hcp4715/rstudio_bayes repository overview

⁠rstudio_docker (working env of the lab)

This repo is to ensure reproducibility of analyses from Hu Chuan-Peng⁠'s team. This tutorial⁠ helped us to build and maintain this repo, many thanks.

The primary purpose of this docker image is to ensure my collaborators can run brms across different platforms, so that they don't need to worry about the installation problems.

This docker image can be used for Bayesian analyses, it includes the following packages: brms, cmdstanr, tidybayes. Also, it includes lme4, tidyverse, metafor. I will include more packages in the future. You can also easily install new packages and save the docker images locally, please see the tutorial⁠ I mentioned above.

⁠About docker

Please see here⁠ for why using docker.

⁠About tags in this series of images

• r4.3.1-arms: for apple chips; R 4.3.1; CmdStan 2.33.1; rstan 2.32.3; tidyverse 2.0.0; lme4 1.1-34
• cmdstanr: for intel chips; R 4.0.4;

⁠How to use this docker images

⁠Step 1: install docker

Linux (ubuntu): https://docs.docker.com/engine/install/ubuntu/⁠

Windows 10: https://docs.docker.com/docker-for-windows/install/⁠

Mac OS: https://docs.docker.com/docker-for-mac/install/⁠

Make sure that docker was successfully installed on your machine

⁠Step 2: pull this image form dock hub

docker pull hcp4715/rstudio_bayes            # this doesn't work, because docker will try to find a image "hcp4715/rstudio_bayes:latest"
docker pull hcp4715/rstudio_bayes:cmdstanr   # docker will try to find a image "hcp4715/rstudio_bayes:cmdstanr"

⁠Step 3: run the docker image:

docker run -e PASSWORD=hulab --cpus=4 -it --rm -p 8787:8787 -v /home/hcp4715/docker_R:/home/rstudio/tutorial hcp4715/rstudio_bayes:cmdstanr

docker run ---- Run a docker image in a container

-e PASSWORD=hulab ---- set a password for rstudio, you can set your own password.

-it ---- Keep STDIN open even if not attached

--rm ---- Automatically remove the container when it exits

--cpus=4 ---- Number of cores will be used by docker. Make sure that your machine has more thread than the number here.

-v ---- Mount a folder to the container

/home/hcp4715/docker_R ---- The directory of a local folder where I stored my data. [For Linux]

/d/hcp4715/docker_R ---- The directory of a local folder under drive D. It appears as D:\hcp4715\docker_R in windows system.

/home/rstudio/tutorial ---- The directory inside the docker image (the mounting point of the local folder in the docker image). Note that the docker container itself likes a mini virtual linux system, so the file system inside it is linux style.

-p ---- Publish a container’s port(s) to the host

hcp4715/rstudio_bayes:cmdstanr ---- The docker image to run. Note that you shoud include the tag :cmdstanr part, otherwise docker will instead use "hcp4715/rstudio_bayes:latest".

After running the code above, you shall see output as below:

[s6-init] making user provided files available at /var/run/s6/etc...exited 0.
[s6-init] ensuring user provided files have correct perms...exited 0.
[fix-attrs.d] applying ownership & permissions fixes...
[fix-attrs.d] done.
[cont-init.d] executing container initialization scripts...
[cont-init.d] userconf: executing... 
[cont-init.d] userconf: exited 0.
[cont-init.d] done.
[services.d] starting services
[services.d] done.
TTY detected. Printing informational message about logging configuration. Logging configuration loaded from '/etc/rstudio/logging.conf'. Logging to 'syslog'.

Then, open your broswer (e.g., firefox, chrome), and try one of the following url in the address:

localhost:8787

http://192.168.99.100:8787

You will be asked to input username and password

Username: rstudio Password: hulab

Now, you will see the familiar interface of rstudio! In the broswer!

⁠Step 4: test the iamge

Run the code below to test whether the image works

# load the libraries
library(brms)
library(tidyverse)

# fit a testing model from brms, use rstan as the backend
fit1 <- brm(count ~ zAge + zBase * Trt + (1|patient), 
	    cores = parallel::detectCores(), # detect how many cpus/threads are available
	    chains = 4,
        data = epilepsy, family = poisson())

# fit a testing model from brms, use cmdstanr as the backend
fit1 <- brm(count ~ zAge + zBase * Trt + (1|patient), 
	    cores = parallel::detectCores(), # detect how many cpus/threads are available
	    chains = 4,  
		backend = 'cmdstanr',
        data = epilepsy, family = poisson())

# check the summary of the model
summary(fit1)

⁠Build docker image from Dockerfile

docker build -t your_user_name/your_docker_image_name:your_tag -f Dockerfile .

Replace the your_user_name/your_docker_image_name:your_tag part with your own information. For example, I useddocker build -t hcp4715/rstudio_bayes:cmdstanr -f Dockerfile .

⁠Final note

Docker is not perfect (yet). First, this docker image doesn't work on Mac with M1 chip. Second, on some Mac book, this docker image does not work perfectly: when running brms, it can only work when using cmdstanr as the backend, not rstan. See issues for the details.