Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Bio Workflow Management Wdl Workflows · FreedomIntelligence | DecimalAI

Bio Workflow Management Wdl Workflows

SkillScore100Maintained

bio-workflow-management-wdl-workflows

Create portable bioinformatics pipelines with Workflow Description Language (WDL) using Cromwell or miniwdl execution engines. Use when running GATK best practices pipelines, working with Terra/AnVIL platforms, or building workflows for cloud execution on Google Cloud or AWS.

scienceby FreedomIntelligence

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Data Engineer pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/bio-workflow-management-wdl-workflows/SKILL.md

decimalai skills pull bio-workflow-management-wdl-workflows --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/bio-workflow-management-wdl-workflows/SKILL.md https://app.decimal.ai/s/bio-workflow-management-wdl-workflows/SKILL.md

Raw file: /s/bio-workflow-management-wdl-workflows/SKILL.md ↗ · machine-readable summary at /s/bio-workflow-management-wdl-workflows.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 50d ago

Yes — +5% lift · at higher cost+134% tokens0% turns100% of whole cases pass with skill (22/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-01	✗→✓	▲ Improved	93%	0%
case-09	✓→✓	＝ Same ✓	143%	0%
case-20	✓→✓	＝ Same ✓	85%	0%
case-02	✓→✓	＝ Same ✓	131%	0%
case-03	✓→✓	＝ Same ✓	222%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

FreedomIntelligence / OpenClaw-Medical-Skills

skills/bio-workflow-management-wdl-workflows/SKILL.md

View on GitHub ↗

Imported 8/1/2026Version v2

Skill contents

SKILL.md

<!--

COPYRIGHT NOTICE

This file is part of the "Universal Biomedical Skills" project.

Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>

All Rights Reserved.

This code is proprietary and confidential.

Unauthorized copying of this file, via any medium is strictly prohibited.

Provenance: Authenticated by MD BABU MIA

-->

WDL Workflows

Basic Task Definition

wdl
version 1.0

task fastqc {
    input {
        File fastq
        Int threads = 2
    }

    command <<<
        fastqc -t ~{threads} ~{fastq}
    >>>

    output {
        File html = glob("*_fastqc.html")[0]
        File zip = glob("*_fastqc.zip")[0]
    }

    runtime {
        docker: "biocontainers/fastqc:v0.11.9"
        cpu: threads
        memory: "4 GB"
    }
}

Simple Workflow

wdl
version 1.0

workflow rnaseq {
    input {
        File fastq_1
        File fastq_2
        File salmon_index
    }

    call fastp {
        input:
            reads_1 = fastq_1,
            reads_2 = fastq_2
    }

    call salmon_quant {
        input:
            reads_1 = fastp.trimmed_1,
            reads_2 = fastp.trimmed_2,
            index = salmon_index
    }

    output {
        File quant_sf = salmon_quant.quant_file
    }
}

Task with All Sections

wdl
version 1.0

task bwa_mem {
    input {
        File reference
        File reference_index
        File reads_1
        File reads_2
        String sample_id
        Int threads = 8
    }

    Int disk_size = ceil(size(reference, "GB") + size(reads_1, "GB") * 3) + 20

    command <<<
        bwa mem -t ~{threads} -R "@RG\tID:~{sample_id}\tSM:~{sample_id}" \
            ~{reference} ~{reads_1} ~{reads_2} | \
            samtools sort -@ ~{threads} -o ~{sample_id}.sorted.bam
        samtools index ~{sample_id}.sorted.bam
    >>>

    output {
        File bam = "~{sample_id}.sorted.bam"
        File bai = "~{sample_id}.sorted.bam.bai"
    }

    runtime {
        docker: "biocontainers/bwa:v0.7.17"
        cpu: threads
        memory: "16 GB"
        disks: "local-disk " + disk_size + " HDD"
    }
}

Scatter (Parallel Execution)

wdl
version 1.0

workflow process_samples {
    input {
        Array[File] fastq_files
        File reference
    }

    scatter (fastq in fastq_files) {
        call align {
            input:
                fastq = fastq,
                reference = reference
        }
    }

    output {
        Array[File] bam_files = align.bam
    }
}

Scatter with Paired Files

wdl
version 1.0

struct SampleFastqs {
    String sample_id
    File fastq_1
    File fastq_2
}

workflow paired_alignment {
    input {
        Array[SampleFastqs] samples
        File reference
    }

    scatter (sample in samples) {
        call align {
            input:
                sample_id = sample.sample_id,
                reads_1 = sample.fastq_1,
                reads_2 = sample.fastq_2,
                reference = reference
        }
    }

    output {
        Array[File] bams = align.bam
    }
}

Conditional Execution

wdl
version 1.0

workflow conditional_qc {
    input {
        File fastq
        Boolean run_qc = true
    }

    if (run_qc) {
        call fastqc {
            input:
                fastq = fastq
        }
    }

    output {
        File? qc_report = fastqc.html
    }
}

Structs and Complex Types

wdl
version 1.0

struct ReferenceData {
    File fasta
    File fasta_index
    File dict
    File? known_sites
}

workflow variant_calling {
    input {
        ReferenceData reference
        Array[File] bam_files
    }

    scatter (bam in bam_files) {
        call haplotype_caller {
            input:
                bam = bam,
                ref_fasta = reference.fasta,
                ref_index = reference.fasta_index,
                ref_dict = reference.dict
        }
    }
}

