|
0
|
1 #!/usr/bin/env Rscript
|
|
|
2
|
|
|
3 suppressPackageStartupMessages(library("optparse"))
|
|
|
4
|
|
|
5 option_list <- list(
|
|
|
6 make_option(c("-c", "--components_input"), action="store", dest="components_input", help="Ks significant components input dataset"),
|
|
|
7 make_option(c("-k", "--kaks_input"), action="store", dest="kaks_input", help="KaKs analysis input dataset"),
|
|
|
8 make_option(c("-o", "--output"), action="store", dest="output", help="Output dataset")
|
|
|
9 )
|
|
|
10
|
|
|
11 parser <- OptionParser(usage="%prog [options] file", option_list=option_list)
|
|
|
12 args <- parse_args(parser, positional_arguments=TRUE)
|
|
|
13 opt <- args$options
|
|
|
14
|
|
|
15
|
|
|
16 get_num_components = function(components_data)
|
|
|
17 {
|
|
|
18 # Get the max of the number_comp column.
|
|
|
19 number_comp = components_data[, 3]
|
|
4
|
20 num_components <- max(number_comp, na.rm=TRUE)
|
|
|
21 return(num_components)
|
|
0
|
22 }
|
|
|
23
|
|
|
24 get_pi_mu_var = function(components_data, num_components)
|
|
|
25 {
|
|
|
26 # FixMe: enhance this to generically handle any integer value for num_components.
|
|
|
27 if (num_components == 1)
|
|
|
28 {
|
|
4
|
29 pi <- c(components_data[1, 9])
|
|
|
30 mu <- c(components_data[1, 7])
|
|
|
31 var <- c(components_data[1, 8])
|
|
0
|
32 }
|
|
|
33 else if (num_components == 2)
|
|
|
34 {
|
|
4
|
35 pi <- c(components_data[2, 9], components_data[3, 9])
|
|
|
36 mu <- c(components_data[2, 7], components_data[3, 7])
|
|
|
37 var <- c(components_data[2, 8], components_data[3, 8])
|
|
0
|
38 }
|
|
|
39 else if (num_components == 3)
|
|
|
40 {
|
|
4
|
41 pi <- c(components_data[4, 9], components_data[5, 9], components_data[6, 9])
|
|
|
42 mu <- c(components_data[4, 7], components_data[5, 7], components_data[6, 7])
|
|
|
43 var <- c(components_data[4, 8], components_data[5, 8], components_data[6, 8])
|
|
0
|
44 }
|
|
|
45 else if (num_components == 4)
|
|
|
46 {
|
|
4
|
47 pi <- c(components_data[7, 9], components_data[8, 9], components_data[9, 9], components_data[10, 9])
|
|
|
48 mu <- c(components_data[7, 7], components_data[8, 7], components_data[9, 7], components_data[10, 7])
|
|
|
49 var <- c(components_data[7, 8], components_data[8, 8], components_data[9, 8], components_data[10, 8])
|
|
0
|
50 }
|
|
4
|
51 else if (num_components == 5)
|
|
|
52 {
|
|
|
53 pi <- c(components_data[11, 9], components_data[12, 9], components_data[13, 9], components_data[14, 9], components_data[15, 9])
|
|
|
54 mu <- c(components_data[11, 7], components_data[12, 7], components_data[13, 7], components_data[14, 7], components_data[15, 7])
|
|
|
55 var <- c(components_data[11, 8], components_data[12, 8], components_data[13, 8], components_data[14, 8], components_data[15, 8])
|
|
|
56 }
|
|
|
57 else if (num_components == 6)
|
|
|
58 {
|
|
|
59 pi <- c(components_data[16, 9], components_data[17, 9], components_data[18, 9], components_data[19, 9], components_data[20, 9], components_data[21, 9])
|
|
|
60 mu <- c(components_data[16, 7], components_data[17, 7], components_data[18, 7], components_data[19, 7], components_data[20, 7], components_data[21, 7])
|
|
|
61 var <- c(components_data[16, 8], components_data[17, 8], components_data[18, 8], components_data[19, 8], components_data[20, 8], components_data[21, 8])
|
|
|
62 }
|
|
|
63 results = c(pi, mu, var)
|
|
|
64 return(results)
|
|
0
|
65 }
|
|
|
66
|
|
4
|
67 plot_ks<-function(kaks_input, output, pi, mu, var, max_ks)
|
|
0
|
68 {
|
|
|
69 # Start PDF device driver to save charts to output.
