diff --git a/README.md b/README.md
index c606c640..322c7fb3 100644
--- a/README.md
+++ b/README.md
@@ -37,6 +37,14 @@ Use DrEval to build drug response models that have an impact
---
+
+
+
+ 
+
+
+---
+
This project is a collaboration of the Technical University of Munich (TUM, Germany)
and the Freie Universität Berlin (FU, Germany).
diff --git a/README.rst b/README.rst
index 20289ac9..9604fcd8 100644
--- a/README.rst
+++ b/README.rst
@@ -68,3 +68,10 @@ Use DrEval to build drug response models that have an impact
This project is a collaboration of the Technical University of Munich (TUM, Germany)
and the Freie Universität Berlin (FU, Germany).
+
+Leaderboard
+-----------
+
+.. image:: docs/_static/img/leaderboard.png
+ :alt: DrEvalPy Leaderboard
+ :align: center
diff --git a/docs/_static/img/DrugResponseEvalLogo.svg b/docs/_static/img/DrugResponseEvalLogo.svg
new file mode 100644
index 00000000..a49fd3b5
--- /dev/null
+++ b/docs/_static/img/DrugResponseEvalLogo.svg
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/docs/_static/img/leaderboard_dark.png b/docs/_static/img/leaderboard_dark.png
new file mode 100644
index 00000000..42716c63
Binary files /dev/null and b/docs/_static/img/leaderboard_dark.png differ
diff --git a/docs/_static/img/leaderboard_light.png b/docs/_static/img/leaderboard_light.png
new file mode 100644
index 00000000..3330824c
Binary files /dev/null and b/docs/_static/img/leaderboard_light.png differ
diff --git a/drevalpy/visualization/create_leaderboard.py b/drevalpy/visualization/create_leaderboard.py
new file mode 100644
index 00000000..4a0de6da
--- /dev/null
+++ b/drevalpy/visualization/create_leaderboard.py
@@ -0,0 +1,478 @@
+#!/usr/bin/env python3
+"""
+DrEvalPy Leaderboard Visualization.
+
+This script generates a leaderboard visualization (normalized PCC and RMSE) from
+the evaluation results CSV file produced by the DrEvalPy evaluation pipeline.
+Usage:
+python create_leaderboard.py --results_path /path/to/results.csv
+"""
+
+import argparse
+from pathlib import Path
+from typing import Optional
+
+import matplotlib.patches as mpatches
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+from matplotlib.patches import FancyBboxPatch
+
+# --- Theme Definitions ---
+DARK_THEME = {
+ "background": "#0d1117",
+ "surface": "#2d2d2d",
+ "text": "#ece7e4",
+ "text_secondary": "#a0a0a0",
+ "grid": "#30363d",
+}
+
+LIGHT_THEME = {
+ "background": "#ffffff",
+ "surface": "#f6f8fa",
+ "text": "#1f2328",
+ "text_secondary": "#57606a",
+ "grid": "#d0d7de",
+}
+
+COLORS = DARK_THEME
+
+COMPETITOR_COLOR = "#6A5ACD"
+
+
+def configure_matplotlib(font_adder: int = 0):
+ """
+ Configure global matplotlib parameters for the current theme.
+
+ :param font_adder: Increment to add to the base font size.
+ """
+ plt.rcParams.update(
+ {
+ "figure.facecolor": COLORS["background"],
+ "axes.facecolor": COLORS["background"],
+ "axes.edgecolor": COLORS["grid"],
+ "axes.labelcolor": COLORS["text"],
+ "text.color": COLORS["text"],
+ "xtick.color": COLORS["text"],
+ "ytick.color": COLORS["text"],
+ "grid.color": COLORS["grid"],
+ "font.family": "sans-serif",
+ "font.size": 11 + font_adder,
+ "axes.spines.top": False,
+ "axes.spines.right": False,
+ }
+ )
+
+
+def load_results(results_path: str, test_mode: str = "LCO") -> pd.DataFrame:
+ """
+ Load and aggregate results from the evaluation CSV.
