Add CCLE Pharmacology Analysis (greenelab#66)

* update colors

* generate pdf

* adjust indent

* add figures generated by ccle pharmacology R script

* add additional plots with and without BRAF included

* add all figures generated by ccle pharm analysis

* update function legend label

* add updated figures

* add fill manual

* update figures with fill manual

scripts/viz/ras_ccle_pharmacology.R

@@ -0,0 +1,161 @@
+# Gregory Way 2017
+# PanCancer Classifier
+# scripts/viz/ras_ccle_pharmacology.R
+#
+# Visualize correlations between Ras classifier score and two MEK inhibitor
+# activity (AZD6244 (Selumetinib) and PD-0325901)
+#
+# Usage: Run in command line
+#
+#     Rscript --vanilla scripts/viz/ras_ccle_pharmacology.R
+#
+# Output:
+# Overall and tissue-specific correlation plots between drug activity and
+# Ras classifier scores
+
+library(dplyr)
+library(ggplot2)
+library(ggpmisc)
+
+# Input Files
+pharm_file <- file.path("data", "pharmacology_predictions_ccle.tsv")
+pharm_full_df <- readr::read_tsv(pharm_file)
+
+# Output Files
+base_path <- file.path("figures", "cell_line")
+selum_file <- file.path(base_path, "selumetinib.pdf")
+selum_braf_file <- file.path(base_path, "selumetinib_braf.pdf")
+selum_tissue_file <- file.path(base_path, "selumetinib_tissues.pdf")
+selum_tissue_braf_file <- file.path(base_path, "selumetinib_tissues_braf.pdf")
+
+pd_file <- file.path(base_path, "pd_0325901.pdf")
+pd_braf_file <- file.path(base_path, "pd_0325901_braf.pdf")
+pd_tissue_file <- file.path(base_path, "pd_0325901_tissues.pdf")
+pd_tissue_braf_file <- file.path(base_path, "pd_0325901_tissues_braf.pdf")
+
+# Define Plotting Function
+plot_drug <- function(pharm_df, compound, tissues = NULL, include_braf = FALSE, 
+                      facet_tissue = TRUE, se = FALSE) {
+  # Output scatter plots with correlations, visualizing drug activity
+  # compared to Ras classifier Score
+  #
+  # Arguments:
+  # pharm_df - dataframe of compound activity by cell line with Ras/BRAF status
+  # compound - a specific compound to visualize
+  # tissues - a list of tissues to consider plotting specifically in facets
+  # include_braf - boolean to include BRAF in considering mutation status
+  # facet_tissue - boolean of tissues to determine to plot in facet_wrap
+  # se - boolean to plot standard error intervals in geom_smooth
+  #
+  # Output:
+  # Scatter plot with correlation information
+
+  pharm_subset_df <- pharm_df[pharm_df$Compound == compound, ]
+  if (!is.null(tissues)) {
+    pharm_subset_df <- pharm_subset_df %>%
+      dplyr::filter(tissue %in% focus_tissues)
+  }
+  if (include_braf) {
+    pharm_subset_df$ras_status[pharm_subset_df$BRAF_MUT == 1] <- 1
+    legend_label <- "Ras/BRAF Status"
+  } else {
+    legend_label <- "Ras Status"
+  }
+
+  if (compound == "AZD6244") {
+    compound <- "Selumetinib"
+  }
+
+  formula <- y ~ x
+
+  p <- ggplot(pharm_subset_df, aes(x = weight, y = ActArea,
+                                   color = as.factor(ras_status),
+                                   fill = as.factor(ras_status))) +
+    geom_point(alpha = 0.5, size = 2) +
+    scale_x_continuous(breaks = c(0, 0.5, 1),
+                       limits = c(-0.1, 1.1)) +
+    geom_smooth(method = "lm", se = se) +
+    geom_segment(aes(x = 0.5, y = -0.1, xend = 0.5, yend = 6),
+                 linetype = "dashed", color = "grey") +
+    scale_fill_manual(values = c("#377eb8", "#ff7f00"),
+                      name = legend_label,
+                      breaks = c(0, 1),
