foundations-training

A conceptual and practical checkpoint for the STaiMIC Training Program Using Nextflow From Biological Question to Nextflow Execution: Foundations for Independent Workflow Building

Independent training and affiliation notice

This educational programme was independently designed and delivered by STaiMIC. Its curriculum is informed by open Nextflow and nf-core community practices and uses the open-source Nextflow workflow system and selected nf-core resources for teaching and demonstration.

This is not an official Seqera, Nextflow, or nf-core course, training session, certification, or product. It is not sponsored, endorsed, reviewed, or certified by Seqera or the nf-core community.

Any instructor’s status as a Nextflow Ambassador is an individual community role and does not imply that Seqera, Nextflow, or nf-core delivers or endorses this STaiMIC programme. Third-party names and marks are used solely for identification and attribution.

Licensing and third-party material

Original STaiMIC material and code in this repository is released under the MIT License. Nextflow is separate open-source software released under the Apache License 2.0. Reused or adapted nf-core material remains subject to its original MIT licence and copyright notices. The STaiMIC licence does not replace or relicense third-party material.

---

What is this?

This repository is the Session 5 practical component of the STaiMIC Nextflow Bioinformatics Training Program (Wednesday 19 August 2026). Unlike Sessions 1–4, this session does not run a new pipeline. It corrects terminology and reproducibility misconceptions surfaced across Sessions 1–4, so students can separate the biology, workflow logic, execution environment, and validation evidence before building independently in Session 6.

Quick Start

Click the badge above, or navigate to Code → Codespaces → Create codespace on main. Wait ~2 minutes for the environment to set up automatically. Then open GETTING_STARTED.md for the full step-by-step guide.

What you'll examine

A deliberately simplified fragment from ampliseq or sarek, followed by one minimal two-process DSL2 workflow. You will trace: biological input → channel → process/task → output → published result → test.

Synthetic Dataset

This session reuses the real samplesheet formats from Session 1 (sarek) and Session 2 (ampliseq), so students recognise the schemas — but the read files themselves are illustrative only, not real FASTQ data.

data/microbiome_samplesheet.csv — mirrors the Session 2 ampliseq format:

sampleID	forwardReads	reverseReads
gut_1	fastq/gut_1_1.fastq.gz	fastq/gut_1_2.fastq.gz
gut_2	fastq/gut_2_1.fastq.gz	fastq/gut_2_2.fastq.gz

data/microbiome_metadata.tsv — mirrors the Session 2 Metadata.tsv:

ID	condition	status
gut_1	healthy	0
gut_2	healthy	0

data/cancer_samplesheet.csv — mirrors the Session 1 sarek format, included for comparison only (not run in this session's workflow):

patient	sex	status	sample	lane	fastq_1	fastq_2
patientX	XX	0	normal_sample	L001	fastq/normal_1.fastq.gz	fastq/normal_2.fastq.gz
patientX	XX	1	tumor_sample	L001	fastq/tumor_1.fastq.gz	fastq/tumor_2.fastq.gz

Why compare both formats? Sarek needs patient, sex, status, and lane because somatic variant calling requires pairing tumor and normal samples from the same patient. Ampliseq only needs sampleID and read paths because 16S profiling doesn't require that pairing. Different biological questions require capturing different metadata — this is the Biology layer in practice.

What you'll learn

Biology layer — research question, samples, reference data, biological output and interpretation.

Workflow layer — Nextflow, nf-core, pipeline, workflow, process, module, task, channel.

Execution layer — executor, container, profile, configuration, work directory, cache, -resume.

Evidence layer — tests, CI, reports, provenance, determinism, reproducibility.

Mandatory Session 6 Entry Gate

• Label a DSL2 workflow correctly
• Explain the biology separately from the computation
• Connect two processes using a channel
• Define typed inputs and named outputs
• Pin software and record provenance correctly
• Run once normally and once as a clean, uncached reproduction
• Explain what a passing test proves — and what it does not

---

Built for the STaiMIC Nextflow Training Program by Nkiruka Cynthia Efenji · Nextflow Ambassador · @Seqera