plots.R

## See the distribution of % Overlap vs. wells

suppressPackageStartupMessages({
    library(ggplot2)
    library(dplyr)
    library(readr)
})

wells_all <- read_csv("rnai-p_values.csv") %>%
    mutate(significant = p_value < 0.05)
binwidth <- 0.025

theme_set(theme_bw())

plot_save <- function(prefix, dir = "plots",
                      extensions = c("png", "pdf", "svg")) {
    invisible(sapply(file.path(dir, paste(prefix, extensions, sep = ".")),
                     ggplot2::ggsave, width = 8, height = 5,
                     units = "in"))
}

ggplot(filter(wells_all, is.na(control)),
       aes(coloc, fill = significant)) +
    geom_histogram(binwidth = binwidth) +
    scale_fill_manual(values = c("grey", "blue")) +
    ggtitle("Histogram of RNAi wells with significant overlap (p-value < 5%)") +
    xlab("% colocalization")
plot_save("rnai-hist-signif")

ggplot(filter(wells_all,
              !is.na(control),
              plate == "504"),
       aes(coloc, fill = symbol)) +
    geom_histogram(binwidth = binwidth) +
    facet_wrap(~ control) +
    ggtitle(paste("Histogram of control wells in plate 504")) +
    xlab("% colocalization")

ggplot(filter(wells_all, !is.na(control)) %>%
       group_by(well) %>% mutate(ymin = min(coloc), ymax = max(coloc),
                                 plate = factor(plate)),
       aes(well, coloc, color = plate)) +
    geom_errorbar(aes(ymin = ymin, ymax = ymax), color = "gray") +
    geom_jitter(aes(shape = plate)) +
    scale_color_brewer(palette = "Set1") +
    ggtitle(paste("Scatter plot of control well colocalization across plates")) +
    xlab("well") + ylab("% colocalization")

ggplot(filter(wells_all, !is.na(control)) %>%
       group_by(well) %>% mutate(ymin = min(p_value), ymax = max(p_value),
                                 plate = factor(plate)),
       aes(well, p_value, color = plate)) +
    geom_errorbar(aes(ymin = ymin, ymax = ymax, color = control)) +
    geom_jitter(aes(shape = plate, size = n)) +
    scale_color_brewer(palette = "Set1") +
    ggtitle(paste("Scatter plot of control well p-values across plates")) +
    xlab("well") + ylab("p-value")

ggplot(wells_all, aes(coloc, color = type)) +
    geom_freqpoly(binwidth = binwidth) +
    scale_y_log10() +
    facet_wrap(~ type) +
    ggtitle("Histogram of all well types") +
    xlab("% colocalization")
plot_save("all-hist")

ggplot(wells_all, aes(coloc, color = type)) +
    geom_density() +
    facet_wrap(~ type) +
    ggtitle("Density plot of wells to show distribution") +
    xlab("% colocalization")
plot_save("all-density")