Input JSON

json
{
    "rnaseq.fastq_1": "data/sample1_R1.fq.gz",
    "rnaseq.fastq_2": "data/sample1_R2.fq.gz",
    "rnaseq.salmon_index": "ref/salmon_index",
    "rnaseq.threads": 8
}

Array Inputs JSON

json
{
    "process_samples.samples": [
        {
            "sample_id": "sample1",
            "fastq_1": "data/sample1_R1.fq.gz",
            "fastq_2": "data/sample1_R2.fq.gz"
        },
        {
            "sample_id": "sample2",
            "fastq_1": "data/sample2_R1.fq.gz",
            "fastq_2": "data/sample2_R2.fq.gz"
        }
    ],
    "process_samples.reference": "ref/genome.fa"
}

Subworkflows

wdl
version 1.0

import "qc.wdl" as qc
import "align.wdl" as align

workflow main_pipeline {
    input {
        File fastq_1
        File fastq_2
        File reference
    }

    call qc.quality_control {
        input:
            reads_1 = fastq_1,
            reads_2 = fastq_2
    }

    call align.alignment {
        input:
            reads_1 = quality_control.trimmed_1,
            reads_2 = quality_control.trimmed_2,
            reference = reference
    }
}

Runtime Options

wdl
runtime {
    docker: "ubuntu:20.04"
    cpu: 4
    memory: "8 GB"
    disks: "local-disk 100 HDD"
    preemptible: 3
    maxRetries: 2
    zones: "us-central1-a us-central1-b"
    bootDiskSizeGb: 15
}

String Interpolation and Expressions

wdl
version 1.0

task process {
    input {
        String sample_id
        Int memory_gb = 8
        Array[File] input_files
    }

    Int memory_mb = memory_gb * 1000
    String output_name = sample_id + ".processed.bam"

    command <<<
        # Access array elements
        process_tool \
            --memory ~{memory_mb} \
            --inputs ~{sep=' ' input_files} \
            --output ~{output_name}
    >>>

    output {
        File result = output_name
    }
}

File Size and Disk Calculation

wdl
version 1.0

task align {
    input {
        File reads_1
        File reads_2
        File reference
    }

    # Calculate disk: input files + 3x for outputs + buffer
    Int disk_gb = ceil(size(reads_1, "GB") + size(reads_2, "GB") +
                       size(reference, "GB") * 2) + 50

    command <<<
        bwa mem ~{reference} ~{reads_1} ~{reads_2} > aligned.sam
    >>>

    runtime {
        disks: "local-disk " + disk_gb + " SSD"
    }
}

Complete RNA-seq Workflow

wdl
version 1.0

workflow rnaseq_pipeline {
    input {
        Array[String] sample_ids
        Array[File] fastq_1_files
        Array[File] fastq_2_files
        File salmon_index
        Int threads = 8
    }

    scatter (idx in range(length(sample_ids))) {
        call fastp {
            input:
                sample_id = sample_ids[idx],
                reads_1 = fastq_1_files[idx],
                reads_2 = fastq_2_files[idx],
                threads = threads
        }

        call salmon_quant {
            input:
                sample_id = sample_ids[idx],
                reads_1 = fastp.trimmed_1,
                reads_2 = fastp.trimmed_2,
                index = salmon_index,
                threads = threads
        }
    }

    output {
        Array[File] quant_files = salmon_quant.quant_sf
        Array[File] fastp_reports = fastp.json_report
    }
}

task fastp {
    input {
        String sample_id
        File reads_1
        File reads_2
        Int threads = 4
    }

    command <<<
        fastp -i ~{reads_1} -I ~{reads_2} \
            -o ~{sample_id}_trimmed_R1.fq.gz \
            -O ~{sample_id}_trimmed_R2.fq.gz \
            --json ~{sample_id}_fastp.json \
            --thread ~{threads}
    >>>

    output {
        File trimmed_1 = "~{sample_id}_trimmed_R1.fq.gz"
        File trimmed_2 = "~{sample_id}_trimmed_R2.fq.gz"
        File json_report = "~{sample_id}_fastp.json"
    }