+
+ :param results_path: Path to evaluation_results.csv.
+ :param test_mode: Filtering mode (e.g., LCO).
+ :raises FileNotFoundError: If path does not exist.
+ :raises ValueError: If no data matches criteria.
+ :return: Processed DataFrame.
+ """
+ path = Path(results_path)
+ if not path.exists():
+ raise FileNotFoundError(f"Results file not found: {results_path}")
+
+ df = pd.read_csv(path, index_col=0)
+ df = df[(df["rand_setting"] == "predictions") & (df["test_mode"] == test_mode)]
+
+ if df.empty:
+ raise ValueError(f"No results found for rand_setting='predictions' and test_mode='{test_mode}'")
+
+ df_agg = (
+ df.groupby("algorithm")
+ .agg(
+ {
+ "Pearson: normalized": ["mean", "std"],
+ "RMSE": ["mean", "std"],
+ }
+ )
+ .reset_index()
+ )
+
+ df_agg.columns = ["algorithm", "PCC", "PCC_std", "RMSE", "RMSE_std"]
+ df_agg["PCC_std"] = df_agg["PCC_std"].fillna(0)
+ df_agg["RMSE_std"] = df_agg["RMSE_std"].fillna(0)
+ df_agg["is_baseline"] = df_agg["algorithm"].str.startswith("Naive")
+
+ return df_agg.sort_values("PCC", ascending=False).reset_index(drop=True)
+
+
+def get_bar_color(rank: int, is_baseline: bool) -> dict:
+ """
+ Assign colors based on model rank and type.
+
+ :param rank: Model index in sorted list.
+ :param is_baseline: Boolean if model is a baseline.
+ :return: Styling dictionary.
+ """
+ if is_baseline:
+ return {"color": "#5a5a5a", "alpha": 1.0}
+
+ medal_gold = "#F4D03F"
+ medal_silver = "#BDC3C7"
+ medal_bronze = "#E67E22"
+
+ if rank == 0:
+ return {"color": medal_gold, "alpha": 1.0}
+ elif rank == 1:
+ return {"color": medal_silver, "alpha": 1.0}
+ elif rank == 2:
+ return {"color": medal_bronze, "alpha": 1.0}
+
+ return {"color": COMPETITOR_COLOR, "alpha": 0.85}
+
+
+def draw_bar(ax, x: float, y: float, width: float, height: float, color: str, alpha: float = 1.0):
+ """
+ Draw a custom rounded rectangle bar.
+
+ :param ax: Matplotlib axis.
+ :param x: Origin X.
+ :param y: Origin Y.
+ :param width: Bar width.
+ :param height: Bar height.
+ :param color: Hex color.
+ :param alpha: Transparency.
+ :return: Patch artist.
+ """
+ bar = FancyBboxPatch(
+ (x, y - height / 2),
+ width,
+ height,
+ boxstyle="round,pad=0.01,rounding_size=0.015",
+ facecolor=color,
+ alpha=alpha,
+ edgecolor="none",
+ zorder=3,
+ )
+ ax.add_patch(bar)
+ return bar
+
+
+def create_leaderboard(
+ df: pd.DataFrame,
+ output_path: str,
+ test_mode: str = "LCO",
+ dataset: str = "CTRPv2",
+ measure: str = "LN_IC50_curvecurator",
+ figsize: tuple = (16, 12),
+ show_top_n: Optional[int] = None,
+ font_adder: int = 6,
+) -> tuple:
+ """
+ Generate the dual-panel leaderboard figure.
+
+ :param df: Input results data.
+ :param output_path: File path for save.
+ :param test_mode: Evaluation mode name.
+ :param dataset: Dataset name.
+ :param measure: Performance measure.
+ :param figsize: Figure dimensions.
+ :param show_top_n: Limit displayed models.
+ :param font_adder: Scale for text.
+ :return: Figure and axes tuple.
