Reproducible, transparent, and adaptable data analysis with Snakemake

Johannes Köster

 

2024

dataset

results

Data analysis

"Let me do that by hand..."

dataset

results

dataset

dataset

dataset

dataset

dataset

"Let me do that by hand..."

Data analysis

  • check computational validity
  • apply same analysis to new data
  • check methodological validity
  • understand analysis

Data analysis

Reproducibility

Transparency

  • modify analysis
  • extend analysis

Adaptability

>1 million downloads since 2015

>2700 citations

>11 citations per week in 2023

  • automation
  • scalability
  • portability
  • readability
  • documentation
  • traceability

Data analysis

Reproducibility

Transparency

  • readability
  • portability
  • scalability

Adaptability

Data analysis

  • automation
  • scalability
  • portability
  • readability
  • documentation
  • traceability

Reproducibility

Transparency

  • readability
  • portability
  • scalability

Adaptability

dataset

results

dataset

dataset

dataset

dataset

dataset

Define workflows

in terms of rules

Define workflows

in terms of rules

rule mytask:
    input:
        "path/to/{dataset}.txt"
    output:
        "result/{dataset}.txt"
    script:
        "scripts/myscript.R"


rule myfiltration:
     input:
        "result/{dataset}.txt"
     output:
        "result/{dataset}.filtered.txt"
     shell:
        "mycommand {input} > {output}"


rule aggregate:
    input:
        "results/dataset1.filtered.txt",
        "results/dataset2.filtered.txt"
    output:
        "plots/myplot.pdf"
    script:
        "scripts/myplot.R"

Define workflows

in terms of rules

Define workflows

in terms of rules

rule mytask:
    input:
        "data/{sample}.txt"
    output:
        "result/{sample}.txt"
    shell:
        "some-tool {input} > {output}"

rule name

how to create output from input

define

  • input
  • output
  • log files
  • parameters
  • resources
rule mytask:
    input:
        "path/to/{dataset}.txt"
    output:
        "result/{dataset}.txt"
    script:
        "scripts/myscript.R"


rule myfiltration:
     input:
        "result/{dataset}.txt"
     output:
        "result/{dataset}.filtered.txt"
     shell:
        "mycommand {input} > {output}"


rule aggregate:
    input:
        "results/dataset1.filtered.txt",
        "results/dataset2.filtered.txt"
    output:
        "plots/myplot.pdf"
    script:
        "scripts/myplot.R"

Automatic inference of DAG of jobs

Boilerplate-free integration of scripts

rule mytask:
    input:
        "data/{sample}.txt"
    output:
        "result/{sample}.txt"
    script:
        "scripts/myscript.py"

reusable scripts:

  • Python
  • R
  • Julia
  • Rust
  • Bash
import pandas as pd

data = pd.read_table(snakemake.input[0])
data = data.sort_values("id")
data.to_csv(snakemake.output[0], sep="\t")

Python:

data <- read.table(snakemake@input[[1]])
data <- data[order(data$id),]
write.table(data, file = snakemake@output[[1]])

Boilerplate-free integration of scripts

R:

import polar as pl

pl.read_csv(&snakemake.input[0])
  .sort()
  .to_csv(&snakemake.output[0])

Rust:

Jupyter notebook integration

rule mytask:
    input:
        "data/{sample}.txt"
    output:
        "result/{sample}.txt"
    notebook:
        "notebooks/mynotebook.ipynb"
  1. Integrated interactive edit mode.
  2. Automatic generalization for reuse in other jobs.

Reusable wrappers

rule map_reads:
    input:
        "{sample}.bam"
    output:
        "{sample}.sorted.bam"
    wrapper:
        "0.22.0/bio/samtools/sort"

reuseable wrappers from central repository

Reusable wrappers

Output handling

rule mytask:
    input:
        "data/{sample}.txt"
    output:
        temp("result/{sample}.txt")
    shell:
        "some-tool {input} > {output}"

Output handling

rule mytask:
    input:
        "data/{sample}.txt"
    output:
        protected("result/{sample}.txt")
    shell:
        "some-tool {input} > {output}"

Output handling

rule mytask:
    input:
        "data/{sample}.txt"
    output:
        pipe("result/{sample}.txt")
    shell:
        "some-tool {input} > {output}"

