From aed11759d14ebf893dae11282a1c7cdc863c6ac4 Mon Sep 17 00:00:00 2001
From: Pariksheet Nanda <pariksheet.nanda@uconn.edu>
Date: Tue, 7 Feb 2017 13:54:39 -0500
Subject: [PATCH] ENH: Control comparison plots (work in progress)

- Need to create p-value column using control references instead of
  cells.
- Need to display on plot in "fold change" e.g. relative to LacZ.
---
 plots.R | 35 ++++++++++++++++++++++++++---------
 1 file changed, 26 insertions(+), 9 deletions(-)

diff --git a/plots.R b/plots.R
index bcda9ba..2e613f7 100644
--- a/plots.R
+++ b/plots.R
@@ -3,13 +3,12 @@
 suppressPackageStartupMessages({
     library(ggplot2)
     library(dplyr)
+    library(readr)
 })
 
-wells_all <- read.csv("rnai-p_values.csv") %>%
+wells_all <- read_csv("rnai-p_values.csv") %>%
     mutate(significant = p_value < 0.05)
 binwidth <- 0.025
-## Bring RNAi to the front of the level list.
-wells_all$type <- relevel(wells_all$type, "RNAi")
 
 theme_set(theme_bw())
 
@@ -20,7 +19,7 @@ plot_save <- function(prefix, dir = "plots",
                      units = "in"))
 }
 
-ggplot(filter(wells_all, type == "RNAi"),
+ggplot(filter(wells_all, is.na(control)),
        aes(coloc, fill = significant)) +
     geom_histogram(binwidth = binwidth) +
     scale_fill_manual(values = c("grey", "blue")) +
@@ -30,15 +29,33 @@ plot_save("rnai-hist-signif")
 
 ggplot(filter(wells_all,
               !is.na(control),
-              plate == "160415_015529-V"),
-       aes(coloc, fill = significant)) +
+              plate == "504"),
+       aes(coloc, fill = symbol)) +
     geom_histogram(binwidth = binwidth) +
     facet_wrap(~ control) +
-    scale_fill_manual(values = c("grey", "blue")) +
-    ggtitle(paste("Histogram of control wells with",
-                  "significant overlap (p-value < 5%)")) +
+    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() +