How to create and manage conda environments for reproducible bioinformatics analysis

Conda environments are the standard solution for this. Use **mamba** (drop-in conda replacement) — it's 10× faster for solving environments. ```bash # Install mamba into base conda install -n base -c conda-forge mamba # Create a new environment with specific Python version mamba create -n rnaseq_env python=3.11 conda activate rnaseq_env # Install tools (always specify channels: conda-forge > bioconda > defaults) mamba install -c conda-forge -c bioconda star=2.7.11 samtools=1.19 trimmomatic=0.39 fastqc multiqc subread # featureCounts # Install R packages in the same environment mamba install -c conda-forge -c bioconda bioconductor-deseq2 r-ggplot2 ``` **Export environment for reproducibility:** ```bash # Export full spec (exact versions + build strings) — 100% reproducible conda env export > environment_full.yml # Export cross-platform spec (no build strings) — more portable conda env export --no-builds > environment.yml # Export only manually installed packages conda env export --from-history > environment_minimal.yml ``` **Share and recreate from yml:** ```bash mamba env create -f environment.yml # or mamba env create -n new_env -f environment.yml ``` **Best practices:** - One environment per project - Always include versions in your yml: `star=2.7.11` not just `star` - Use `conda-forge` before `bioconda` in channel priority - For Docker-based reproducibility, use Snakemake + conda + `--use-conda` flag - Add `environment.yml` to your git repository ```bash # Snakemake auto-installs per-rule environments: # rule star_align: # conda: 'envs/star.yaml' # ... snakemake --use-conda --cores 8 ```