config.yaml

samples: samples.tsv
units: units.tsv
reheader: reheader.tsv
internal_sid: "LIMS"
read_type: "pe"

references:
    basepath: "path_to_references_data"
    provider: "ucsc"
    release: "hg19"
genome_fasta: "ucsc.hg19.fasta"
genes_gtf: "Annotation/Genes/genes.gtf"

rseqc_reference:
    basepath: "path_to_references_data"
    provider: "rseqc"
    release: "hg19"
housekeeping: "hg19.HouseKeepingGenes.bed"
refseq: "hg19_RefSeq.bed"

transcriptome_reference:
    basepath: "path_to_references_data"
    provider: "ensembl"
    release: "GRCh38"
transcriptome_fasta: "Homo_sapiens.GRCh38.cdna.all.fa.gz"

strand: "reverse"

rules:
    multiqc:
        arguments: "--force "
    featureCounts_run:
        gtf_feature_type: "exon"
        gtf_attribute_type: "gene_id"
    kallisto:
        arguments_pe: ""
        arguments_se: "--single -b 30 -l 280 -s 80 "
    trim_galore_pe:
        arguments: "--paired -q 20 --cores 4 --fastqc --fastqc_args '--outdir qc/fastqc --quiet'"
    trim_galore_se:
        arguments: "-q 20 --fastqc --fastqc_args '--outdir qc/fastqc --quiet'"
    fastq_screen:
        params: "--force --aligner bowtie2 --subset 100000 --nohits"
        config_file: "../data/fastq_screen.config"