Render the Main Manuscript Figures

Command

python scripts/render_docs.py

render_docs.py installs a project local Jupyter kernel, removes stale rendered artifacts from previous runs, executes the Quarto notebooks in docs/, and writes the rendered site to docs/_site/.

Figure sources

The figure notebooks are grouped by input layer.

Truth table based

• Figure 1

Staged manuscript asset based

• Figure 2

Benchmark output based

• Figure 3
• Figure 4
• Figure 6
• Figure 7

Figure 6 also reads the no-bulge ML comparison recall-curve table for the ML panel. Figure 7 uses the no-bulge pairwise Reciprocal Rank Fusion output.

Prediction-contract based

• Figure 5

Figure 5 compares validated true sites with the per-tool prediction contracts, because coverage is defined before applying any top-k rank cutoff.

Notebook structure

Each figure notebook contains the figure specific data preparation and plotting code. Generic runtime helpers remain in src/offtarget_benchmark/, but the figure transformations themselves are executed in the notebook source.