preprocess.yaml

{% set MB_PREFIX = "mockinbird.stammp_modules" %}


{% set ref_gen = "/usr/users/ssohrab/CLIP/Reference_genome" %}

{% set star_index = ref_gen + "/STAR_generatedGenome" %}
{% set bowtie_index = ref_gen + "/BowtieIndex/genome" %}

{% set pileup = "/usr/users/ssohrab/CLIP/additional_files/RNAseq.mpileup" %}


general:
    adapter5prime: GTTCAGAGTTCTACAGTCCGACGATC
    adapter3prime: TGGAATTCTCGGGTGCCAAGG
    genomefasta: {{ ref_gen }}/WholeGenomeFasta/genome.fa
    normalization_pileup: {{ pileup }}
    rmTemp: no
    n_threads: 3

reads:
    fx_Q33: yes
    bc_5prime: 5
    bc_3prime: 0
    min_len: 20
    reference_nucleotide: T
    mutation_nucleotide: C

pipeline:
    - FastQCModule: &FASTQC_CFG
        kmer_length: 7         # int: used in fastqc's kmer statistics
        extra_flags: []        # list of additional flags
        outdir_name: fastQC_raw

    - UmiToolsExtractModule

    - SkewerAdapterClippingModule:
        extra_args:
            - -k 1

    - ClippyAdapterClippingModule:
        clip_len: 10
        clipped_5prime_bc: true

    - BowtieMapModule:
        genome_index: {{ bowtie_index }}
        n_multimap: 1
        n_mismatch: 1


    - SortIndexModule:
       keep_all: yes

    - UmiToolsDedupModule

    - BamPPModule:
        remove_n_edge_mut: 1
        plot_transition_profiles: yes
        max_mut_per_read: 1
        min_base_quality: 0
        min_avg_ali_quality: 0
        min_mismatch_quality: 20
        dump_raw_data: no
        outdir_name: bam_analysis

    - SoftclipAnalysisModule:
       outdir_name: bam_analysis

    - SortIndexModule:
       keep_all: yes

    - PileupModule:
       keep_all: yes

    - BSFinderModule:
        pval_threshold: 0.005
        min_cov: 2

    - NormalizationModule:
        mut_snp_ratio: 0.75

    - QuantileCapModule:
        max_quantile: 0.95