Sarek Metro

Gallery / Realistic Pipelines

Integration showcase: a sarek-style variant-calling pipeline drawn with diagonal (45-degree) labels for tight column packing, an off-track FASTQ input, file termini, marker styles, and a `%%metro line_spread: rails` panel where each caller keeps its own rail and shared stations render as interchanges.

Mermaid source

examples/sarek_metro.mmd

%%metro title: Example analysis pathways
%%metro caption: Adapted from: Fellows Yates, James A., et al. PeerJ 9 (2021)
%%metro style: dark
%%metro logo: examples/nf-core-sarek_logo_light.png | examples/nf-core-sarek_logo_dark.png
%%metro logo_scale: 1.0
%%metro font_scale: 1.3
%%metro legend_logo_gap: 40
%%metro legend: br
%%metro label_angle: 45
%%metro line_spread: rails | calling
%%metro grid: preprocessing | 0,0
%%metro grid: variantcalling | 1,0
%%metro grid: annotation | 2,0
%%metro grid: calling | 0,1,1,3

%%metro line: core | Core workflow | #2db572
%%metro line: germline | Germline | #0570b0
%%metro line: tumor_only | Tumor only | #d62728
%%metro line: pair_n | Tumor-normal (normal) | #0570b0
%%metro line: pair_t | Tumor-normal (tumor) | #d62728

%% Tumor-normal pair is the normal + tumor lines bundled together; give the
%% combination a single legend entry instead of two separate rows.
%%metro legend_combo: pair_n, pair_t | Tumor-normal pair

%%metro files: ubam_in | uBAM
%%metro files: spring_in | spring
%%metro files: fastq_in | FASTQ
%%metro file: vcf_ann | VCF
%%metro file: vcf_vc | VCF
%%metro files: cram_call | CRAM
%%metro file: out | VCF
%%metro file: out | TXT
%%metro files: out | ...

%% Intermediate alignment artefacts written off the trunk at the step that
%% produces them (mapping, samtools, recalibration prep, applybqsr).
%%metro files: mapped_out | BAM/CRAM
%%metro files: samtools_out | CRAM
%%metro files: recalprep_out | CRAM
%%metro files: recal_out | CRAM

%% ubam and spring need conversion, so they feed convert; fastq is already in
%% the right format and joins the trunk at FastQC, past convert.  spring and
%% fastq lift off the trunk as off-track inputs with an S-curve.  The bam/cram
%% outputs hang off their producer the same way (off_track anchors a sink to
%% its source).
%%metro off_track: fastq_in
%%metro off_track: spring_in
%%metro off_track: mapped_out
%%metro off_track: samtools_out
%%metro off_track: recalprep_out
%%metro off_track: recal_out
%%metro off_track: vcf_vc

%% Marker key. Optional steps use the standard station pill (undistinguished);
%% only mandatory / accelerated steps get a marker shape. Acceleration is shown
%% by fill colour: Parabricks (green), Sentieon (navy), or both (teal) on
%% whatever shape the step already carries.
%%metro marker_legend: square, solid | Mandatory
%%metro marker_legend: square, #4CAF50 | Parabricks accelerated
%%metro marker_legend: square, #1b3a6b | Sentieon accelerated
%%metro marker_legend: square, #2f7f74 | Parabricks & Sentieon accelerated
%%metro marker_legend: pill, open | Expanded in pathways panel

%% Caller families along the pathways trunk.
%%metro group: SNPs & Indels | deepvariant, freebayes, haplotypecaller, haplotyper, dnascope, tnscope, lofreq, muse, mpileup, mutect2, strelka2
%%metro group: SV & CNV | indexcov, manta, tiddit, ascat, controlfreec, cnvkit
%%metro group: MSI | msisensor2, msisensorpro

%% --- markers: mandatory steps get a square; acceleration is the fill colour.
%% mapping and markduplicates are accelerable by both Parabricks and Sentieon
%% (teal); the BQSR steps are Parabricks-only (green); the Sentieon callers in
%% the pathways panel are Sentieon-only (navy).
%%metro marker: convert | square, solid
%%metro marker: bam_convert | square, solid
%%metro marker: mapping | square, #2f7f74
%%metro marker: markduplicates | square, #2f7f74
%%metro marker: prepare_recal | circle, #4CAF50
%%metro marker: applybqsr | circle, #4CAF50
%%metro marker: vc_anchor | pill, open

%% Sentieon-accelerated callers in the pathways panel (navy circles).  Single-rail
%% callers (haplotyper, dnascope) show the marker; tnscope spans two rails and
%% renders as a plain interchange (a marker on a spanning rail station is drawn
%% as the interchange bar, not the glyph).
%%metro marker: haplotyper | circle, #1b3a6b
%%metro marker: dnascope | circle, #1b3a6b
%%metro marker: tnscope | circle, #1b3a6b


graph LR
    subgraph preprocessing [Pre-processing]
        ubam_in[ ]
        spring_in[ ]
        fastq_in[ ]
        convert[convert]
        fastqc[FastQC]
        umi[UMI]
        fastp[FastP]
        bbsplit[BBsplit]
        mapping[mapping]
        bam_convert[convert]
        markduplicates[markduplicates]
        mosdepth[mosdepth, samtools]
        prepare_recal[prepare recalibration]
        applybqsr[applybqsr]
        mosdepth_qc[mosdepth, samtools]
        ngscheckmate[NGSCheckmate]
        mapped_out[ ]
        samtools_out[ ]
        recalprep_out[ ]
        recal_out[ ]

