traj-dockers

This repo holds Dockerfiles for containers of trajectory inference algorithms. The containers can be found on the Stuart lab's docker hub. Each one of these containers holds the converter functions in order to parse the output of trajectory algorithms into common formats.

The main purpose of the containers is to minimize installation woes by facilitating development and analysis inside the container. This has the desirable side effect of reproducible computation environments. I'm willing to work with people if the container design doesn't fit a particular workflow (e.g. doesn't do jupyter notebook at the moment). Please submit a 'would be nice' issue and I'll have a peak and see what I can do.

The docker containers can also be run stand alone and exec'd into. If that's more your slice of pie see the second workflow example below.

For instructions below I suggest creating a shared volume, ./shared in the directory you are launching the docker from. This volume acts as shared storage. The shared storage is the easiest way to access input and write output from the container.

Example workflow analysis with monocle outputing common formats

starting from a bash shell:

# pull the docker image down to your machine
docker pull dmccoll/monocle:0.1

# Move into some directory and make a space for shared storage.
cd some/directory && mkdir shared

# Move the data you'd like into shared storage
cp ../../some-data ./shared

# Run the containers Rstudio server
docker run -v $(pwd)/shared:/home/rstudio/shared -d -p 8787:8787 -e ROOT=TRUE dmccoll/monocle:0.1

From there open your favorite browser (tested on chrome) at http://localhost:8787/.

Now use the editor to create a script that runs on your data, and save the script to /home/rstudio/shared inside the container. An example script could be something as short as:

library(monocle)
# Make random data for testing
#rd <- replicate(110, rnbinom(500, c(3, 10, 45, 100), .1))
#colnames(rd)<- 1:110
#Astrocyte_cds <- newCellDataSet(rd)

# Contains an Astrocyte_cds CellDataSet 
load("/home/rstudio/shared/Astrocyte_cds.rda")

# Monocle analysis pipeline
Astrocyte_cds <- estimateSizeFactors(Astrocyte_cds)
Astrocyte_cds <- estimateDispersions(Astrocyte_cds)
Astrocyte_ordering_genes <- subset(dispersionTable(Astrocyte_cds), mean_expression>=0.1)
Astrocyte_cds <- setOrderingFilter(Astrocyte_cds, Astrocyte_ordering_genes)
Astrocyte_cds <- reduceDimension(Astrocyte_cds, max_components = 2, method = "DDRTree")
Astrocyte_cds <- orderCells(Astrocyte_cds)
# Astrocyte_cds is now a completed monocle S4 class instance.

# source the r script with the converters for monocle.
source("/traj-converters/src/R/monocle_convert.r")

# output the common json and cell_x_branch matrix from the monocle object
write_cell_x_branch(Astrocyte_cds, file="/home/rstudio/shared/Astro.monocle.cellxbranch.tab")
write_common_json(Astrocyte_cds, file="/home/rstudio/shared/Astro.monocle.json")

After executing each line of that script in rstudio's console (e.g. using ctrl + enter shortcut of Rstudio's editor) you will have a ./shared/Astro.monocle.json and a ./shared/Astro.monocle.cellxbranch.tab file on your local machine.