bioviz-kit 0.3.9

Framework-agnostic plotting utilities for biological and clinical data visualization

bioviz-kit

Framework-agnostic visualization library for publication-ready clinical and biological data plots.

Features

• Publication-ready styling – Clean, professional visualizations out of the box
• Framework-agnostic – Works with any data pipeline or analysis framework
• Pydantic configurations – Type-safe, validated configuration objects
• Clinical & bioinformatics focused – Specialized plot types for common analyses:
  • Kaplan-Meier survival curves with risk tables
  • Volcano plots for differential expression/enrichment
  • Oncoplots (mutation landscapes)
  • Forest plots for hazard ratios
  • Waterfall plots for tumor response
  • Grouped bar charts with confidence intervals
  • Distribution plots (histogram + boxplot)
  • Styled tables

Installation

Install from PyPI:

pip install bioviz-kit

Or install in development mode:

git clone https://github.com/yourusername/bioviz-kit.git
cd bioviz-kit
pip install -e .

Requirements

• Python 3.11+
• pandas
• matplotlib
• pydantic
• lifelines (for KM plots)
• adjustText (for label placement)
• seaborn (optional, for some plot types)

Quick Start

Kaplan-Meier Survival Plots

import pandas as pd
from bioviz.configs import KMPlotConfig
from bioviz.plots import KMPlotter

# Your survival data
df = pd.DataFrame({
    "time": [5, 10, 15, 8, 12, 20, 6, 14],
    "event": [1, 0, 1, 1, 0, 1, 1, 0],
    "arm": ["Treatment", "Treatment", "Treatment", "Treatment",
            "Control", "Control", "Control", "Control"],
})

# Configure the plot
config = KMPlotConfig(
    time_col="time",
    event_col="event",
    group_col="arm",
    title="Overall Survival by Treatment Arm",
    xlabel="Time (months)",
    ylabel="Survival Probability",
    show_risktable=True,
    show_pvalue=True,
    color_dict={"Treatment": "#009E73", "Control": "#D55E00"},
)

# Generate the plot
plotter = KMPlotter(df, config)
fig, ax, pval = plotter.plot(output_path="km_plot.pdf")
print(f"Log-rank p-value: {pval:.4f}")

Volcano Plots

from bioviz.configs import VolcanoConfig
from bioviz.plots import VolcanoPlotter

# Differential expression results
df = pd.DataFrame({
    "gene": ["TP53", "KRAS", "EGFR", "BRCA1", "MYC"],
    "log2fc": [2.5, -1.8, 0.3, 3.1, -2.2],
    "pvalue": [0.001, 0.01, 0.5, 0.0001, 0.005],
})

config = VolcanoConfig(
    x_col="log2fc",
    y_col="pvalue",
    label_col="gene",
    title="Differential Expression Analysis",
    y_col_thresh=0.05,
    abs_x_thresh=1.5,
)

plotter = VolcanoPlotter(df, config)
fig, ax = plotter.plot()
fig.savefig("volcano.pdf", bbox_inches="tight")

Oncoplots

from bioviz.configs import OncoplotConfig
from bioviz.plots import OncoPlotter

# Mutation data (long format)
df = pd.DataFrame({
    "sample": ["S1", "S1", "S2", "S2", "S3"],
    "gene": ["TP53", "KRAS", "TP53", "EGFR", "KRAS"],
    "variant_class": ["Missense", "Missense", "Nonsense", "Amplification", "Missense"],
})

config = OncoplotConfig(
    sample_col="sample",
    gene_col="gene",
    variant_col="variant_class",
    title="Mutation Landscape",
)

plotter = OncoPlotter(df, config)
fig, axes = plotter.plot()
fig.savefig("oncoplot.pdf", bbox_inches="tight")

Forest Plots

from bioviz.configs import ForestPlotConfig
from bioviz.plots import ForestPlotter

# Hazard ratio data
df = pd.DataFrame({
    "comparator": ["Age >= 65", "Male", "Stage III-IV", "ECOG 1"],
    "reference": ["Age < 65", "Female", "Stage I-II", "ECOG 0"],
    "hr": [1.45, 0.92, 2.10, 1.68],
    "ci_lower": [1.10, 0.72, 1.65, 1.25],
    "ci_upper": [1.91, 1.18, 2.67, 2.26],
    "p_value": [0.008, 0.52, 0.001, 0.001],
})

config = ForestPlotConfig(
    title="Multivariate Cox Regression",
    xlabel="Hazard Ratio (95% CI)",
)

plotter = ForestPlotter(df, config)
fig, ax = plotter.plot()
fig.savefig("forest.pdf", bbox_inches="tight")

Grouped Bar Charts with CIs

from bioviz.configs import GroupedBarConfig
from bioviz.plots import GroupedBarPlotter

# Create from proportions (computes Clopper-Pearson CIs automatically)
config = GroupedBarConfig(
    title="Response Rates by Treatment",
    xlabel="Response Rate (%)",
    orientation="horizontal",
    ci_method="clopper-pearson",
)

plotter = GroupedBarPlotter.from_proportions(
    category_list=["CR", "PR", "SD", "PD"],
    group_configs=[
        {"name": "Treatment", "k": {"CR": 15, "PR": 25, "SD": 30, "PD": 10}, "n": 80},
        {"name": "Control", "k": {"CR": 5, "PR": 15, "SD": 35, "PD": 25}, "n": 80},
    ],
    config=config,
)
fig, ax = plotter.plot()

Architecture

bioviz-kit follows a consistent pattern across all plot types:

bioviz/
├── configs/           # Pydantic configuration classes
│   ├── km_cfg.py      # KMPlotConfig
│   ├── volcano_cfg.py # VolcanoConfig
│   └── ...
├── plots/             # Plotter classes
│   ├── km.py          # KMPlotter
│   ├── volcano.py     # VolcanoPlotter
│   └── ...
└── utils/             # Shared utilities

Each plot type has:

1. A Config class (e.g., KMPlotConfig) - Pydantic model with validated fields
2. A Plotter class (e.g., KMPlotter) - Takes data + config, produces matplotlib figure

Configuration Philosophy

All configurations use Pydantic models with:

• Type hints for IDE autocompletion
• Validation to catch errors early
• Defaults that produce publication-ready output
• Documentation via field descriptions

Font sizes default to None to inherit from matplotlib's rcParams, making it easy to apply global themes:

import matplotlib.pyplot as plt

# Set global style
plt.rcParams.update({
    "font.size": 12,
    "axes.labelsize": 14,
    "axes.titlesize": 16,
})

# Now all bioviz plots will use these sizes by default

Examples

See the examples/ directory for complete, runnable examples:

• minimal_bioviz_smoke.py - Line plots, oncoplots, tables
• oncoplot_example.py - Detailed oncoplot customization
• volcano_smoke.py - Volcano plot variations
• distribution_examples.py - Histogram + boxplot combinations
• km_survival_example.py - Kaplan-Meier survival analysis

Documentation

Full documentation is available at bioviz-kit.readthedocs.io.

Integration with tm-modeling

bioviz-kit serves as the visualization backend for tm-modeling. Users of tm-modeling can continue using the existing API (KMPlotConfig, generate_km_plot_and_risk_table) while benefiting from bioviz-kit's improved rendering.

License

bioviz-kit is released under the MIT License (c) 2025 Victoria Cheung.

Acknowledgments

This package was spun out of internal tooling developed at Revolution Medicines. Many thanks to the team there for allowing the code to be open sourced.