Chapter 5: Package Management with Conda, Mamba & Seqera Wave

Chapter 5: Package Management with Conda, Mamba & Seqera Wave

While containers offer maximum isolation, Conda and Mamba remain ubiquitous for rapid prototyping and local testing. Nextflow provides native directives to provision Conda environments dynamically at runtime.

1. Using Conda in Nextflow Processes

process BCFTOOLS_CALL {
    // Specify packages directly from bioconda and conda-forge
    conda 'bioconda::bcftools=1.19 conda-forge::gsl=2.7'

    input:
    tuple val(sample_id), path(bam), path(bai)
    path fasta

    output:
    tuple val(sample_id), path("${sample_id}.vcf.gz")

    script:
    """
    bcftools mpileup -Ou -f ${fasta} ${bam} | \
    bcftools call -mv -Oz -o ${sample_id}.vcf.gz
    """
}

2. Mamba Acceleration & Conda Lockfiles

The standard Conda SAT solver can take hours to resolve dependency trees. Nextflow supports Mamba (a C++ solver implementation) for 10x faster environment builds. For production pipelines, generate deterministic lockfiles using conda-lock to pin every transient sub-dependency.

# nextflow.config configuration for Conda
conda {
    enabled = true
    useMamba = true
    cacheDir = "$HOME/.conda_cache"
}

3. Nextflow Wave: On-Demand Dynamic Containers

Seqera Wave is a cloud service that automatically synthesizes, builds, and serves Docker containers on-the-fly directly from Conda or Dockerfile recipes in your process. You specify Conda packages in Nextflow, and Wave builds and caches an optimized minimal OCI container in real-time, eliminating local Conda builds entirely.

# Enable Wave in nextflow.config
wave.enabled = true
fusion.enabled = true // Enables high-speed S3/GCS virtual POSIX client