Mercurial > repos > greg > plant_tribes_ks_distribution
view ks_distribution.R @ 0:c5846258c458 draft
Uploaded
author | greg |
---|---|
date | Thu, 08 Jun 2017 12:55:49 -0400 |
parents | |
children | 56f42cc1dd58 |
line wrap: on
line source
#!/usr/bin/env Rscript suppressPackageStartupMessages(library("optparse")) option_list <- list( make_option(c("-c", "--components_input"), action="store", dest="components_input", help="Ks significant components input dataset"), make_option(c("-k", "--kaks_input"), action="store", dest="kaks_input", help="KaKs analysis input dataset"), make_option(c("-o", "--output"), action="store", dest="output", help="Output dataset") ) parser <- OptionParser(usage="%prog [options] file", option_list=option_list) args <- parse_args(parser, positional_arguments=TRUE) opt <- args$options get_num_components = function(components_data) { # Get the max of the number_comp column. number_comp = components_data[, 3] num_components <- max(number_comp, na.rm=TRUE) return(num_components) } get_pi_mu_var = function(components_data, num_components) { # FixMe: enhance this to generically handle any integer value for num_components. if (num_components == 1) { pi <- c(components_data[1, 9]) mu <- c(components_data[1, 7]) var <- c(components_data[1, 8]) } else if (num_components == 2) { pi <- c(components_data[2, 9], components_data[3, 9]) mu <- c(components_data[2, 7], components_data[3, 7]) var <- c(components_data[2, 8], components_data[3, 8]) } else if (num_components == 3) { pi <- c(components_data[4, 9], components_data[5, 9], components_data[6, 9]) mu <- c(components_data[4, 7], components_data[5, 7], components_data[6, 7]) var <- c(components_data[4, 8], components_data[5, 8], components_data[6, 8]) } else if (num_components == 4) { pi <- c(components_data[7, 9], components_data[8, 9], components_data[9, 9], components_data[10, 9]) mu <- c(components_data[7, 7], components_data[8, 7], components_data[9, 7], components_data[10, 7]) var <- c(components_data[7, 8], components_data[8, 8], components_data[9, 8], components_data[10, 8]) } else if (num_components == 5) { pi <- c(components_data[11, 9], components_data[12, 9], components_data[13, 9], components_data[14, 9], components_data[15, 9]) mu <- c(components_data[11, 7], components_data[12, 7], components_data[13, 7], components_data[14, 7], components_data[15, 7]) var <- c(components_data[11, 8], components_data[12, 8], components_data[13, 8], components_data[14, 8], components_data[15, 8]) } else if (num_components == 6) { 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]) 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]) 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]) } results = c(pi, mu, var) return(results) } plot_ks<-function(kaks_input, output, pi, mu, var) { # Start PDF device driver to save charts to output. pdf(file=output, bg="white") kaks <- read.table(file=kaks_input, header=T) max_ks <- max(kaks$Ks, na.rm=TRUE) # Change bin width max_bin_range <- as.integer(max_ks / 0.05) bin <- 0.05 * seq(0, (max_bin_range + 1 )) kaks <- kaks[kaks$Ks<max_ks,] h.kst <- hist(kaks$Ks, breaks=bin, plot=F) nc <- h.kst$counts vx <- h.kst$mids ntot <- sum(nc) # Set margin for plot bottom, left top, right. par(mai=c(0.5, 0.5, 0, 0)) # Plot dimension in inches. par(pin=c(3.0, 3.0)) g <- calculate_fitted_density(pi, mu, var, max_ks) h <- ntot * 1.5 / sum(g) vx <- seq(1, 100) * (max_ks / 100) ymax <- max(nc) barplot(nc, space=0.25, offset=0, width=0.04, xlim=c(0, max_ks), ylim=c(0, ymax), col="lightpink1", border="lightpink3") # Add x-axis. axis(1) color <- c('red', 'yellow','green','black','blue', 'darkorange' ) for (i in 1:length(mu)) { lines(vx, g[,i] * h, lwd=2, col=color[i]) } } calculate_fitted_density <- function(pi, mu, var, max_ks) { comp <- length(pi) var <- var/mu^2 mu <- log(mu) # Calculate lognormal density. vx <- seq(1, 100) * (max_ks / 100) fx <- matrix(0, 100, comp) for (i in 1:100) { for (j in 1:comp) { fx[i, j] <- pi[j] * dlnorm(vx[i], meanlog=mu[j], sdlog=(sqrt(var[j]))) if (is.nan(fx[i,j])) fx[i,j]<-0 } } return(fx) } # Read in the components data. components_data <- read.delim(opt$components_input, header=TRUE) # Get the number of components. num_components <- get_num_components(components_data) # Set pi, mu, var. items <- get_pi_mu_var(components_data, num_components) if (num_components == 1) { pi <- items[1] mu <- items[2] var <- items[3] } if (num_components == 2) { pi <- items[1:2] mu <- items[3:4] var <- items[5:6] } if (num_components == 3) { pi <- items[1:3] mu <- items[4:6] var <- items[7:9] } if (num_components == 4) { pi <- items[1:4] mu <- items[5:8] var <- items[9:12] } if (num_components == 5) { pi <- items[1:5] mu <- items[6:10] var <- items[11:15] } if (num_components == 6) { pi <- items[1:6] mu <- items[7:12] var <- items[13:18] } # Plot the output. plot_ks(opt$kaks_input, opt$output, pi, mu, var)