library(ArchR)
setwd('...')

addArchRGenome('hg38')
addArchRThreads(threads = 10) 

# fragment_files
fragment_files = c(Control1 = "Control1_fragments.tsv.gz",
                   Control2 = "Control2_fragments.tsv.gz",
                   Control3 = "Control3_fragments.tsv.gz",
                   Control4 = "Control4_fragments.tsv.gz",
                   Control5 = "Control5_fragments.tsv.gz")
# validBarcodes_list
# validBarcodes_list中的barcode需要与fragment文件第四行的细胞barcode一致
# 在Signac中，barcode存放在以下位置：
validBarcodes_list <- SeuratObj_ATAC@assays$ATAC@fragments %>% lapply(function(x){x@cells})
names(validBarcodes_list) <- names(fragment_files)
sapply(validBarcodes_list,length)

# 在当前路径下为每个fragment建.arrow文件
ArrowFiles <- c()
for(fragment_name in names(fragment_files)){
  # fragment_name <- 'Control1'
  fragment <- fragment_files[fragment_name]
  print(fragment)
  ArrowFile <- createArrowFiles(fragment ,validBarcodes = validBarcodes_list[[fragment_name]],
                                force = T, minTSS = 0,minFrags = 0, maxFrags = 1e+09)
  ArrowFiles <- c(ArrowFiles,ArrowFile)
}

# 创建proj
proj <- ArchRProject(ArrowFiles,outputDirectory = ".", copyArrows = F,showLogo = F)
proj <- addIterativeLSI(proj,dimsToUse = 1:50,iterations = 3,force = T)
proj <- addClusters(proj,reducedDims = 'IterativeLSI')
proj <- addGroupCoverages(proj)
proj <- addReproduciblePeakSet(proj,pathToMacs2 = '/home/timo/anaconda3/bin/macs3')
proj <- addPeakMatrix(proj)

# 提取GeneScoreMatrix
GeneScoreMatrix <- getMatrixFromProject(proj,useMatrix = "GeneScoreMatrix")
ArchR_GA <- GeneScoreMatrix@assays@data@listData$GeneScoreMatrix
rownames(ArchR_GA) <- GeneScoreMatrix@elementMetadata@listData$name
colnames(ArchR_GA) <- colnames(GeneScoreMatrix)