|
26
|
1 library(ggplot2)
|
|
|
2
|
|
|
3 args <- commandArgs(trailingOnly = TRUE)
|
|
|
4
|
|
107
|
5 mutations.by.id.file = args[1]
|
|
|
6 absent.aa.by.id.file = args[2]
|
|
|
7 genes = strsplit(args[3], ",")[[1]]
|
|
|
8 genes = c(genes, "")
|
|
|
9 outdir = args[4]
|
|
|
10
|
|
26
|
11
|
|
56
|
12 print("---------------- read input ----------------")
|
|
|
13
|
|
107
|
14 mutations.by.id = read.table(mutations.by.id.file, sep="\t", fill=T, header=T, quote="")
|
|
|
15 absent.aa.by.id = read.table(absent.aa.by.id.file, sep="\t", fill=T, header=T, quote="")
|
|
26
|
16
|
|
107
|
17 for(gene in genes){
|
|
|
18
|
|
|
19 if(gene == ""){
|
|
|
20 mutations.by.id.gene = mutations.by.id[!grepl("unmatched", mutations.by.id$best_match),]
|
|
|
21 absent.aa.by.id.gene = absent.aa.by.id[!grepl("unmatched", absent.aa.by.id$best_match),]
|
|
|
22 } else {
|
|
|
23 mutations.by.id.gene = mutations.by.id[grepl(paste("^", gene, sep=""), mutations.by.id$best_match),]
|
|
|
24 absent.aa.by.id.gene = absent.aa.by.id[grepl(paste("^", gene, sep=""), absent.aa.by.id$best_match),]
|
|
|
25 }
|
|
|
26 if(nrow(mutations.by.id.gene) == 0){
|
|
|
27 next
|
|
|
28 }
|
|
108
|
29
|
|
|
30 print(paste("nrow", gene, nrow(absent.aa.by.id.gene)))
|
|
|
31
|
|
107
|
32 mutations.at.position = colSums(mutations.by.id.gene[,-c(1,2)])
|
|
|
33 aa.at.position = colSums(absent.aa.by.id.gene[,-c(1,2,3,4)])
|
|
56
|
34
|
|
107
|
35 dat_freq = mutations.at.position / aa.at.position
|
|
|
36 dat_dt = data.frame(i=1:length(dat_freq), freq=dat_freq)
|
|
|
37
|
|
|
38 print("---------------- plot ----------------")
|
|
57
|
39
|
|
107
|
40 m = ggplot(dat_dt, aes(x=i, y=freq)) + theme(axis.text.x = element_text(angle = 90, hjust = 1))
|
|
|
41 m = m + geom_bar(stat="identity", colour = "black", fill = "darkgrey", alpha=0.8) + scale_x_continuous(breaks=1:length(dat_freq), labels=1:length(dat_freq))
|
|
|
42 m = m + annotate("segment", x = 0.5, y = -0.05, xend=26.5, yend=-0.05, colour="darkgreen", size=1) + annotate("text", x = 13, y = -0.1, label="FR1")
|
|
|
43 m = m + annotate("segment", x = 26.5, y = -0.07, xend=38.5, yend=-0.07, colour="darkblue", size=1) + annotate("text", x = 32.5, y = -0.15, label="CDR1")
|
|
|
44 m = m + annotate("segment", x = 38.5, y = -0.05, xend=55.5, yend=-0.05, colour="darkgreen", size=1) + annotate("text", x = 47, y = -0.1, label="FR2")
|
|
|
45 m = m + annotate("segment", x = 55.5, y = -0.07, xend=65.5, yend=-0.07, colour="darkblue", size=1) + annotate("text", x = 60.5, y = -0.15, label="CDR2")
|
|
|
46 m = m + annotate("segment", x = 65.5, y = -0.05, xend=104.5, yend=-0.05, colour="darkgreen", size=1) + annotate("text", x = 85, y = -0.1, label="FR3")
|
|
|
47 m = m + expand_limits(y=c(-0.1,1)) + xlab("AA position") + ylab("Frequency") + ggtitle(paste(gene, "AA mutation frequency"))
|
|
57
|
48
|
|
107
|
49 print("---------------- write/print ----------------")
|
|
108
|
50
|
|
109
|
51 png(filename=paste(outdir, "/aa_histogram_", gene, ".png", sep=""), width=1280, height=720)
|
|
|
52 print(m)
|
|
|
53 dev.off()
|
|
|
54
|
|
108
|
55 dat.sums = data.frame(index=1:length(mutations.at.position), mutations.at.position=mutations.at.position, aa.at.position=aa.at.position)
|
|
|
56
|
|
|
57 write.table(dat.sums, paste(outdir, "/aa_histogram_sum_", gene, ".txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)
|
|
|
58 write.table(mutations.by.id.gene, paste(outdir, "/aa_histogram_count_", gene, ".txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)
|
|
|
59 write.table(absent.aa.by.id.gene, paste(outdir, "/aa_histogram_absent_", gene, ".txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)
|
|
107
|
60 write.table(dat_dt, paste(outdir, "/aa_histogram_", gene, ".txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)
|
|
|
61 }
|
