Figure Generation¶

The final stage of the pipeline turns the statistics tables written by the earlier steps into the figures used in the paper.

All panels are produced by a single notebook, pipeline/11_ploting/all_figures-raidhani.ipynb, which covers main figures 1, 2, 4 and 5 panel by panel, together with supplementary figures 1–7. Figure 3 is a schematic and is not generated here.

Each panel reads a statistics table produced upstream — graph composition and data-source counts from KG construction, loss curves and split summaries from training, verdict distributions from the hypothesis pipeline, and the yeast thermal-screen results — rather than recomputing anything from the knowledge graph itself. Keeping plotting separate from computation means a figure can be restyled without rerunning any part of the pipeline.

The notebook is committed with its outputs stored, so every panel can be inspected exactly as published without executing a cell.