CellCODE with gene lists

geneListCellCode.R

library(CellCODE)
library(dplyr)
library(purrr)
library(magrittr)
library(markerGeneProfile)
data('mouseMarkerGenesCombined')

# this function will turn gene lists into CellCODE inputs
cellCodeInput = function(genes){
    geneCount = genes %>% map(length)
    allGenes = genes %>% unlist %>% unique
    
    tags = genes %>% lapply(function(x){
        out = rep(0,length(allGenes))
        out[allGenes %in% x] = 1
        return(out)
    }) %>% as.data.frame()
    
    rownames(tags) = allGenes
    return(tags)
}

# this is marker gene input
geneList = mouseMarkerGenesCombined$Cortex %>% map(mouse2human) %>% map('humanGene')
# remove genes that are not in the dataset
geneList %<>% map(function(x){
    x[x %in% rownames(cortex_white)]
})
# code works with gene count > 2. Not sure why but it is a decent cutoff anyway
geneList = geneList[sapply(geneList,length)>2]


tagList = cellCodeInput(geneList)

# dataset input is a MATRIX (not data frame) with genes as rownames
cortex_white = trabzuniRegionsExp[, c(trabzuniRegionsMeta$brainRegion %in% c('frontal cortex', 'white matter'))]
groups = trabzuniRegionsMeta$brainRegion[trabzuniRegionsMeta$brainRegion %in% c('frontal cortex', 'white matter')]

SPVs=getAllSPVs(data = cortex_white %>% as.matrix(), 
                 grp = groups %>% as.factor,
                 dataTag = tagList, 
                 method = "mixed", 
                 plot =  T)