+ """
+ configure_matplotlib(font_adder=font_adder)
+
+ if show_top_n:
+ df = df.head(show_top_n)
+
+ n_models = len(df)
+ y_positions = np.arange(n_models - 1, -1, -1)
+ bar_height = 0.65
+
+ fig, (ax1, ax2) = plt.subplots(1, 2, figsize=figsize, facecolor=COLORS["background"])
+ fig.subplots_adjust(wspace=0.4)
+
+ ax1.set_facecolor(COLORS["background"])
+ df_pcc = df.sort_values("PCC", ascending=False).reset_index(drop=True)
+ max_pcc = (df_pcc["PCC"] + df_pcc["PCC_std"]).max() * 1.18
+
+ for i, (_, row) in enumerate(df_pcc.iterrows()):
+ style = get_bar_color(i, row["is_baseline"])
+ draw_bar(ax1, 0, y_positions[i], row["PCC"], bar_height, style["color"], style["alpha"])
+
+ label_color = style["color"] if not row["is_baseline"] else COLORS["text_secondary"]
+ label_x = row["PCC"] + max_pcc * 0.02
+ ax1.text(
+ label_x,
+ y_positions[i],
+ f"{row['PCC']:.3f}",
+ va="center",
+ ha="left",
+ fontsize=9 + font_adder,
+ fontweight="bold",
+ color=label_color,
+ zorder=5,
+ )
+
+ if i < 3 and not row["is_baseline"]:
+ medals = ["①", "②", "③"]
+ ax1.text(
+ -max_pcc * 0.03,
+ y_positions[i],
+ medals[i],
+ va="center",
+ ha="center",
+ fontsize=14 + font_adder,
+ fontweight="bold",
+ color=style["color"],
+ zorder=5,
+ )
+
+ ax1.set_xlim(-max_pcc * 0.06, max_pcc)
+ ax1.set_ylim(-0.8, n_models - 0.2)
+ ax1.set_yticks(y_positions)
+ ax1.set_yticklabels(df_pcc["algorithm"].values, fontsize=10 + font_adder)
+
+ for i, label in enumerate(ax1.get_yticklabels()):
+ if i < 3 and not df_pcc.iloc[i]["is_baseline"]:
+ label.set_fontweight("bold")
+ label.set_color(get_bar_color(i, False)["color"])
+ elif df_pcc.iloc[i]["is_baseline"]:
+ label.set_style("italic")
+ label.set_color(COLORS["text_secondary"])
+ else:
+ label.set_color(COLORS["text"])
+
+ ax1.set_xlabel("Normalized PCC", fontsize=12 + font_adder, fontweight="bold", labelpad=10)
+ ax1.xaxis.grid(True, linestyle="--", alpha=0.3, color=COLORS["grid"])
+ ax1.set_axisbelow(True)
+ ax1.tick_params(axis="x", colors=COLORS["text_secondary"])
+ ax1.set_title(
+ "Normalized Pearson ↑ higher is better",
+ fontsize=14 + font_adder,
+ fontweight="bold",
+ color="#29ABCA",
+ pad=15,
+ )
+
+ ax2.set_facecolor(COLORS["background"])
+ df_rmse = df.sort_values("RMSE", ascending=True).reset_index(drop=True)
+ max_rmse = (df_rmse["RMSE"] + df_rmse["RMSE_std"]).max() * 1.18
+
+ for i, (_, row) in enumerate(df_rmse.iterrows()):
+ style = get_bar_color(i, row["is_baseline"])
+ draw_bar(ax2, 0, y_positions[i], row["RMSE"], bar_height, style["color"], style["alpha"])
+
+ label_color = style["color"] if not row["is_baseline"] else COLORS["text_secondary"]
+ label_x = row["RMSE"] + max_rmse * 0.02
+ ax2.text(
+ label_x,
+ y_positions[i],
+ f"{row['RMSE']:.3f}",
+ va="center",
+ ha="left",
+ fontsize=9 + font_adder,
+ fontweight="bold",
+ color=label_color,
+ zorder=5,
+ )
+
+ if i < 3 and not row["is_baseline"]:
+ medals = ["①", "②", "③"]
+ ax2.text(
+ -max_rmse * 0.03,
+ y_positions[i],
+ medals[i],
+ va="center",
+ ha="center",
+ fontsize=14 + font_adder,
+ fontweight="bold",
+ color=style["color"],
+ zorder=5,
+ )
+
+ ax2.set_xlim(-max_rmse * 0.06, max_rmse)
