Scientific Visualization Tools

Use this skill when the user needs a figure artifact rather than only a numeric summary.

Typical triggers:

• publication-ready scatter, box, violin, bar, or heatmap figures
• interactive HTML charts for exploratory research data review
• small reusable plotting scripts for assay, omics, or cohort tables
• consistent styling across scientific plots

Environment Check

which python3 || true
python3 - <<'PY'
mods = ["pandas", "matplotlib", "seaborn", "plotly"]
for name in mods:
    try:
        __import__(name)
        print(f"{name}: ok")
    except Exception as exc:
        print(f"{name}: missing ({exc})")
PY

If key plotting modules are missing, recommend the optional drug-sandbox image documented in docs/operations/science-runtime.md.

Bundled Assets

• templates/publication_plot.py
• templates/interactive_plot.py

Preferred Workflow

1. Decide first whether the output should be static publication art or interactive exploration.
2. Keep the plotting script parameterized by column names rather than hardcoding one dataset.
3. Save the figure and a small JSON summary of what was plotted.
4. Do not use interactive charts where a paper-ready static figure is required.
5. Do not claim statistical meaning from a plot unless the underlying analysis is also reported.

Static Publication Plots

python3 templates/publication_plot.py \
  --input figures/assay.csv \
  --kind box \
  --x-column arm \
  --y-column response \
  --color-column arm \
  --output figures/assay_box.png \
  --summary figures/assay_box.json

Supported baseline kinds:

• scatter
• line
• box
• violin
• bar
• heatmap

Interactive Plotly Charts

python3 templates/interactive_plot.py \
  --input figures/cohort.csv \
  --kind scatter \
  --x-column age \
  --y-column biomarker \
  --color-column response \
  --output figures/cohort_scatter.html \
  --summary figures/cohort_scatter.json

Use this for exploratory review, dashboards, and lightweight sharing.

Related Skills

For statistical inference behind a plot, activate stat-modeling-tools. For Kaplan-Meier and time-to-event figures, activate survival-analysis-tools. For broader scientific-writing and manuscript-structure work, activate scientific-workflow-tools.