    runtime {
        docker: "quay.io/biocontainers/fastp:0.23.4--hadf994f_2"
        cpu: threads
        memory: "4 GB"
    }
}

task salmon_quant {
    input {
        String sample_id
        File reads_1
        File reads_2
        File index
        Int threads = 8
    }

    command <<<
        salmon quant -i ~{index} -l A \
            -1 ~{reads_1} -2 ~{reads_2} \
            -o ~{sample_id}_salmon \
            --threads ~{threads} --validateMappings
    >>>

    output {
        File quant_sf = "~{sample_id}_salmon/quant.sf"
        File quant_dir = "~{sample_id}_salmon"
    }

    runtime {
        docker: "quay.io/biocontainers/salmon:1.10.0--h7e5ed60_0"
        cpu: threads
        memory: "16 GB"
    }
}

Run Commands

bash
# Validate WDL syntax
womtool validate workflow.wdl

# Generate inputs template
womtool inputs workflow.wdl > inputs.json

# Run with Cromwell (local)
java -jar cromwell.jar run workflow.wdl -i inputs.json

# Run with miniwdl (simpler local runner)
miniwdl run workflow.wdl -i inputs.json

# Run on Terra
# Upload WDL and inputs.json to Terra workspace

Execution Engines

| Engine | Use Case | |--------|----------| | Cromwell | Full-featured, Google Cloud, AWS, HPC | | miniwdl | Lightweight local execution | | Terra | Cloud platform with Cromwell backend | | AnVIL | NIH cloud platform (Terra-based) | | dxWDL | DNAnexus platform |

Related Skills

workflow-management/cwl-workflows - CWL alternative
workflow-management/snakemake-workflows - Python-based alternative
workflow-management/nextflow-pipelines - Groovy-based alternative

Illustrative Want to see what bio-workflow-management-wdl-workflows would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/2/2026, 17:07 UTC · 22 cases

