Introduction to Bioinformatics workflows with Nextflow and nf-core: All Images

Getting Started with Nextflow

Figure 1

Image 1 of 1: ‘Flowchart illustrating a simple bioinformatics RNA-Seq pipeline for transcript expression quantification. The process begins with inputs at the top: 'Fastq' files, a 'Reference sequence', and 'Grch38 Ensembl 91' annotations. The first step is 'quality control', using 'fastQC' software version 0.11.9. The second step is 'index creation' with 'Salmon' software version 1.3.0 using the '-i' option. The third step is 'quantification', again with 'Salmon' version 1.3.0, this time using the '-l A' option. There are two outputs from this process: 'Output 1' is a 'QC report' and 'Output 2' is 'Transcript expression' data.’

Flowchart illustrating a simple bioinformatics RNA-Seq pipeline for transcript expression quantification. The process begins with inputs at the top: 'Fastq' files, a 'Reference sequence', and 'Grch38 Ensembl 91' annotations. The first step is 'quality control', using 'fastQC' software version 0.11.9. The second step is 'index creation' with 'Salmon' software version 1.3.0 using the '-i' option. The third step is 'quantification', again with 'Salmon' version 1.3.0, this time using the '-l A' option. There are two outputs from this process: 'Output 1' is a 'QC report' and 'Output 2' is 'Transcript expression' data.

An example of a simple bioinformatics RNA-Seq pipeline.

Figure 2

Image 1 of 1: ‘A comparison of three bioinformatics pipeline diagrams. Panel A shows an 'Analysis workflow' for transcript expression quantification with three main steps: 1) quality control using fastQC v0.11.9, 2) index creation with Salmon v.1.3.0, and 3) quantification also using Salmon v.1.3.0. Inputs include Fastq files, a Reference sequence, and Grch38 Ensembl 91, leading to outputs of a QC report and transcript expression data. Panel B illustrates a 'Traditional pipeline' emphasizing platform-specific requirements and local execution with steps leading to two outputs. Panel C depicts a 'Workflow manager', highlighting platform-independent requirements, portability, local and cloud execution options, scalability, and containerized steps for automatic resource management, leading to an output and an execution report. The color-coding indicates input data (gray), output data (yellow), and software, versions, parameters (green and blue)’

A comparison of three bioinformatics pipeline diagrams. Panel A shows an 'Analysis workflow' for transcript expression quantification with three main steps: 1) quality control using fastQC v0.11.9, 2) index creation with Salmon v.1.3.0, and 3) quantification also using Salmon v.1.3.0. Inputs include Fastq files, a Reference sequence, and Grch38 Ensembl 91, leading to outputs of a QC report and transcript expression data. Panel B illustrates a 'Traditional pipeline' emphasizing platform-specific requirements and local execution with steps leading to two outputs. Panel C depicts a 'Workflow manager', highlighting platform-independent requirements, portability, local and cloud execution options, scalability, and containerized steps for automatic resource management, leading to an output and an execution report. The color-coding indicates input data (gray), output data (yellow), and software, versions, parameters (green and blue)

An example of differences between running a specific analysis workflow using a traditional pipeline or a WfMS-based pipeline. Source: Wratten, L., Wilm, A. & Göke, J. Reproducible, scalable, and shareable analysis pipelines with bioinformatics workflow managers. Nat Methods 18, 1161–1168 (2021). https://doi.org/10.1038/s41592-021-01254-9

Figure 3

Image 1 of 1: ‘Infographic illustrating the components and supported platforms of a nextflow pipeline. The top section 'nextflow pipeline' is divided into three: writing code in any language (represented by R, Python, and Bash icons), orchestrating tasks with dataflow programming (represented by papers marked 'Data Flow' and 'Programming Model'), and defining software dependencies via containers (represented by Conda, Docker, and Singularity icons) and built-in version control with Git (represented by Git, GitHub, GitLab, and Bitbucket icons). Below, in the 'nextflow runtime' section, is 'Task orchestration and execution'. Arrows point downwards to the 'Supported Platforms' section, showcasing various platforms such as AWS, Google Cloud, Azure, Grid Engine, Slurm, HTCondor, Platform Computing, Kubernetes, and PBS Works.’

Infographic illustrating the components and supported platforms of a nextflow pipeline. The top section 'nextflow pipeline' is divided into three: writing code in any language (represented by R, Python, and Bash icons), orchestrating tasks with dataflow programming (represented by papers marked 'Data Flow' and 'Programming Model'), and defining software dependencies via containers (represented by Conda, Docker, and Singularity icons) and built-in version control with Git (represented by Git, GitHub, GitLab, and Bitbucket icons). Below, in the 'nextflow runtime' section, is 'Task orchestration and execution'. Arrows point downwards to the 'Supported Platforms' section, showcasing various platforms such as AWS, Google Cloud, Azure, Grid Engine, Slurm, HTCondor, Platform Computing, Kubernetes, and PBS Works.

Overview of Nextflow core features.

Figure 4

Image 1 of 1: ‘Diagram depicting part of a bioinformatics data processing workflow. On the left, there is a 'channel' labeled 'samples' containing three items: Fastq1, Fastq2, and Fastq3. This channel flows into a 'process' called 'fastqc' represented by a rounded rectangle containing the command 'fastqc -o out ${reads}'. The output of this process goes into a channel named 'out_ch', which lists 'outdir' three times as its contents. This channel then flows into a channel operator 'collect' and then into another 'process' called 'multiqc', indicated by a rounded rectangle with the command 'multiqc -o mqc_res .'. The output of 'multiqc' goes into a channel called 'mqc_ch', which also lists 'outdir' one time.’

$Diagram depicting part of a bioinformatics data processing workflow. On the left, there is a 'channel' labeled 'samples' containing three items: Fastq1, Fastq2, and Fastq3. This channel flows into a 'process' called 'fastqc' represented by a rounded rectangle containing the command 'fastqc -o out ${reads}'. The output of this process goes into a channel named 'out_ch', which lists 'outdir' three times as its contents. This channel then flows into a channel operator 'collect' and then into another 'process' called 'multiqc', indicated by a rounded rectangle with the command 'multiqc -o mqc_res .'. The output of 'multiqc' goes into a channel called 'mqc_ch', which also lists 'outdir' one time.$
Nextflow process flow diagram.

Figure 5

Image 1 of 1: ‘Diagram of a computational process within a bioinformatics workflow. The image features a large, central, rounded rectangle labeled 'process' with a smaller rectangle inside it labeled 'script', indicating the code or commands that are being executed. Above the script box, there is a smaller inset labeled 'Executors' with three icons: a desktop computer labeled 'Local', a stack of servers labeled 'High Performance Compute Cluster', and a cloud symbol labeled 'Cloud Compute'. These represent the different computing environments where the script can be executed. To the left of the process box is a green left-pointing arrowhead, suggesting input into the process, and to the right is a yellow right-pointing arrowhead, indicating the direction of output from the process.’

Nextflow Executors

Workflow parameterisation

Channels

Figure 1

Processes

Figure 1

Processes Part 2

Workflow

Operators

Reporting

Nextflow configuration

Figure 1

Workflow caching and checkpointing

Simple RNA-Seq pipeline

Figure 1

The INDEX process should be the longest running process. dag.png dag The vertices in the graph represent the pipeline’s processes and operators, while the edges represent the data connections (i.e. channels) between them.

Deploying nf-core pipelines

Figure 1

Figure 2