ALIGNERS,
               patient=PATIENTS,
               genome=["GRCh37", "GRCh38"]),
        expand("calls/merged/{aligner}/{patient}_{genome}/merged_variants.vcf", aligner=ALIGNERS,
               patient=PATIENTS,
               genome=["GRCh37", "GRCh38"]),
        expand("calls/merged/{aligner}/{patient}_{genome}/venn",
               aligner=ALIGNERS, patient=PATIENTS, genome=["GRCh37", "GRCh38"])