Hemoglobin predictor wrapped in a Docker container. Download
the docker image with command docker pull toivoja/hb-predictor.
A specific version (for example 0.23) can be downloaded with docker pull toivoja/hb-predictor:0.23.
The predictor is described in the following article:
Toivonen, J, Koski, Y, Turkulainen, E, Prinsze, F, della Briotta Parolo, P, Heinonen, M, et al.
Prediction and impact of personalized donation intervals.
Vox Sang. 2021; 1 – 9.
You can run the final image using
docker container run -it --rm -p 127.0.0.1:8080:8080 toivoja/hb-predictor
Then go to the following address in the browser:
http://localhost:8080/hb-predictor
Kill the server (and the server) by pressing control-c.
Note, this section contains obsolete information! The option to do the analyses from command line are not currently supported. Use the browser user interface instead.
docker container run -i --rm -v /whicheverfolderinputisin:/input toivoja/hb-predictor src/hb-predictor.sh date=2020-07-15 gender=female sample_fraction=0.002 method=icp-fix hlen=1 input_file=/input/inputfilename.rdata
Options:
- gender is either 'male', 'female', or 'both'
- method is either 'no-fix', 'icp-fix', or 'both'
- The input file can be specified with input_file option
The results will be stored to file results.tar.gz.
You can also build the docker image from source code yourself.
One can compose a new docker image from several docker images building on top
of each other. Building images in this layered fashion eases the development.
There exists a ready-made docker image containing a recent base version of R
called rocker/r-ubuntu.
Using command
docker build -f Dockerfile-r-base-tidyverse -t toivoja/r-base-tidyverse .
we build an intermediate image that adds many additional packages to the above image.
Some R packages are also installed from sources.
These packages are listed in file src/docker-install-packages2.R.
This second phase takes about 40 minutes on my machine.
The third docker image adds our web application built with the Plumber library. Use command
docker build -f Dockerfile -t toivoja/hb-predictor .
to build it. This phase is very quick (currently 3 minutes 14 seconds), which makes it easy to test changes to the script. You can list the docker images on this machine using the command docker images
The docker images are normally stored somewhere on the file system or on docker hub in an unspecified form. You can however export a docker image into a single tar file with
docker save toivoja/hb-predictor > hb-predictor.tar
Importing an image from a tar file is done by
docker load -i hb-predictor.tar
- The following page contains examples on building R script based docker applications: https://www.r-bloggers.com/running-your-r-script-in-docker/
- And the following page shows how a simple web application can be build with R package Rook: https://anythingbutrbitrary.blogspot.com/2012/12/a-simple-web-application-using-rook.html
- Why Rook does not work well with docker: https://pythonspeed.com/articles/docker-connection-refused/
- Plumber for creating a webserver that can listen to all interfaces, instead of just localhost: https://www.rplumber.io/docs/index.html
- Installing rstan: https://github.com/stan-dev/rstan/wiki/Installing-RStan-on-Linux