view aa_histogram.r @ 57:16c7fc1c4bf8 draft

Uploaded

library(ggplot2)

args <- commandArgs(trailingOnly = TRUE)

input = args[1]
outfile = args[2]

print("---------------- read input ----------------")

dat = read.table(input, sep="\t", fill=T, header=T, quote="")

print("---------------- as numeric ----------------")

mutations.at.position = as.numeric(dat[1,])
aa.at.position = as.numeric(dat[2,])

print("---------------- freq data.frame ----------------")

dat_freq = mutations.at.position / aa.at.position
dat_dt = data.frame(i=1:length(dat_freq), freq=dat_freq)

print("---------------- weird stuff ----------------")

options(width=220)

print(dat[,20:40])

print(dat_dt) #need this or it will fail????

print("---------------- plot ----------------")

m = ggplot(dat_dt, aes(x=i, y=freq)) + theme(axis.text.x = element_text(angle = 90, hjust = 1))
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))
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")
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")
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")
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")
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")
m = m + expand_limits(y=c(-0.1,1)) + xlab("AA position") + ylab("Frequency") + ggtitle("AA mutation frequency")

print("---------------- write/print ----------------")

write.table(dat_dt, paste(dirname(outfile), "/aa_histogram.txt", sep=""), sep="\t",quote=F,row.names=F,col.names=T)

png(filename=outfile, width=1280, height=720)
print(m)
dev.off()