July 1st, 2015

Analyzing the data - manual shift



Analyzing the data - fully automated

Cell tracking in 4 steps

Overview of our workflow

Overview of our workflow

Meet the penalty function

Augmenting videos in interactive interfaces

  • Existing workaround:
    • export.Frames to png and gif format (via ImageMagick)
    • locator function used to behave weirdly
  • snap()!
    • Delivers integrated information on detected objects, trajectories, …

snap()!

flowcatchR in a simple script

library("flowcatchR")
data("MesenteriumSubset")

plateletsMesenterium <- channel.Frames(MesenteriumSubset, mode="red")

preprocessedPlatelets <- preprocess.Frames(plateletsMesenterium,brush.size=3, brush.shape="disc",
  at.wwidth=10, at.wheight=10,kern.size=3, kern.shape="disc",ws.tolerance=1, ws.radius=1)

platelets <- particles(plateletsMesenterium, preprocessedPlatelets)

linkedPlatelets <- link.particles(platelets,L=26, R=2,lambda1=1, lambda2=1,include.area=TRUE)

trajPlatelets <- trajectories(linkedPlatelets)

plot2D.TrajectorySet(trajPlatelets, MesenteriumSubset)

Deployment plan

  • How it turned out to be
    • simple
    • fast
    • painless to maintain
  • A long time ago in a galaxy far, far away…
    • dependencies to satisfy, many programs needed
    • complex installations
    • not working on particular machines

Need of a common computing environment

flowcatchR in Docker containers

  • Install Docker (that's it!) on a 64 bit machine
    • boot2docker also required on Windows
# go to the folder containing the files (<< 1 Mb)
docker build -t "jupyflow" /path/to/dockerfile

# or once it is available online
docker pull jupyflow # might take a while
docker run -p 8888:8888 jupyflow

… and just open the browser on localhost:8888!

Additional containers: RStudio server and Shiny server (combined with docker-compose)

flowcatchR and the IPython/Jupyter Notebook

shinyFlow - flowcatchR as a Shiny app

Try it out!

Acknowledgements

IMBEI

  • Johanna Mazur
  • Isabella Zwiener
  • Harald Binder


CTH Mainz

  • Sven Jäckel
  • Mareike Döhrmann
  • Kerstin Jurk

Thank you for your attention!

Manuscript in preparation, flowcatchR: an end-user-friendly workflow solution for automated analysis of fast moving blood cells, Marini F., Mazur J., Binder H.

marinif@uni-mainz.de