justold/immunopipe-base repository overview

⁠immunopipe-base

Base Docker image for the Immunopipe single-cell RNA-seq and TCR-seq analysis pipeline.

⁠Overview

immunopipe-base is the foundational layer in a multi-stage Docker build system for Immunopipe⁠. This image contains all core system dependencies, Python packages, R base environment, and Bioconductor packages required by the pipeline, providing a stable foundation that changes infrequently.

⁠Purpose

This base image serves as:

• Layer 1 in a 3-tier build strategy (base → rpkgs → immunopipe)
• Dependency foundation containing ~100+ conda packages including Python 3.12, R 4.3, and essential bioinformatics tools
• Cache optimization layer that rebuilds only when core dependencies change, significantly reducing build times

⁠Contents

⁠System Tools

• Node.js 22 (for pipen-report)
• Build essentials (gcc, git, curl, wget)
• System libraries (fontconfig, graphviz, libiconv)
• Utilities (procps-ng, time)

⁠Python Ecosystem

• Python 3.12
• Poetry (package management)
• Core scientific packages (via pip): liana, celltypist2, tensorflow 2.20, keras, scikit-learn, biopython

⁠R Ecosystem

• R 4.3 base
• Bioconductor core packages (fgsea, scater, glmgampoi, destiny, DESeq2, edgeR)
• Seurat ecosystem (seurat, seuratobject, seuratdisk, seuratwrappers)
• Single-cell analysis tools (doubletfinder, scdblfinder, scRepertoire)
• Visualization packages (ggplot2, complexheatmap, patchwork, plotroc)
• TCR/BCR analysis (immunarch via scRepertoire dependencies)
• Statistical packages (harmony, presto, glmnet, metap)

⁠Specialized Tools

• FAISS (CPU version 1.13) for similarity search
• ClusTCR 2025 for TCR clustering
• HDF5R for reading 10x data

⁠Build Process

This image is built from Dockerfile.base using environment_base.yml as the conda environment specification. The build includes aggressive cleanup to minimize image size:

• Removal of static libraries, headers, and build tools
• Deletion of Python bytecode and JavaScript source maps
• Stripping of R package documentation and help files
• Removal of compiler compatibility files

Automated builds: Triggered automatically via GitHub Actions when environment_base.yml or Dockerfile.base changes on the dev branch.

⁠Usage

⁠As a Base for immunopipe-rpkgs

FROM justold/immunopipe-base:latest

# Install custom R packages built from source
COPY --from=builder /opt/conda/lib/R/library/gglogger /opt/conda/lib/R/library/gglogger
# ...additional R packages...

⁠For Development/Testing

docker pull justold/immunopipe-base:latest
docker run -it justold/immunopipe-base:latest bash

⁠Image Size

Optimized through multi-stage cleanup, typically ~5-8GB (before R custom packages).

⁠Maintenance

• Source: immunopipe/docker/environment_base.yml⁠
• Dockerfile: immunopipe/Dockerfile.base⁠
• Update frequency: Changes only when core dependencies need version updates or new system-level packages are required

⁠Related Images

• justold/immunopipe-rpkgs (builds on this image): Adds custom R packages (gglogger, enrichit, plotthis, scplotter, biopipen.utils)
• justold/immunopipe (final image): Complete pipeline with Python code and configuration

⁠License

See the Immunopipe repository⁠ for license information.

⁠Support

For issues or questions:

• GitHub Issues: https://github.com/pwwang/immunopipe/issues⁠
• Documentation: https://pwwang.github.io/immunopipe/⁠