Using and combining workflows

configfile: "config/config.yaml"

module some_workflow:
    snakefile:
        github("some-org/some-workflow", path="workflow/Snakefile", tag="v1.17.0")
    config:
        config["some-workflow"]

use rule * from some_workflow
# easily extend the workflow
rule some_plot:
    input:
        "results/some-table.h5"
    output:
        "results/plots/some-plot.pdf"
    notebook:
        "notebooks/some-plot.py.ipynb"
module some_other_workflow:
    snakefile:
        github("some-org/some-other-workflow", path="workflow/Snakefile", tag="v2.5.0")
    config:
        config["some-other-workflow"]

use rule * from some_other_workflow

use rule process_xy from some_other_workflow with:
  	params:
      	some_threshold=0.3

Data analysis

  • automation
  • scalability
  • portability
  • readability
  • documentation
  • traceability

Reproducibility

Transparency

  • readability
  • portability
  • scalability

Adaptability

\max U_t \cdot 2S \cdot \sum_{j \in J} x_j \cdot p_j + 2S \cdot \sum_{j \in J} x_j \cdot (u_{t,j}) + S \cdot \sum_{f \in F} \gamma_f \cdot S_f\\ + \sum_{f \in F} \delta_f \cdot S_f
\sum_{j \in J} x_j \cdot u_{r,j} \leq U_r \quad \forall r \in R
\delta_f \leq \frac{\sum_{j \in J} x_j \cdot z_{f,j}}{\sum_{j \in J} z_{f,j}} \quad\forall f \in F
\text{subject to:}

job selection

job resource usage

free resources

job temp file consumption

temp file lifetime fraction

job priority

job thread usage

Scheduling

temp file size

temp file deletion

\gamma_f \leq \delta_f \quad\forall f \in F
\gamma_f \in \{0,1\}
\delta_f \in [0,1]
x_f \in \{0,1\}

DAG partitioning

--groups a=g1 b=g1
--groups a=g1 b=g1
--group-components g1=2
--groups a=g1 b=g1
--group-components g1=5

Scalable to any platform

workstation

compute server

cluster

grid computing

cloud computing

Data analysis

  • automation
  • scalability
  • portability
  • readability
  • documentation
  • traceability

Reproducibility

Transparency

  • readability
  • portability
  • scalability

Adaptability

rule mytask:
    input:
        "path/to/{dataset}.txt"
    output:
        "result/{dataset}.txt"
    conda:
        "envs/some-tool.yaml"
    shell:
        "some-tool {input} > {output}"

Conda integration

channels:
 - conda-forge
dependencies:
  - some-tool =2.3.1
  - some-lib =1.1.2

Container integration

rule mytask:
    input:
        "path/to/{dataset}.txt"
    output:
        "result/{dataset}.txt"
    container:
        "docker://biocontainers/some-tool#2.3.1"
    shell:
        "some-tool {input} > {output}"

Containerization

containerized:
    "docker://username/myworkflow:1.0.0"


rule mytask:
    input:
        "path/to/{dataset}.txt"
    output:
        "result/{dataset}.txt"
    conda:
        "envs/some-tool.yaml"
    shell:
        "some-tool {input} > {output}"
snakemake --containerize > Dockerfile

Data analysis

  • automation
  • scalability
  • portability
  • readability
  • documentation
  • traceability

Reproducibility

Transparency

  • readability
  • portability
  • scalability

Adaptability

Self-contained HTML reports

Many more features

  • dynamic DAG rewiring
  • service jobs (providing sockets, loading databases, or ramdisks)
  • semantic helper functions for minimizing boilerplate code
  • fallible rules
  • caching of shared results across workflows
  • transparent handling of remote storage
  • interoperability (CWL tool wrappers, integration of Nextflow workflows)

Extensible architecture

Extensible architecture

Snakemake workflow catalog

Conclusion

Snakemake covers all aspects of fully reproducible, transparent, and adaptable data analysis, offering

  • maximum readability
  • ad-hoc integration with scripting and high performance languages
  • an extensible architecture
  • a plethora of advanced features

https://snakemake.github.io

Eager to chat and learn about use cases, issues, challenges!

Available today, tomorrow, Friday morning

Introduction to Snakemake

By Johannes Köster

Introduction to Snakemake

Introduction to Snakemake @ CERN

  • 453