From 98ee52b122f41db594494d5f9f77b4bafd71cb34 Mon Sep 17 00:00:00 2001
From: Elias Projahn <elias@johrpan.de>
Date: Sun, 14 Aug 2022 16:20:54 +0200
Subject: [PATCH] plots: Add rankings correlation plot

---
 NAMESPACE                        |   1 +
 R/plots.R                        | 106 +++++++++++++++++++++++++++++++
 man/plot_rankings_correlation.Rd |  32 ++++++++++
 3 files changed, 139 insertions(+)
 create mode 100644 man/plot_rankings_correlation.Rd

diff --git a/NAMESPACE b/NAMESPACE
index 8a00b14..60d16ad 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -22,6 +22,7 @@ export(plot_boxplot)
 export(plot_chromosomes)
 export(plot_positions)
 export(plot_rankings)
+export(plot_rankings_correlation)
 export(plot_scores)
 export(plot_scores_by_position)
 export(preset)
diff --git a/R/plots.R b/R/plots.R
index 49e63b5..ee70814 100644
--- a/R/plots.R
+++ b/R/plots.R
@@ -204,6 +204,112 @@ plot_rankings <- function(rankings, gene_sets) {
   plot
 }
 
+#' Plot a scatter plot to compare two rankings.
+#'
+#' This function requires the package `plotly`.
+#'
+#' @param ranking_x The ranking to be shown on the X axis.
+#' @param ranking_y The ranking to be shown on the Y axis.
+#' @param name_x Title of the X axis.
+#' @param name_y Title of the Y axis.
+#' @param gene_sets A named list of vectors of gene IDs to highlight. The names
+#'   will be used to distinguish the sets and in the legend.
+#' @param use_ranks Show ranks instead of scores.
+#'
+#' @export
+plot_rankings_correlation <- function(ranking_x,
+                                      ranking_y,
+                                      name_x,
+                                      name_y,
+                                      gene_sets = NULL,
+                                      use_ranks = TRUE) {
+  if (!requireNamespace("plotly", quietly = TRUE)) {
+    stop("Please install \"plotly\" to use this function.")
+  }
+
+  data <- merge(ranking_x, ranking_y, by = "gene")
+  data <- merge(data, geposan::genes, by.x = "gene", by.y = "id")
+
+  data[, `:=`(
+    x = if (use_ranks) rank.x else score.x,
+    y = if (use_ranks) rank.y else score.y
+  )]
+
+  model <- stats::lm(y ~ x, data)
+  model_data <- data.table(x = seq(min(data$x), max(data$x), length = 100))
+  model_data[, c("y", "lower", "upper") := data.table(
+    stats::predict(model, model_data, interval = "confidence")
+  )]
+
+  # Take a random sample to actually plot.
+  sample_data <- data[!gene %chin% unlist(gene_sets)][sample(.N, 1000)]
+
+  fig <- plotly::plot_ly() |>
+    plotly::add_markers(
+      data = sample_data,
+      x = ~x,
+      y = ~y,
+      name = "All genes",
+      marker = list(
+        color = base_color(),
+        size = 5
+      ),
+      hoverinfo = "skip"
+    ) |>
+    plotly::add_lines(
+      data = model_data,
+      x = ~x,
+      y = ~y,
+      line = list(color = base_color()),
+      showlegend = FALSE,
+      hoverinfo = "skip"
+    ) |>
+    plotly::add_ribbons(
+      data = model_data,
+      x = ~x,
+      ymin = ~lower,
+      ymax = ~upper,
+      fillcolor = base_color_transparent(),
+      line = list(width = 0),
+      showlegend = FALSE,
+      hoverinfo = "skip"
+    )
+
+  gene_set_index <- 1
+  for (gene_set_name in names(gene_sets)) {
+    gene_set <- gene_sets[[gene_set_name]]
+
+    fig <- fig |>
+      plotly::add_markers(
+        data = data[gene %chin% gene_set],
+        x = ~x,
+        y = ~y,
+        name = gene_set_name,
+        text = ~name,
+        marker = list(
+          color = gene_set_color(gene_set_index),
+          size = 8
+        )
+      )
+
+    gene_set_index <- gene_set_index + 1
+  }
+
+  fig <- fig |> plotly::layout(
+    xaxis = list(title = name_x),
+    yaxis = list(title = name_y)
+  )
+
+  if (use_ranks) {
+    fig <- fig |> plotly::layout(
+      xaxis = list(autorange = "reversed"),
+      yaxis = list(autorange = "reversed")
+    )
+  }
+
+  fig
+}
+
 
 #' Plot a ranking as a scatter plot of scores.
 #'
diff --git a/man/plot_rankings_correlation.Rd b/man/plot_rankings_correlation.Rd
new file mode 100644
index 0000000..aa7e690
--- /dev/null
+++ b/man/plot_rankings_correlation.Rd
@@ -0,0 +1,32 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/plots.R
+\name{plot_rankings_correlation}
+\alias{plot_rankings_correlation}
+\title{Plot a scatter plot to compare two rankings.}
+\usage{
+plot_rankings_correlation(
+  ranking_x,
+  ranking_y,
+  name_x,
+  name_y,
+  gene_sets = NULL,
+  use_ranks = TRUE
+)
+}
+\arguments{
+\item{ranking_x}{The ranking to be shown on the X axis.}
+
+\item{ranking_y}{The ranking to be shown on the Y axis.}
+
+\item{name_x}{Title of the X axis.}
+
+\item{name_y}{Title of the Y axis.}
+
+\item{gene_sets}{A named list of vectors of gene IDs to highlight. The names
+will be used to distinguish the sets and in the legend.}
+
+\item{use_ranks}{Show ranks instead of scores.}
+}
+\description{
+This function requires the package \code{plotly}.
+}