annotate aa_histogram.r @ 89:3c9d4d976c47 draft

"planemo upload commit 8b702d0cf8f2355307eb8f95989f3d2488b2c963"
author rhpvorderman
date Fri, 05 Nov 2021 13:41:03 +0000
parents 729738462297
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
83
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
1 library(ggplot2)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
2
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
3 args <- commandArgs(trailingOnly = TRUE)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
4
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
5 mutations.by.id.file = args[1]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
6 absent.aa.by.id.file = args[2]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
7 genes = strsplit(args[3], ",")[[1]]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
8 genes = c(genes, "")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
9 outdir = args[4]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
10
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
11
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
12 print("---------------- read input ----------------")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
13
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
14 mutations.by.id = read.table(mutations.by.id.file, sep="\t", fill=T, header=T, quote="")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
15 absent.aa.by.id = read.table(absent.aa.by.id.file, sep="\t", fill=T, header=T, quote="")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
16
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
17 for(gene in genes){
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
18 graph.title = paste(gene, "AA mutation frequency")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
19 if(gene == ""){
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
20 mutations.by.id.gene = mutations.by.id[!grepl("unmatched", mutations.by.id$best_match),]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
21 absent.aa.by.id.gene = absent.aa.by.id[!grepl("unmatched", absent.aa.by.id$best_match),]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
22
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
23 graph.title = "AA mutation frequency all"
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
24 } else {
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
25 mutations.by.id.gene = mutations.by.id[grepl(paste("^", gene, sep=""), mutations.by.id$best_match),]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
26 absent.aa.by.id.gene = absent.aa.by.id[grepl(paste("^", gene, sep=""), absent.aa.by.id$best_match),]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
27 }
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
28 print(paste("nrow", gene, nrow(absent.aa.by.id.gene)))
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
29 if(nrow(mutations.by.id.gene) == 0){
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
30 next
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
31 }
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
32
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
33 mutations.at.position = colSums(mutations.by.id.gene[,-c(1,2)])
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
34 aa.at.position = colSums(absent.aa.by.id.gene[,-c(1,2,3,4)])
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
35
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
36 dat_freq = mutations.at.position / aa.at.position
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
37 dat_freq[is.na(dat_freq)] = 0
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
38 dat_dt = data.frame(i=1:length(dat_freq), freq=dat_freq)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
39
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
40
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
41 print("---------------- plot ----------------")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
42
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
43 m = ggplot(dat_dt, aes(x=i, y=freq)) + theme(axis.text.x = element_text(angle = 90, hjust = 1), text = element_text(size=13, colour="black"))
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
44 m = m + geom_bar(stat="identity", colour = "black", fill = "darkgrey", alpha=0.8) + scale_x_continuous(breaks=dat_dt$i, labels=dat_dt$i)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
45 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")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
46 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")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
47 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")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
48 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")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
49 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")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
50 m = m + expand_limits(y=c(-0.1,1)) + xlab("AA position") + ylab("Frequency") + ggtitle(graph.title)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
51 m = m + theme(panel.background = element_rect(fill = "white", colour="black"), panel.grid.major.y = element_line(colour = "black"), panel.grid.major.x = element_blank())
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
52 #m = m + scale_colour_manual(values=c("black"))
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
53
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
54 print("---------------- write/print ----------------")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
55
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
56
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
57 dat.sums = data.frame(index=1:length(mutations.at.position), mutations.at.position=mutations.at.position, aa.at.position=aa.at.position)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
58
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
59 write.table(dat.sums, paste(outdir, "/aa_histogram_sum_", gene, ".txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
60 write.table(mutations.by.id.gene, paste(outdir, "/aa_histogram_count_", gene, ".txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
61 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)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
62 write.table(dat_dt, paste(outdir, "/aa_histogram_", gene, ".txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
63
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
64 png(filename=paste(outdir, "/aa_histogram_", gene, ".png", sep=""), width=1280, height=720)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
65 print(m)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
66 dev.off()
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
67
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
68 ggsave(paste(outdir, "/aa_histogram_", gene, ".pdf", sep=""), m, width=14, height=7)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 81
diff changeset
69 }