annotate aa_histogram.r @ 58:8bb4d6009e08 draft

Uploaded
author davidvanzessen
date Tue, 01 Mar 2016 06:53:40 -0500
parents cb66d6dd1e66
children 67a9ddf6a8f5
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
26
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
1 library(ggplot2)
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
2
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
3 args <- commandArgs(trailingOnly = TRUE)
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
4
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
5 input = args[1]
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
6 outfile = args[2]
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
7
56
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
8 print("---------------- read input ----------------")
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
9
49
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
10 dat = read.table(input, sep="\t", fill=T, header=T, quote="")
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
11
56
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
12 print("---------------- as numeric ----------------")
49
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
13
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
14 mutations.at.position = as.numeric(dat[1,])
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
15 aa.at.position = as.numeric(dat[2,])
26
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
16
56
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
17 print("---------------- freq data.frame ----------------")
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
18
49
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
19 dat_freq = mutations.at.position / aa.at.position
26
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
20 dat_dt = data.frame(i=1:length(dat_freq), freq=dat_freq)
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
21
56
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
22 print("---------------- weird stuff ----------------")
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
23
49
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
24 options(width=220)
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
25
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
26 print(dat[,20:40])
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
27
51
d4e72eeea640 Uploaded
davidvanzessen
parents: 49
diff changeset
28 print(dat_dt) #need this or it will fail????
d4e72eeea640 Uploaded
davidvanzessen
parents: 49
diff changeset
29
56
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
30 print("---------------- plot ----------------")
a4317b006d70 Uploaded
davidvanzessen
parents: 51
diff changeset
31
26
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
32 m = ggplot(dat_dt, aes(x=i, y=freq)) + theme(axis.text.x = element_text(angle = 90, hjust = 1))
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
33 m = m + geom_histogram(stat="identity", colour = "black", fill = "darkgrey", alpha=0.8) + scale_x_continuous(breaks=1:length(dat_freq), labels=1:length(dat_freq))
49
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
34 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")
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
35 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")
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
36 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")
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
37 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")
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
38 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")
5c6b9e99d576 Uploaded
davidvanzessen
parents: 29
diff changeset
39 m = m + expand_limits(y=c(-0.1,1)) + xlab("AA position") + ylab("Frequency") + ggtitle("AA mutation frequency")
57
cb66d6dd1e66 Uploaded
davidvanzessen
parents: 56
diff changeset
40
cb66d6dd1e66 Uploaded
davidvanzessen
parents: 56
diff changeset
41 print("---------------- write/print ----------------")
cb66d6dd1e66 Uploaded
davidvanzessen
parents: 56
diff changeset
42
26
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
43 write.table(dat_dt, paste(dirname(outfile), "/aa_histogram.txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
44
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
45 png(filename=outfile, width=1280, height=720)
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
46 print(m)
2433a1e110e1 Uploaded
davidvanzessen
parents:
diff changeset
47 dev.off()