+ ax2.set_ylim(-0.8, n_models - 0.2)
+ ax2.set_yticks(y_positions)
+ ax2.set_yticklabels(df_rmse["algorithm"].values, fontsize=10 + font_adder)
+ ax2.set_xlabel("Root Mean Square Error", fontsize=12 + font_adder, fontweight="bold", labelpad=10)
+
+ for i, label in enumerate(ax2.get_yticklabels()):
+ if i < 3 and not df_rmse.iloc[i]["is_baseline"]:
+ label.set_fontweight("bold")
+ label.set_color(get_bar_color(i, False)["color"])
+ elif df_rmse.iloc[i]["is_baseline"]:
+ label.set_style("italic")
+ label.set_color(COLORS["text_secondary"])
+ else:
+ label.set_color(COLORS["text"])
+
+ ax2.xaxis.grid(True, linestyle="--", alpha=0.3, color=COLORS["grid"])
+ ax2.set_axisbelow(True)
+ ax2.tick_params(axis="x", colors=COLORS["text_secondary"])
+ ax2.set_title("RMSE ↓ lower is better", fontsize=14 + font_adder, fontweight="bold", color="#FF6B9D", pad=15)
+
+ title_text = "DrEval Challenge Leaderboard"
+ n_chars = len(title_text)
+ gradient_colors = []
+ for j in range(n_chars):
+ t = j / max(n_chars - 1, 1)
+ if t < 0.5:
+ t2 = t * 2
+ r = int(0x14 + (0x29 - 0x14) * t2)
+ g = int(0xB8 + (0xAB - 0xB8) * t2)
+ b = int(0xA6 + (0xCA - 0xA6) * t2)
+ else:
+ t2 = (t - 0.5) * 2
+ r = int(0x29 + (0x9D - 0x29) * t2)
+ g = int(0xAB + (0x4E - 0xAB) * t2)
+ b = int(0xCA + (0xDD - 0xCA) * t2)
+ gradient_colors.append(f"#{r:02x}{g:02x}{b:02x}")
+
+ title_x_start = 0.5 - len(title_text) * 0.012
+ for j, char in enumerate(title_text):
+ fig.text(
+ title_x_start + j * 0.024,
+ 0.97,
+ char,
+ fontsize=24 + font_adder,
+ fontweight="bold",
+ color=gradient_colors[j],
+ ha="center",
+ )
+ fig.text(
+ 0.5,
+ 0.92,
+ f"{dataset} Dataset • {measure} • {_get_test_mode_name(test_mode)}",
+ ha="center",
+ fontsize=12 + font_adder,
+ color=COLORS["text_secondary"],
+ )
+
+ logo_path = Path("docs/_static/img/DrugResponseEvalLogo.svg")
+ if logo_path.exists():
+ try:
+ from io import BytesIO
+
+ import cairosvg
+ from PIL import Image
+
+ png_data = cairosvg.svg2png(url=str(logo_path))
+ logo_img = Image.open(BytesIO(png_data))
+ logo_ax = fig.add_axes((0.8, 0.94, 0.15, 0.06))
+ logo_ax.imshow(logo_img)
+ logo_ax.axis("off")
+ except Exception as e:
+ print(e)
+ pass
+
+ legend_elements = [
+ mpatches.Patch(facecolor="#F4D03F", label="#1 Champion", edgecolor="none"),
+ mpatches.Patch(facecolor="#BDC3C7", label="#2 Runner-up", edgecolor="none"),
+ mpatches.Patch(facecolor="#E67E22", label="#3 Third Place", edgecolor="none"),
+ mpatches.Patch(facecolor=COMPETITOR_COLOR, alpha=0.85, label="Competitor", edgecolor="none"),
+ mpatches.Patch(facecolor="#5a5a5a", alpha=1, label="Baseline", edgecolor="none"),
+ ]
+
+ legend = fig.legend(
+ handles=legend_elements,
+ loc="lower center",
+ ncol=5,
+ frameon=True,
+ facecolor=COLORS["surface"],
+ edgecolor=COLORS["grid"],
+ fontsize=10 + font_adder,
+ bbox_to_anchor=(0.5, 0.02),
+ )
+ legend.get_frame().set_alpha(0.9)
+ for text in legend.get_texts():
+ text.set_color(COLORS["text"])
+
+ footer_text = (
+ "Submit your model → https://drevalpy.readthedocs.io/en/latest/. "
+ "Send us your results.\n\n"
+ "If you significantly outperform the RandomForest, we send you chocolate!"
