|
0
|
1 #!/usr/bin/env Rscript
|
|
|
2
|
|
|
3 suppressPackageStartupMessages(library("optparse"))
|
|
|
4
|
|
|
5 option_list <- list(
|
|
3
|
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")
|
|
0
|
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
|
|
3
|
16 get_num_components = function(components_data)
|
|
0
|
17 {
|
|
|
18 # Get the max of the number_comp column.
|
|
3
|
19 number_comp = components_data[, 3]
|
|
|
20 num_components <- max(number_comp, na.rm=TRUE);
|
|
|
21 num_components
|
|
0
|
22 }
|
|
|
23
|
|
3
|
24 get_pi_mu_var = function(components_data, num_components)
|
|
|
25 {
|
|
0
|
26 # FixMe: enhance this to generically handle any integer value for num_components.
|
|
3
|
27 if (num_components == 1)
|
|
|
28 {
|
|
0
|
29 pi <- c(components_data[1, 9]);
|
|
|
30 mu <- c(components_data[1, 7]);
|
|
|
31 var <- c(components_data[1, 8]);
|
|
3
|
32 }
|
|
|
33 else if (num_components == 2)
|
|
|
34 {
|
|
0
|
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]);
|
|
3
|
38 }
|
|
|
39 else if (num_components == 3)
|
|
|
40 {
|
|
0
|
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]);
|
|
3
|
44 }
|
|
|
45 else if (num_components == 4)
|
|
|
46 {
|
|
0
|
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]);
|
|
3
|
50 }
|
|
0
|
51 return = c(pi, mu, var)
|
|
|
52 return
|
|
|
53 }
|
|
|
54
|
|
3
|
55 plot_ks<-function(kaks_input, output, pi, mu, var)
|
|
|
56 {
|
|
|
57 # Start PDF device driver to save charts to output.
|
|
|
58 pdf(file=output, bg="white")
|
|
|
59 # Change bin width
|
|
|
60 bin <- 0.05 * seq(0, 40);
|
|
|
61 kaks <- read.table(file=kaks_input, header=T);
|
|
|
62 kaks <- kaks[kaks$Ks<2,];
|
|
|
63 h.kst <- hist(kaks$Ks, breaks=bin, plot=F);
|
|
|
64 nc <- h.kst$counts;
|
|
|
65 vx <- h.kst$mids;
|
|
|
66 ntot <- sum(nc);
|
|
|
67 # Set margin for plot bottom, left top, right.
|
|
|
68 par(mai=c(0.5, 0.5, 0, 0));
|
|
|
69 # Plot dimension in inches.
|
|
|
70 par(pin=c(2.5, 2.5));
|
|
|
71 g <- calculate_fitted_density(pi, mu, var);
|
|
|
72 h <- ntot * 2.5 / sum(g);
|
|
|
73 vx <- seq(1, 100) * 0.02;
|
|
|
74 ymax <- max(nc) + 5;
|
|
|
75 barplot(nc, space=0.25, offset=0, width=0.04, xlim=c(0,2), ylim=c(0, ymax));
|
|
|
76 # Add x-axis.
|
|
|
77 axis(1);
|
|
|
78 color <- c('green', 'blue', 'black', 'red');
|
|
|
79 for (i in 1:length(mu))
|
|
|
80 {
|
|
|
81 lines(vx, g[,i] * h, lwd=2, col=color[i]);
|
|
|
82 }
|
|
0
|
83 };
|
|
|
84
|
|
3
|
85 calculate_fitted_density <- function(pi, mu, var)
|
|
|
86 {
|
|
|
87 comp <- length(pi);
|
|
|
88 var <- var/mu^2;
|
|
|
89 mu <- log(mu);
|
|
|
90 #calculate lognormal density
|
|
|
91 vx <- seq(1, 100) * 0.02;
|
|
|
92 fx <- matrix(0, 100, comp);
|
|
|
93 for (i in 1:100)
|
|
|
94 {
|
|
|
95 for (j in 1:comp)
|
|
|
96 {
|
|
|
97 fx[i, j] <- pi[j] * dlnorm(vx[i], meanlog=mu[j], sdlog=(sqrt(var[j])));
|
|
|
98 };
|
|
|
99 };
|
|
|
100 fx;
|
|
0
|
101 }
|
|
|
102
|
|
3
|
103 # Read in the components data.
|
|
|
104 components_data <- read.delim(opt$components_input, header=TRUE);
|
|
|
105 # Get the number of components.
|
|
|
106 num_components <- get_num_components(components_data)
|
|
0
|
107
|
|
|
108 # Set pi, mu, var.
|
|
3
|
109 items <- get_pi_mu_var(components_data, num_components);
|
|
|
110 pi <- items[1];
|
|
|
111 mu <- items[2];
|
|
|
112 var <- items[3];
|
|
0
|
113
|
|
|
114 # Plot the output.
|
|
3
|
115 plot_ks(opt$kaks_input, opt$output, pi, mu, var);
|