|
|
|
70 pdf(file=output, bg="white")
|
|
4
|
71 kaks <- read.table(file=kaks_input, header=T)
|
|
|
72 max_ks <- max(kaks$Ks, na.rm=TRUE)
|
|
0
|
73 # Change bin width
|
|
4
|
74 max_bin_range <- as.integer(max_ks / 0.05)
|
|
|
75 bin <- 0.05 * seq(0, max_bin_range)
|
|
|
76 kaks <- kaks[kaks$Ks<max_ks,];
|
|
|
77 h.kst <- hist(kaks$Ks, breaks=bin, plot=F)
|
|
|
78 nc <- h.kst$counts
|
|
|
79 vx <- h.kst$mids
|
|
|
80 ntot <- sum(nc)
|
|
0
|
81 # Set margin for plot bottom, left top, right.
|
|
4
|
82 par(mai=c(0.5, 0.5, 0, 0))
|
|
0
|
83 # Plot dimension in inches.
|
|
4
|
84 par(pin=c(2.5, 2.5))
|
|
|
85 g <- calculate_fitted_density(pi, mu, var)
|
|
|
86 h <- ntot * 2.5 / sum(g)
|
|
|
87 vx <- seq(1, 100) * 0.02
|
|
|
88 ymax <- max(nc) + 5
|
|
|
89 barplot(nc, space=0.25, offset=0, width=0.04, xlim=c(0, max_ks), ylim=c(0, ymax), col="lightpink1", border="lightpink3")
|
|
0
|
90 # Add x-axis.
|
|
4
|
91 axis(1)
|
|
|
92 color <- c('red', 'yellow','green','black','blue', 'darkorange' )
|
|
0
|
93 for (i in 1:length(mu))
|
|
|
94 {
|
|
4
|
95 lines(vx, g[,i] * h, lwd=2, col=color[i])
|
|
0
|
96 }
|
|
4
|
97 }
|
|
0
|
98
|
|
|
99 calculate_fitted_density <- function(pi, mu, var)
|
|
|
100 {
|
|
4
|
101 comp <- length(pi)
|
|
|
102 var <- var/mu^2
|
|
|
103 mu <- log(mu)
|
|
|
104 # Calculate lognormal density.
|
|
|
105 vx <- seq(1, 100) * 0.02
|
|
|
106 fx <- matrix(0, 100, comp)
|
|
0
|
107 for (i in 1:100)
|
|
|
108 {
|
|
|
109 for (j in 1:comp)
|
|
|
110 {
|
|
4
|
111 fx[i, j] <- pi[j] * dlnorm(vx[i], meanlog=mu[j], sdlog=(sqrt(var[j])))
|
|
|
112 if (is.nan(fx[i,j])) fx[i,j]<-0
|
|
|
113 }
|
|
|
114 }
|
|
|
115 return(fx)
|
|
0
|
116 }
|
|
|
117
|
|
|
118 # Read in the components data.
|
|
4
|
119 components_data <- read.delim(opt$components_input, header=TRUE)
|
|
0
|
120 # Get the number of components.
|
|
|
121 num_components <- get_num_components(components_data)
|
|
|
122
|
|
|
123 # Set pi, mu, var.
|
|
4
|
124 items <- get_pi_mu_var(components_data, num_components)
|
|
|
125 pi <- items[1:3]
|
|
|
126 mu <- items[4:6]
|
|
|
127 var <- items[7:9]
|
|
0
|
128
|
|
|
129 # Plot the output.
|
|
4
|
130 plot_ks(opt$kaks_input, opt$output, pi, mu, var, max_ks)
|
