--- a
+++ b/paper/Application on scRNA and scATAC datasets/rds_to_h5.R
@@ -0,0 +1,77 @@
+library(chromVAR) # BiocManager::install("chromVAR") DirichletMultinomial TFBSTools
+library(SummarizedExperiment)
+
+scpeaks <- readRDS('data_raw/scATAC-Healthy-Hematopoiesis-191120.rds')
+scATAC <- readRDS('data_raw/scATAC-Cicero-GA-Hematopoiesis-191120.rds')
+scRNA <- readRDS('data_raw/scRNA-Healthy-Hematopoiesis-191120.rds')
+
+scchromVAR <- readRDS('data_raw/scATAC-chromVAR-Hematopoiesis-191120.rds')
+dev <- deviations(scchromVAR)
+rownames(dev) <- as.vector(rowData(scchromVAR)$name)
+
+# remove unkown cell type
+dev <- dev[,!endsWith(colData(scATAC)$BioClassification, 'Unk')]
+scATAC <- scATAC[,!endsWith(colData(scATAC)$BioClassification, 'Unk')]
+scRNA <- scRNA[,!endsWith(colData(scRNA)$BioClassification, 'Unk')]
+
+
+# keep shared genes
+gene_names <- unique(intersect(row.names(scATAC), row.names(scRNA)))
+scATAC <- scATAC[match(gene_names, row.names(scATAC)), ]
+scRNA <- scRNA[match(gene_names, row.names(scRNA)), ]
+
+
+dev <- dev[,colSums(assays(scATAC)$gA)!=0]
+scATAC <- scATAC[,colSums(assays(scATAC)$gA)!=0]
+scRNA <- scRNA[,colSums(assays(scRNA)$counts)!=0]
+
+
+cell_types_ATAC <- substr(colData(scATAC)$BioClassification, 4, 20)
+cell_types_RNA <- substr(colData(scRNA)$BioClassification, 4, 20)
+
+rename_ct <- function(ct){
+    ct[startsWith(ct, 'CD4.N')] <- 'CD4.N'
+    #ct[startsWith(ct, 'CD14.Mono')] <- 'CD14.Mono'
+    ct[startsWith(ct, 'CLP')] <- 'CLP'
+    return(ct)
+}
+cell_types_ATAC <- rename_ct(cell_types_ATAC)
+cell_types_RNA <- rename_ct(cell_types_RNA)
+
+uni_cell_types <- union(unique(cell_types_ATAC),unique(cell_types_RNA))
+
+
+covariates_RNA <- scRNA$Group
+covariates_ATAC <- scATAC$Group
+
+
+library(Seurat)
+library(hdf5r)
+library(Matrix)
+
+file.h5 <- H5File$new('human_hematopoiesis_scATAC.h5', mode = "w")
+file.h5[["count"]] <- as.matrix(assays(scATAC)$gA)
+# expression <- NormalizeData(as.matrix(assays(scATAC)$gA))
+# file.h5[["expression"]] <- expression
+file.h5[["grouping"]] <- as.vector(cell_types_ATAC)
+file.h5[["cell_ids"]] <- colnames(scATAC)
+file.h5[["gene_names"]] <- rownames(scATAC)
+file.h5[["covariates"]] <- matrix(scATAC$Group, ncol=1)
+file.h5$close_all()
+
+file.h5 <- H5File$new('human_hematopoiesis_scRNA.h5', mode = "w")
+file.h5[["count"]] <- as.matrix(assays(scRNA)$counts)
+# expression <- NormalizeData(assays(scRNA)$counts)
+# file.h5[["expression"]] <- as.matrix(expression)
+file.h5[["grouping"]] <- as.vector(cell_types_RNA)
+file.h5[["cell_ids"]] <- colnames(scRNA)
+file.h5[["gene_names"]] <- rownames(scRNA)
+file.h5[["covariates"]] <- matrix(scRNA$Group, ncol=1)
+file.h5$close_all()
+
+
+file.h5 <- H5File$new('human_hematopoiesis_motif.h5', mode = "w")
+file.h5[["count"]] <- as.matrix(dev)
+file.h5[["cell_ids"]] <- colnames(dev)
+file.h5[["motif_names"]] <- rownames(dev)
+file.h5$close_all()
\ No newline at end of file