Closed lcolladotor closed 3 years ago
Note that one of the steps takes nearly a day to run as in https://github.com/LieberInstitute/HumanPilot/blob/master/Analysis/sce_scran.R#L460-L466. So you might want to use a screen
here first (see https://linuxize.com/post/how-to-use-linux-screen/).
Still needs to be done by adding more code at https://github.com/LieberInstitute/spatialDLPFC/blob/main/analysis/01_build_SPE.R prior to saving spe.Rdata
.
Use load(here::here("rdata", "spe", "spe.Rdata"), verbose = TRUE)
to re-load the current data before adding the code related to clustering.
We don't know where to save the clusters to. Do we just create a list of characters called sce_discrete_vars? What kind of list should it be?
For plotting the clusters, you might want to try something like https://github.com/LieberInstitute/spatialDLPFC/blob/main/analysis/02_marker_genes.R#L43-L51 where we create a directory and store 1 PDF per grid plot (so it's not a HUUUUGE single PDF).
Failed after almost 6 days.
After #10, we can run some graph-based clustering algorithms as in https://github.com/LieberInstitute/HumanPilot/blob/master/Analysis/sce_scran.R#L452-L468 + https://github.com/LieberInstitute/HumanPilot/blob/master/Analysis/sce_scran.R#L1185-L1191 and add them to the
sce_discrete_vars
in #8 as in https://github.com/LieberInstitute/spatialLIBD/blob/master/R/run_app.R#L63.