        ubam_in -->|core| convert
        spring_in -->|core| convert
        fastq_in -->|core| fastqc
        convert -->|core| fastqc
        fastqc -->|core| umi
        umi -->|core| fastp
        fastp -->|core| bbsplit
        bbsplit -->|core| mapping
        mapping -->|core| bam_convert
        mapping -->|core| markduplicates
        bam_convert -->|core| mosdepth
        markduplicates -->|core| mosdepth
        mosdepth -->|core| prepare_recal
        prepare_recal -->|core| applybqsr
        applybqsr -->|core| mosdepth_qc
        mosdepth_qc -->|core| ngscheckmate

        mapping -->|core| mapped_out
        mosdepth -->|core| samtools_out
        prepare_recal -->|core| recalprep_out
        applybqsr -->|core| recal_out
    end

    subgraph variantcalling [Variant calling]
        vc_anchor[variant calling]
        bcftools[bcftools]
        samtools_vc[samtools]
        varlociraptor[Varlociraptor]
        finalise[finalise]
        normalise[normalise]
        consensus[consensus]
        _vc_merge[ ]
        vcf_vc[ ]

        vc_anchor -->|core| bcftools
        bcftools -->|core| samtools_vc
        samtools_vc -->|core| varlociraptor
        samtools_vc -->|core| finalise
        finalise -->|core| normalise
        normalise -->|core| consensus
        varlociraptor -->|core| _vc_merge
        consensus -->|core| _vc_merge
        _vc_merge -->|core| vcf_vc
    end

    subgraph annotation [Annotation]
        snpeff[snpEff]
        ensemblvep[ensemblVEP]
        bcftools_annotate[bcftools annotate]
        multiqc[MultiQC]
        vcf_ann[ ]

        snpeff -->|core| ensemblvep
        ensemblvep -->|core| bcftools_annotate
        bcftools_annotate -->|core| multiqc
        multiqc -->|core| vcf_ann
    end

    subgraph calling [Example analysis pathways]
        cram_call[ ]
        deepvariant[DeepVariant]
        freebayes[FreeBayes]
        haplotypecaller[HaplotypeCaller]
        haplotyper[Sentieon Haplotyper]
        dnascope[Sentieon DNAscope]
        tnscope[Sentieon TNscope]
        lofreq[LoFreq]
        muse[MuSE]
        mpileup[mpileup]
        mutect2[Mutect2]
        strelka2[Strelka2]
        indexcov[indexcov]
        manta[Manta]
        tiddit[TIDDIT]
        ascat[ASCAT]
        controlfreec[Control-FREEC]
        cnvkit[CNVkit]
        msisensor2[MSIsensor2]
        msisensorpro[MSIsensor-pro]
        out[ ]

        %% Germline route (top rail): deepvariant, haplotypecaller, haplotyper,
        %% dnascope are germline-only; freebayes/mpileup/strelka2/manta/tiddit/
        %% cnvkit/indexcov are shared.
        cram_call -->|germline| deepvariant
        deepvariant -->|germline| freebayes
        freebayes -->|germline| haplotypecaller
        haplotypecaller -->|germline| haplotyper
        haplotyper -->|germline| dnascope
        dnascope -->|germline| mpileup
        mpileup -->|germline| strelka2
        strelka2 -->|germline| indexcov
        indexcov -->|germline| manta
        manta -->|germline| tiddit
        tiddit -->|germline| cnvkit
        cnvkit -->|germline| out

        %% Tumour-only route (middle rail): tnscope, lofreq, mutect2, msisensor2
        %% join here; mpileup/indexcov run tumour but not the combined pair.
        cram_call -->|tumor_only| freebayes
        freebayes -->|tumor_only| tnscope
        tnscope -->|tumor_only| lofreq
        lofreq -->|tumor_only| mpileup
        mpileup -->|tumor_only| mutect2
        mutect2 -->|tumor_only| indexcov
        indexcov -->|tumor_only| manta
        manta -->|tumor_only| tiddit
        tiddit -->|tumor_only| controlfreec
        controlfreec -->|tumor_only| cnvkit
        cnvkit -->|tumor_only| msisensor2
        msisensor2 -->|tumor_only| out

        %% Tumour-normal pair route (bottom rail): muse, ascat, msisensorpro are
        %% combined-only; strelka2 runs germline + combined (not tumour-only).
        cram_call -->|pair_n,pair_t| freebayes
        freebayes -->|pair_n,pair_t| tnscope
        tnscope -->|pair_n,pair_t| muse
        muse -->|pair_n,pair_t| mutect2
        mutect2 -->|pair_n,pair_t| strelka2
        strelka2 -->|pair_n,pair_t| manta
        manta -->|pair_n,pair_t| tiddit
        tiddit -->|pair_n,pair_t| ascat
        ascat -->|pair_n,pair_t| controlfreec
        controlfreec -->|pair_n,pair_t| cnvkit
        cnvkit -->|pair_n,pair_t| msisensorpro
        msisensorpro -->|pair_n,pair_t| out
    end

    %% Inter-section: core workflow flows through the top three sections.
    ngscheckmate -->|core| vc_anchor
    _vc_merge -->|core| snpeff

CLI command

nf-metro render examples/sarek_metro.mmd -o sarek_metro.svg

Rendered map