+ )
+
+ fig.text(
+ 0.5,
+ -0.01,
+ footer_text,
+ ha="center",
+ va="top",
+ fontsize=14 + font_adder,
+ color=COLORS["text_secondary"],
+ style="italic",
+ linespacing=1.0,
+ )
+
+ plt.tight_layout(rect=(0, 0.06, 1, 0.90))
+ fig.savefig(output_path, dpi=150, bbox_inches="tight", facecolor=COLORS["background"], transparent=False)
+ plt.close(fig)
+ print(f"Saved leaderboard to: {output_path}")
+
+ return fig, (ax1, ax2)
+
+
+def _get_test_mode_name(test_mode: str) -> str:
+ """
+ Map shorthand mode codes to full descriptive names.
+
+ :param test_mode: Suffix code (LCO, etc).
+ :return: Full string name.
+ """
+ names = {
+ "LCO": "10-Fold Leave-Cell-Out Cross Validation",
+ "LDO": "10-Fold Leave-Drug-Out Cross Validation",
+ "LPO": "10-Fold Leave-Pair-Out Cross Validation",
+ "LTO": "10-Fold Leave-Tissue-Out Cross Validation",
+ }
+ return names.get(test_mode, test_mode)
+
+
+def main():
+ """Execute dual-theme leaderboard generation."""
+ parser = argparse.ArgumentParser(
+ description="Generate DrEvalPy leaderboard visualization (Dark & Light modes)",
+ formatter_class=argparse.RawDescriptionHelpFormatter,
+ )
+ parser.add_argument("--results_path", "-r", type=str, required=True, help="Path to evaluation_results.csv")
+ parser.add_argument("--output_dir", "-o", type=str, default="docs/_static/img", help="Directory to save images")
+ parser.add_argument("--test_mode", "-t", type=str, default="LCO", choices=["LCO", "LDO", "LPO", "LTO"])
+ parser.add_argument("--dataset", "-d", type=str, default="CTRPv2", help="Dataset name")
+ parser.add_argument("--measure", "-m", type=str, default="LN_IC50_curvecurator", help="Response measure")
+ parser.add_argument("--top_n", "-n", type=int, default=None, help="Top N models")
+ parser.add_argument("--font_adder", type=int, default=6, help="Font size increment")
+
+ args = parser.parse_args()
+
+ df = load_results(args.results_path, test_mode=args.test_mode)
+
+ out_dir = Path(args.output_dir)
+ out_dir.mkdir(parents=True, exist_ok=True)
+
+ global COLORS
+
+ COLORS = DARK_THEME
+ create_leaderboard(
+ df=df,
+ output_path=str(out_dir / "leaderboard_dark.png"),
+ test_mode=args.test_mode,
+ dataset=args.dataset,
+ measure=args.measure,
+ show_top_n=args.top_n,
+ font_adder=args.font_adder,
+ )
+
+ COLORS = LIGHT_THEME
+ create_leaderboard(
+ df=df,
+ output_path=str(out_dir / "leaderboard_light.png"),
+ test_mode=args.test_mode,
+ dataset=args.dataset,
+ measure=args.measure,
+ show_top_n=args.top_n,
+ font_adder=args.font_adder,
+ )
+
+
+if __name__ == "__main__":
+ main()