+                      labels = c("Wild-Type", "Mutant")) +
+    scale_color_manual(values = c("#377eb8", "#ff7f00"),
+                       name = legend_label,
+                       breaks = c(0, 1),
+                       labels = c("Wild-Type", "Mutant")) +
+    stat_poly_eq(aes(label = paste(..rr.label..)),
+                 label.x.npc = 0.17, label.y.npc = 0.92,
+                 formula = formula,
+                 parse = TRUE, size = 4, na.rm = TRUE,
+                 rr.digits = 1) +
+    stat_fit_glance(method = "lm", geom = "text",
+                    label.x.npc = 0.8, label.y.npc = 0.97,
+                    method.args = list(formula = formula), size = 4,
+                    aes(label = paste("P = ",
+                                      signif(..p.value.., digits = 1),
+                                      sep = ""))) +
+    xlab("Ras Classifier Score") +
+    ylab("Activity Area") +
+    ggtitle(compound, subtitle = "CCLE Response") + 
+    theme_bw() +
+    theme(axis.text = element_text(size = 12),
+          axis.title = element_text(size = 14),
+          plot.title = element_text(hjust = 0.5),
+          plot.subtitle = element_text(hjust = 0.5))
+
+  if (facet_tissue) {
+    p <- p + facet_wrap("tissue")
+  }
+
+  return(p)
+}
+
+# Set constants - tissues are set based on sample size and class size balance
+focus_tissues <- c("LUNG", "SKIN", "BREAST", "OVARY", "LARGE INTESTINE",
+                   "STOMACH")
+
+# Plot and output figures
+# Selumetinib all tissues - Ras only
+sel_fig <- plot_drug(pharm_full_df, "AZD6244", facet_tissue = FALSE, se = TRUE)
+ggplot2::ggsave(selum_file, plot = sel_fig, dpi = 600, height = 4, width = 5.25)
+
+# Selumetinib all tissues - Ras and BRAF
+sel_braf_fig <- plot_drug(pharm_full_df, "AZD6244", facet_tissue = FALSE,
+                          include_braf = TRUE, se = TRUE)
+ggplot2::ggsave(selum_braf_file, plot = sel_braf_fig, dpi = 600, height = 4,
+                width = 5.25)
+
+# PD-0325901 all tissues - Ras only
+p_fig <- plot_drug(pharm_full_df, "PD-0325901", facet_tissue = FALSE, se = TRUE)
+ggplot2::ggsave(pd_file, plot = p_fig, dpi = 600, height = 4,
+                width = 5.25)
+
+# PD-0325901 all tissues - Ras and BRAF
+p_braf_fig <- plot_drug(pharm_full_df, "PD-0325901", facet_tissue = FALSE,
+                        include_braf = TRUE, se = TRUE)
+ggplot2::ggsave(pd_braf_file, plot = p_braf_fig, dpi = 600,
+                height = 4.5, width = 8)
+
+# Selumetinib stratified by tissue - Ras only
+sel_tis_fig <- plot_drug(pharm_full_df, "AZD6244", tissues = focus_tissues)
+ggplot2::ggsave(selum_tissue_file, plot = sel_tis_fig, dpi = 600,
+                height = 4.5, width = 8)
+
+# Selumetinib stratified by tissue - Ras and BRAF
+sel_tis_braf_fig <- plot_drug(pharm_full_df, "AZD6244",
+                              tissues = focus_tissues, include_braf = TRUE)
+ggplot2::ggsave(selum_tissue_braf_file, plot = sel_tis_braf_fig, dpi = 600,
+                height = 4.5, width = 8)
+
+# PD-0325901 stratified by tissue - Ras only
+p_tis_fig <- plot_drug(pharm_full_df, "PD-0325901", tissues = focus_tissues)
+ggplot2::ggsave(pd_tissue_file, plot = p_tis_fig, dpi = 600,
+                height = 4.5, width = 8)
+
+# PD-0325901 stratified by tissue - Ras and BRAF
+p_tis_braf_fig <- plot_drug(pharm_full_df, "PD-0325901",
+                              tissues = focus_tissues, include_braf = TRUE)
+ggplot2::ggsave(pd_tissue_braf_file, plot = p_tis_braf_fig,
+                dpi = 600, height = 4.5, width = 8)