CoGAPS Case Study 4 — Docker + Quarto + PyCoGAPS

This repo provides a reproducible environment for Case Study 4 using:

• Docker (RStudio Server + Quarto)
• A project-local Python environment (.venv)
• PyCoGAPS built from source (includes the C++ CoGAPS core)

Dataset strategy (user-provided mount)

This project assumes you download the dataset once via your browser and mount it into the container.

The notebook expects the dataset at:

• /opt/data/kang_counts_25k.h5ad

---

1) Download the dataset (one-time)

Download the file from Figshare (use a web browser if curl fails):

• https://figshare.com/ndownloader/files/34464122

Save it as:

• kang_counts_25k.h5ad

Create a local folder (in the repo root) and place the file there:

mkdir -p data_mount
# move or save the downloaded file here:
# data_mount/kang_counts_25k.h5ad

---

2) Build the Docker image

From the repo root (where the Dockerfile is):

docker build --platform linux/amd64 -t cogaps-case-study4:dev .

Apple Silicon Macs: the --platform linux/amd64 flag is required.

---

3) Run RStudio Server (recommended: mount repo + dataset)

This run command mounts:

• your local repo → /home/rstudio/project (so edits + renders persist on your machine)
• your dataset folder → /opt/data (read-only)

docker run --platform linux/amd64 -it --rm \
  -p 8787:8787 \
  -e PASSWORD="Password12" \
  -v "$PWD":/home/rstudio/project \
  -w /home/rstudio/project \
  -v "$PWD/data_mount":/opt/data:ro \
  cogaps-case-study4:dev

Open:

• http://localhost:8787
• User: rstudio
• Password: Password12 (or whatever you set)

---

4) One-time setup inside the container

In RStudio → Terminal:

cd /home/rstudio/project
bash setup_venv.sh

(Optional sanity check)

source _environment
quarto check jupyter

---

5) Run the case study

Interactive mode (recommended for students)

Open the .qmd file in RStudio and run chunks one by one:

• cogaps_case_study4_student.qmd (render-safe version recommended)

Render mode (HTML report)

cd /home/rstudio/project
source _environment
quarto render cogaps_case_study4_student.qmd

---

Optional: Distributed CoGAPS (advanced)

Run distributed CoGAPS as a standalone script (not during Quarto render):

cd /home/rstudio/project
source _environment
./.venv/bin/python run_pycogaps_distributed.py

This should write:

• data/cogaps_result_distributed.h5ad

---