flowEMMi is a tool for automated model-based clustering of microbial data. It is particularly useful in two cases (i) your FCS data contains only few (maybe two!) channels and (ii) you need a fast tool.
Case (i) will happen quite often with non-human data, when no antibodies are available and only staining is, yielding foward- and side-scatter and no additional channels.
Case (ii) happens to everybody working with lots of data, typically those people dealing with problem (i).
Here are a number of ways how to run flowEMMi for your data.
Install the dependencies (as given in the em.R header), afterwards continue
with the ./em.R --help line below.
Should work with nixpkgs under any linux, as well.
- Download the newest package and extract
- cd AutoGating
- nix-shell (installation of missing dependencies)
- ./em.R --help (to show help)
- ./em.R --log --separation -f <fcs-file.fcs> (will compile the *cpp file and produce output files)
After running the program, there will be:
- png's (optionally svg's) with the cluster visualization
- bic.png with the graph of BIC for each number of clusters: choose the first number of cluster, once the graph flattens, for a sparse, good solution
- bic.txt with the textual data
- *txt files with relative fill data for each cluster
- cluster-positions-X.txt with the positions and covariance matrices for each cluster