rm(list = ls())
setwd("D:\\the_first_paper\\ZRC_the_first_paper\\ZPP")
options(stringsAsFactors = F)
Sys.setenv("VROOM_CONNECTION_SIZE" = 131072 * 60)

library(WGCNA)
library(tibble)
library(GEOquery)
library(limma)
library(affy)
library(clusterProfiler)
library(org.Hs.eg.db)
library(AnnoProbe)

gse <- getGEO("GSE30529", destdir = ".",
              getGPL = T,
              AnnotGPL = T)

exp <- exprs(gse[[1]])
clip <- pData(gse[[1]]) 

group <- ifelse(clip$source_name_ch1 == "Tubuli of control kidney", "control", "DKD")

gse[[1]]

GPL <- fData(gse[[1]])
gpl <- GPL[, c(1, 3)]
gpl$`Gene symbol` <- data.frame(sapply(gpl$`Gene symbol`,
                                 function(x)unlist(strsplit(x, "///"))[1]),
                          stringsAsFactors = F)[,1]

exp <- as.data.frame(exp)
exp$ID <- rownames(exp)

exp_symbol <- merge(exp, gpl, by = "ID")
exp_symbol <- na.omit(exp_symbol)
table(duplicated(exp_symbol))
exp_unique <- avereps(exp_symbol[, -c(1, ncol(exp_symbol))],
                      ID = exp_symbol$`Gene symbol`)
table(duplicated(rownames(exp_unique)))
save(clip, exp, exp_symbol, exp_unique, gpl, GPL, gse, group, file = "GSE30529.Rdata")

## raw_data
setwd("D:\\the_first_paper\\ZRC_the_first_paper\\ZPP\\GSE30529_RAW")
raw <- ReadAffy()
raw_rma <- rma(raw)
dim(raw_rma)
data <- t(data.frame(raw_rma))
save(data, file = "raw.Rdata")