Lift

+5%

100% vs 95%

Agent turns

1.0 vs 1.0

Tokens

+134%

4,307 vs 1,838

fail → pass · 1pass → pass · 21

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-09 I want to process multiple paired-end samples in parallel in WDL 1.0. Should I pass two separate Array[File] inputs and iterate by numerical index, or define a struct for sample metadata and paired FASTQ paths and scatter over an array of that struct? Show how the workflow inputs and scatter block look.	pass→pass	9,436	8,475	-10%	1	1	0%	1,883	4,575	+143%	0	0	—
case-20 I am building an RNA-seq workflow in Nextflow DSL2. How do I define a process block that takes a channel of paired FASTQ files and emits trimmed FASTQ files using fastp?	pass→pass	10,605	7,675	-28%	1	1	0%	2,257	4,175	+85%	0	0	—
case-01 Can you write a WDL v1.0 task for aligning paired-end reads using BWA MEM and sorting with samtools? I need the task to accept fastq files, sample ID, reference files, and thread counts. It should dynamically compute the required disk space based on input file sizes to ensure enough storage, pipe BWA output directly into sorted samtools BAM output along with index generation, and include standard runtime configurations like Docker and memory.	fail→pass	13,876	11,399	-18%	1	1	0%	2,842	5,480	+93%	0	0	—
case-02 I'm writing a WDL v1.0 task for FastQC. Should I explicitly hardcode the expected FastQC output filename in the output section, or use glob patterns? Show me how the output section should be defined in WDL v1.0 for fastqc.	pass→pass	10,518	7,619	-28%	1	1	0%	1,970	4,555	+131%	0	0	—
case-03 In a WDL v1.0 workflow, I have an optional quality control step controlled by a Boolean flag 'run_qc'. If 'run_qc' is false, the fastqc task won't execute. How should the workflow output declaration represent the quality control report file since it might not exist?	pass→pass	6,584	5,038	-23%	1	1	0%	1,234	3,969	+222%	0	0	—
case-04 I need to organize genome reference files (FASTA, FASTA index, dictionary, and optional known sites VCF) into a single reusable structure in WDL 1.0 rather than passing four separate workflow inputs. How is this defined?	pass→pass	10,229	7,118	-30%	1	1	0%	2,080	4,494	+116%	0	0	—
case-05 I have two separate WDL files, 'qc.wdl' and 'align.wdl'. I want to call 'quality_control' from qc.wdl and 'alignment' from align.wdl inside a master workflow in WDL v1.0. How should the import statements and task calls be structured?	pass→pass	7,732	8,375	+8%	1	1	0%	1,601	4,686	+193%	0	0	—
case-06 I need to pass an Array[File] named 'input_files' into a shell command line inside a WDL 1.0 task block. Should I write a bash for-loop to iterate over input_files, or is there a standard WDL string interpolation function to join array items with space separators?	pass→pass	9,204	4,435	-52%	1	1	0%	1,772	3,855	+118%	0	0	—
case-07 In a WDL v1.0 task for alignment, I want to compute required disk space dynamically based on input FASTQ files (reads_1 and reads_2) and reference genome size. Should I estimate memory or disk space using float arithmetic directly in runtime, or define an Int expression using ceil and size()? Show the disk size expression.	pass→pass	11,539	6,063	-47%	1	1	0%	2,153	4,146	+93%	0	0	—
case-08 I have a complex WDL workflow file named 'workflow.wdl'. I want to auto-generate a JSON template containing all required workflow input parameters. What command-line tool command should I run?	pass→pass	8,797	2,123	-76%	1	1	0%	1,420	3,371	+137%	0	0	—
case-10 I have separate parallel arrays for sample IDs (sample_ids), read 1 files (fastq_1_files), and read 2 files (fastq_2_files) in a WDL 1.0 RNA-seq pipeline. How do I scatter over the samples by integer index to process each triplet together?	pass→pass	8,432	6,929	-18%	1	1	0%	1,715	4,567	+166%	0	0	—
case-11 I need to construct a fastp trimming task in WDL 1.0 that accepts paired FASTQ files, outputs trimmed FASTQ files and a JSON report, and runs inside a container. What container image repository tag and runtime block attributes should be defined for this task?	pass→pass	8,735	8,107	-7%	1	1	0%	1,986	4,846	+144%	0	0	—
case-12 In a WDL 1.0 task running salmon quantification for paired reads, I need to capture both the quant.sf file and the entire salmon output directory as task outputs. How should the output section be declared?	pass→pass	9,781	6,219	-36%	1	1	0%	1,698	4,226	+149%	0	0	—
case-13 How are structured array inputs formatted in Cromwell/WDL input JSON files when supplying sample metadata (sample ID, R1, R2) to a scattered workflow named 'process_samples'?	pass→pass	8,806	7,701	-13%	1	1	0%	1,885	4,696	+149%	0	0	—
case-14 I want to execute a WDL workflow locally with an input JSON file without starting a full Java Cromwell server or setting up database backends. What command line runner can I use?	pass→pass	8,439	3,836	-55%	1	1	0%	1,499	3,667	+145%	0	0	—
case-15 Before running a WDL pipeline on a cloud cluster, how do I check the syntax and validity of the WDL file locally from the command line?	pass→pass	8,569	6,171	-28%	1	1	0%	1,569	4,135	+164%	0	0	—
case-16 In a production Google Cloud Cromwell deployment, I want a WDL task runtime block to request preemptible VM instances up to 3 times before falling back to non-preemptible, and allow up to 2 task retries on failure. How are these options named in the runtime block?	pass→pass	7,229	4,918	-32%	1	1	0%	1,447	3,819	+164%	0	0	—
case-17 When a WDL workflow scatters tasks across samples, how are the outputs gathered across all scattered iterations into workflow-level outputs?	pass→pass	12,598	9,849	-22%	1	1	0%	2,204	4,821	+119%	0	0	—
case-18 I'm setting up a BWA MEM command inside a WDL v1.0 task. I need to set the read group header using sample ID for both ID and SM tags. What command string expression handles this string formatting in WDL?	pass→pass	9,507	6,585	-31%	1	1	0%	1,754	4,218	+140%	0	0	—
case-19 When specifying the disk attribute in a WDL runtime block on cloud platforms like GCP, how is the storage disk string formatted to request local disk space of a dynamic size with SSD or HDD type?	pass→pass	10,148	7,276	-28%	1	1	0%	1,906	4,350	+128%	0	0	—
case-21 I need a Snakemake rule to index a reference FASTA file using samtools faidx. How should the rule definition specify input, output, and shell command?	pass→pass	7,735	3,760	-51%	1	1	0%	1,578	3,729	+136%	0	0	—
case-22 How do I write a CWL (Common Workflow Language) CommandLineTool in YAML for running FastQC on an input File parameter?	pass→pass	9,309	7,168	-23%	1	1	0%	1,983	4,374	+121%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 95.5%
Pass rate with the skill: 100%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v2
Run completed: 8/2/2026, 17:07 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-01

Prompt

Can you write a WDL v1.0 task for aligning paired-end reads using BWA MEM and sorting with samtools? I need the task to accept fastq files, sample ID, reference files, and thread counts. It should dynamically compute the required disk space based on input file sizes to ensure enough storage, pipe BWA output directly into sorted samtools BAM output along with index generation, and include standard runtime configurations like Docker and memory.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run100

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+138% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 49 days ago

Model	Method	Date	Lift
gemini-3.6-flash	verified	7/27/2026	+5%

Other measured skills in the registry, with their headline benchmark lift.