DeepHealth/plot_model_comparison_1year.R

# Plot AUC comparisons (1-year gap) between models and Delphi using ggplot2
# Usage:
#   Rscript plot_model_comparison_1year.R [path_to_csv] [output_dir]
# Defaults:
#   path_to_csv = "model_comparison_auc_1year.csv"
#   output_dir  = current working directory (".")

suppressPackageStartupMessages({
  library(ggplot2)
  library(cowplot)
})

args <- commandArgs(trailingOnly = TRUE)
csv_path <- if (length(args) >= 1) args[1] else "model_comparison_auc_1year.csv"
out_dir  <- if (length(args) >= 2) args[2] else "."

if (!dir.exists(out_dir)) {
  dir.create(out_dir, recursive = TRUE, showWarnings = FALSE)
}

# Read data
# Expect columns including: auc_delphi, auc_256, auc_120, Colour (hex color), name, etc.
df <- tryCatch({
  read.csv(csv_path, check.names = FALSE)
}, error = function(e) {
  stop(sprintf("Failed to read CSV at '%s': %s", csv_path, e$message))
})


# Helper to compare any two AUC columns (x vs y)
make_xy_plot <- function(data, x_col, y_col, title_text, x_label, y_label) {
  ggplot(data, aes(x = .data[[x_col]], y = .data[[y_col]])) +
    geom_abline(slope = 1, intercept = 0, color = "black", linetype = "dashed", linewidth = 0.5) +
    geom_vline(xintercept = 0.5, color = "gray50", linetype = "dashed", linewidth = 0.4) +
    geom_hline(yintercept = 0.5, color = "gray50", linetype = "dashed", linewidth = 0.4) +
    geom_point(aes(fill = Colour), shape = 21, color = "white", stroke = 0.65, size = 2.2, alpha = 0.95, show.legend = FALSE) +
    scale_fill_identity() +
    coord_cartesian(xlim = c(0.3, 1.05), ylim = c(0.3, 1.05)) +
    coord_fixed(ratio = 1) +
    labs(title = title_text, x = x_label, y = y_label) +
    theme_minimal(base_size = 10) +
    theme(
      plot.title = element_text(hjust = 0.5),
      panel.grid.minor = element_blank()
    )
}


# Helper to compare model AUC vs Delphi AUC (x = auc_delphi)
make_delphi_plot <- function(data, y_col, title_text, y_label) {
  ggplot(data, aes(x = auc_delphi, y = .data[[y_col]])) +
    geom_abline(slope = 1, intercept = 0, color = "black", linetype = "dashed", linewidth = 0.5) +
    geom_vline(xintercept = 0.5, color = "gray50", linetype = "dashed", linewidth = 0.4) +
    geom_hline(yintercept = 0.5, color = "gray50", linetype = "dashed", linewidth = 0.4) +
    geom_point(aes(fill = Colour), shape = 21, color = "white", stroke = 0.65, size = 2.2, alpha = 0.95, show.legend = FALSE) +
    scale_fill_identity() +
    coord_cartesian(xlim = c(0.3, 1.05), ylim = c(0.3, 1.05)) +
    coord_fixed(ratio = 1) +
    labs(title = title_text, x = "AUC_Delphi", y = y_label) +
    theme_minimal(base_size = 10) +
    theme(
      plot.title = element_text(hjust = 0.5),
      panel.grid.minor = element_blank()
    )
}

# Placeholder empty plot if a required column is missing
empty_plot <- function(msg) {
  ggplot() + theme_void() + ggtitle(msg) + theme(plot.title = element_text(hjust = 0.5))
}


# Plot: AUC_120 vs AUC_120_L (1 year gap)
if (!all(c("auc_120", "auc_120_l") %in% names(df))) {
  warning("Columns 'auc_120' and/or 'auc_120_l' not found in CSV; skipping AUC_120 vs AUC_120_L plot.")
} else {
  p120_vs_120l <- make_xy_plot(
    data = df,
    x_col = "auc_120",
    y_col = "auc_120_l",
    title_text = "AUC_120 vs AUC_120_L 1 year gap",
    x_label = "AUC_120",
    y_label = "AUC_120_L"
  )
  out_120_vs_120l <- file.path(out_dir, "model_comparison_auc_120_vs_120_l_1year.png")
  ggsave(filename = out_120_vs_120l, plot = p120_vs_120l, width = 7, height = 4, dpi = 600, bg = "white")
  cat(sprintf("Saved: %s\n", out_120_vs_120l))
}

# Plot: AUC_256 vs AUC_256_L (1 year gap)
if (!all(c("auc_256", "auc_256_l") %in% names(df))) {
  warning("Columns 'auc_256' and/or 'auc_256_l' not found in CSV; skipping AUC_256 vs AUC_256_L plot.")
} else {
  p256_vs_256l <- make_xy_plot(
    data = df,
    x_col = "auc_256",
    y_col = "auc_256_l",
    title_text = "AUC_256 vs AUC_256_L 1 year gap",
    x_label = "AUC_256",
    y_label = "AUC_256_L"
  )
  out_256_vs_256l <- file.path(out_dir, "model_comparison_auc_256_vs_256_l_1year.png")
  ggsave(filename = out_256_vs_256l, plot = p256_vs_256l, width = 7, height = 4, dpi = 600, bg = "white")
  cat(sprintf("Saved: %s\n", out_256_vs_256l))
}

# ---- Combined 2x2 grid: (auc_120 vs delphi), (auc_256 vs delphi), (auc_120_l vs delphi), (auc_256_l vs delphi) ----

has_cols <- function(cols) all(cols %in% names(df))

p_120_vs_delphi   <- if (has_cols(c("auc_delphi", "auc_120")))   make_delphi_plot(df, "auc_120",   "AUC_120 vs Delphi (1 year)",   "AUC_120")   else empty_plot("Missing auc_120 or auc_delphi")
p_256_vs_delphi   <- if (has_cols(c("auc_delphi", "auc_256")))   make_delphi_plot(df, "auc_256",   "AUC_256 vs Delphi (1 year)",   "AUC_256")   else empty_plot("Missing auc_256 or auc_delphi")
p_120l_vs_delphi  <- if (has_cols(c("auc_delphi", "auc_120_l"))) make_delphi_plot(df, "auc_120_l", "AUC_120_L vs Delphi (1 year)", "AUC_120_L") else empty_plot("Missing auc_120_l or auc_delphi")
p_256l_vs_delphi  <- if (has_cols(c("auc_delphi", "auc_256_l"))) make_delphi_plot(df, "auc_256_l", "AUC_256_L vs Delphi (1 year)", "AUC_256_L") else empty_plot("Missing auc_256_l or auc_delphi")

grid_plot <- plot_grid(
  p_120_vs_delphi, p_256_vs_delphi,
  p_120l_vs_delphi, p_256l_vs_delphi,
  labels = c("A", "B", "C", "D"),
  ncol = 2, align = "hv"
)

out_grid <- file.path(out_dir, "model_comparison_auc_vs_delphi_1year_grid.png")
ggsave(filename = out_grid, plot = grid_plot, width = 12, height = 8, dpi = 300, bg = "white")
cat(sprintf("Saved grid: %s\n", out_grid))