|
4
|
1 args <- commandArgs(trailingOnly = TRUE)
|
|
|
2
|
|
|
3 naive.file = args[1]
|
|
|
4 shm.file = args[2]
|
|
|
5 output.file.ca = args[3]
|
|
|
6 output.file.cg = args[4]
|
|
|
7 output.file.cm = args[5]
|
|
|
8
|
|
|
9 naive = read.table(naive.file, sep="\t", header=T, quote="", fill=T)
|
|
|
10 shm.merge = read.table(shm.file, sep="\t", header=T, quote="", fill=T)
|
|
|
11
|
|
|
12
|
|
|
13 final = merge(naive, shm.merge[,c("Sequence.ID", "best_match")], by.x="ID", by.y="Sequence.ID")
|
|
|
14 print(paste("nrow final:", nrow(final)))
|
|
|
15 names(final)[names(final) == "best_match"] = "Sample"
|
|
|
16 final.numeric = final[,sapply(final, is.numeric)]
|
|
|
17 final.numeric[is.na(final.numeric)] = 0
|
|
|
18 final[,sapply(final, is.numeric)] = final.numeric
|
|
|
19
|
|
|
20 final.ca = final[grepl("^ca", final$Sample),]
|
|
|
21 final.cg = final[grepl("^cg", final$Sample),]
|
|
|
22 final.cm = final[grepl("^cm", final$Sample),]
|
|
|
23
|
|
|
24 if(nrow(final.ca) > 0){
|
|
|
25 final.ca$Replicate = 1
|
|
|
26 }
|
|
|
27
|
|
|
28 if(nrow(final.cg) > 0){
|
|
|
29 final.cg$Replicate = 1
|
|
|
30 }
|
|
|
31
|
|
|
32 if(nrow(final.cm) > 0){
|
|
|
33 final.cm$Replicate = 1
|
|
|
34 }
|
|
|
35
|
|
|
36 #print(paste("nrow final:", nrow(final)))
|
|
|
37 #final2 = final
|
|
|
38 #final2$Sample = gsub("[0-9]", "", final2$Sample)
|
|
|
39 #final = rbind(final, final2)
|
|
|
40 #final$Replicate = 1
|
|
|
41
|
|
|
42 write.table(final.ca, output.file.ca, quote=F, sep="\t", row.names=F, col.names=T)
|
|
|
43 write.table(final.cg, output.file.cg, quote=F, sep="\t", row.names=F, col.names=T)
|
|
|
44 write.table(final.cm, output.file.cm, quote=F, sep="\t", row.names=F, col.names=T)
|
|
|
45
|
