Diff of /Data_preprocessing.R [000000] .. [a50134]

Switch to unified view

a b/Data_preprocessing.R
1
setwd('D:/brca')
2
library('imputeMissings')
3
data=read.csv("brca_multitest.csv",row.names = 1)
4
5
#################delete missing#######################
6
sum(is.na(data))
7
miss=c()
8
for (i in 1:nrow(data)){
9
  miss=c(miss,sum(is.na(data[i,])))
10
}
11
missrate=miss/ncol(data)
12
data1=data[missrate<0.2,]
13
data1[is.na(data1)] <- 0
14
###################delete zero#################
15
nz=c()
16
for (i in 1:nrow(data1)){
17
  nz=c(nz,sum(data1[i,]==0))
18
}
19
zerorate=nz/ncol(data1)
20
data2=data1[zerorate<0.2,]
21
###############impute######################
22
23
data3=t(data2)
24
data3=data.frame(data3)
25
data3[data3==0]=NA
26
data4<-impute(data3)
27
##########normalize##############
28
data5=t(data4)
29
data6=data5
30
data7=t(data6)
31
data8=scale(data7, center = T, scale = T)
32
data9=t(data8)
33
write.csv(data9,'brcatest_go.csv')
34