Closed lcolladotor closed 1 year ago
@lahuuki prior to meeting with @heenadivecha, check with lobstr::obj_size()
that your SCE
object is about 2 GB of RAM or less. You might have to drop the counts
and keep only the logcounts
. So you'll end up making a SCE
object only for shiny.
We do want to prioritize making an initial version of this app since @kmartinow @kmaynard12 et al could then use it to explore violin plots of select genes. Either split by the current cell type annotations at the specific or broad resolutions, but also results from https://github.com/LieberInstitute/DLPFC_snRNAseq/issues/12 and merged clusters based on https://github.com/LieberInstitute/spatialDLPFC/issues/117 (I'll make an issue about this).
@heenadivecha and I have gotten iSEE to work with the local copy of the sce. We are getting stuck deploying the shiny app, and have some questions with how to use the HDF5 style data. The sce object appears to be a iSEE friendly size, but I think thats because the assays are still on the disk.
sce <- loadHDF5SummarizedExperiment("sce_DLPFC_annotated/")
lobstr::obj_size(sce)
# 171.96 MB
When we go to upload all the necessary files from the HDF5 dir we get this error. So we may have to subset?
library(“rsconnect”)
source(“token.R”)
options(repos = BiocManager::repositories())
rsconnect::deployApp(
appFiles = c(“app.R”, “sce_DLPFC_annotated/“, “initial.R”),
appName = “DLPFC_snRNA-seq_2022",
account = “libd”,
server = “[shinyapps.io](http://shinyapps.io/)”
)
# Error: The directory /Users/heenadivecha/Documents/code/DLPFC_snRNAseq/code/06_iSEE_apps cannot be deployed # # because it is too large (the maximum size is 3145728000 bytes). Remove some files or adjust the rsconnect.max.bundle.size option.
From https://docs.rstudio.com/shinyapps.io/applications.html, looks like we can upload up to 5 GB since we have a Professional shinyapps.io account. So let's try with options(rsconnect.max.bundle.size= 1024^3 * 5)
or options(rsconnect.max.bundle.size= 1024^3 * 4.9)
.
I think this one has mostly been resolved, right @heenadivecha? Though note that I did change the URL and title at https://github.com/LieberInstitute/DLPFC_snRNAseq/commit/abfb4bdb3baaab214870ad2c895dde977f93a375, so it has to be re-deployed.
The old app has to be archived, though I recommend doing this after you have deployed the new one, so you can copy the settings if needed.
Yes, @lcolladotor this issue has been resolved. The iSEE app has been re-deployed.
Thanks @heenadivecha!
I deleted the old app from our shinyapps.io account to avoid any confusion moving forward ^^
Related to #80
This one will be challenging due to memory issues as it's a large dataset. We might have to do this one outside shinyapps.io and recruit Bill's help to setup an AWS machine to run this shiny app. In any case, prepare the
app_dir
files like in #80 and test it locally.Unlike the apps from https://github.com/LieberInstitute/10xPilot_snRNAseq-human#explore-the-data-interactively, we'll want to include in the
colData()
both the broad and specific cell types columns. Actually, it'd be great if we updated the ones at https://github.com/LieberInstitute/10xPilot_snRNAseq-human#explore-the-data-interactively with that, since those apps only have the specific cell types and as we noticed with TREG, sometimes it's useful to be able to plot along the broad cell types. CC'ing @mattntran.Colors again are a big one here and it'll be important to use colors consistent with the prior DLPFC apps for the cell types.