inst/workflow/roc/code_plot/s3-plotdiag_bin_est.r

### We can also plot against estimated values from multinomial logistic
### regression, i.e. no measurement error.

rm(list = ls())

suppressMessages(library(cubfits, quietly = TRUE))

### Load environment and set data.
source("00-set_env.r")
fn.in <- paste(prefix$data, "pre_process.rda", sep = "")
load(fn.in)
fn.in <- paste(prefix$data, "init_", model, ".rda", sep = "")
load(fn.in)
fn.in <- paste(prefix$data, "simu_true_", model, ".rda", sep = "")
if(file.exists(fn.in)){
  load(fn.in)
} else{
  stop(paste(fn.in, " is not found.", sep = ""))
}

### Arrange data.
phi.Obs.lim <- range(phi.Obs)
aa.names <- names(reu13.df.obs)
ret.EPhi <- prop.bin.roc(reu13.df.obs, EPhi,
                         bin.class = run.info$bin.class)
predict.roc <- prop.model.roc(fitlist, phi.Obs.lim)

### Fix xlim at log10 scale.
lim.bin <- range(log10(ret.EPhi[[1]]$center))
xlim <- c(lim.bin[1] - diff(lim.bin) / 4, lim.bin[2] + diff(lim.bin) / 4)

### Plot bin and model for measurements.
fn.out <- paste(prefix$plot.diag, "bin_est_EPhi.pdf", sep = "")
pdf(fn.out, width = 16, height = 11)
  mat <- matrix(c(rep(1, 5), 2:21, rep(22, 5)),
                nrow = 6, ncol = 5, byrow = TRUE)
  mat <- cbind(rep(23, 6), mat, rep(24, 6))
  nf <- layout(mat, c(3, rep(8, 5), 2), c(3, 8, 8, 8, 8, 3), respect = FALSE)
  ### Plot title.
  par(mar = c(0, 0, 0, 0))
  plot(NULL, NULL, xlim = c(0, 1), ylim = c(0, 1), axes = FALSE)
  text(0.5, 0.6, paste(workflow.name, ", bin: true phi", sep = ""))
  text(0.5, 0.4, date(), cex = 0.6)

  ### Plot results.
  for(i.aa in 1:length(aa.names)){
    tmp.obs <- ret.EPhi[[i.aa]]
    tmp.roc <- predict.roc[[i.aa]]
    plotbin(tmp.obs, tmp.roc, main = "", xlab = "", ylab = "",
            lty = 2, axes = FALSE, xlim = xlim)
    box()
    text(mean(xlim), 1, aa.names[i.aa], cex = 1.5)
    if(i.aa %in% c(1, 6, 11, 16)){
      axis(2)
    }
    if(i.aa %in% 16:19){
      axis(1)
    }
    if(i.aa %in% 1:5){
      axis(3)
    }
    if(i.aa %in% c(5, 10, 15)){
      axis(4)
    }
    axis(1, tck = 0.02, labels = FALSE)
    axis(2, tck = 0.02, labels = FALSE)
    axis(3, tck = 0.02, labels = FALSE)
    axis(4, tck = 0.02, labels = FALSE)
  }

  ### For cases with less aa.
  i.aa <- 19 - i.aa
  if(i.aa > 0){
    for(i.plot in 1:i.aa){
      plot(NULL, NULL, xlim = c(0, 1), ylim = c(0, 1),
           xlab = "", ylab = "", main = "", axes = FALSE)
    }
  }

  ### Add histogram.
  p.1 <- hist(log10(EPhi), nclass = 40, plot = FALSE)
  hist.ylim <- range(p.1$counts)
  hist.ylim[2] <- hist.ylim[2] + 0.2 * diff(hist.ylim)
  plot(p.1, xlim = xlim, ylim = hist.ylim, main = "", xlab = "", ylab = "",
       axes = FALSE)
  axis(1)
  axis(4)

  ### Add label.
  model.label <- c("Logistic Regression")
  model.lty <- 2
  legend(xlim[1], hist.ylim[2],
         model.label, lty = model.lty, box.lty = 0, cex = 0.8)

  ### Plot xlab.
  plot(NULL, NULL, xlim = c(0, 1), ylim = c(0, 1), axes = FALSE)
  text(0.5, 0.5,
       expression(paste(log[10], "(True Production Rate)", sep = "")))

  ### Plot ylab.
  plot(NULL, NULL, xlim = c(0, 1), ylim = c(0, 1), axes = FALSE)
  text(0.5, 0.5, "Codon Frequency", srt = 90)
dev.off()

Try the cubfits package in your browser

Any scripts or data that you put into this service are public.

cubfits documentation built on May 2, 2019, 4:08 a.m.