UpSet Plot Examples

This example demonstrates UpSet plot functionality in PubliPlots. UpSet plots are an alternative to Venn diagrams for visualizing set intersections, especially useful when dealing with many sets (4 or more) where Venn diagrams become hard to read.

import publiplots as pp
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# Set style
pp.set_notebook_style()

Basic UpSet Plot

Create a basic UpSet plot from a dictionary of sets.

# Create sample sets with meaningful overlaps
np.random.seed(100)
upset_sets = {
    'Gene Set A': set(np.random.randint(1, 100, 50)),
    'Gene Set B': set(np.random.randint(30, 130, 55)),
    'Gene Set C': set(np.random.randint(60, 140, 45)),
    'Gene Set D': set(np.random.randint(20, 110, 48))
}

# Create basic UpSet plot
fig, axes = pp.upsetplot(
    data=upset_sets,
    title='Gene Set Intersection Analysis',
    show_counts=15,
)
plt.show()

UpSet Plot with Custom Colors

Customize the color and transparency of bars.

# Create UpSet plot with custom styling
fig, axes = pp.upsetplot(
    data=upset_sets,
    sort_by='size',  # Sort by intersection size
    title='Customized UpSet Plot',
    color='#E67E7E',
    alpha=0.3,
    bar_linewidth=1.5,
    show_counts=12,
)
plt.show()

UpSet Plot with Many Sets

UpSet plots excel at showing intersections between many sets, where Venn diagrams become impractical.

# Create 6 sets with various overlaps
np.random.seed(200)
many_sets = {
    'Pathway A': set(np.random.randint(1, 120, 60)),
    'Pathway B': set(np.random.randint(40, 160, 65)),
    'Pathway C': set(np.random.randint(80, 180, 55)),
    'Pathway D': set(np.random.randint(20, 140, 58)),
    'Pathway E': set(np.random.randint(60, 160, 62)),
    'Pathway F': set(np.random.randint(30, 150, 60))
}

# Create UpSet plot with many sets
fig, axes = pp.upsetplot(
    data=many_sets,
    sort_by='size',
    title='Pathway Overlap Analysis (6 Pathways)',
    color='#75B375',
    show_counts=20,
)
plt.show()

UpSet Plot from DataFrame

Create an UpSet plot from a binary membership DataFrame. This is useful when working with tabular data.

# Create binary membership DataFrame
# Rows are elements, columns are sets, values are 0/1 indicating membership
np.random.seed(300)
n_elements = 100
upset_df = pd.DataFrame({
    'Cluster_1': np.random.choice([0, 1], n_elements, p=[0.6, 0.4]),
    'Cluster_2': np.random.choice([0, 1], n_elements, p=[0.5, 0.5]),
    'Cluster_3': np.random.choice([0, 1], n_elements, p=[0.7, 0.3]),
    'Marker_Set': np.random.choice([0, 1], n_elements, p=[0.65, 0.35]),
})

# Create UpSet plot from DataFrame
fig, axes = pp.upsetplot(
    data=upset_df,
    sort_by='size',
    title='Cluster and Marker Overlaps',
    color='#E6B375',
    alpha=0.4,
    show_counts=15,
)
plt.show()

Comparing UpSet vs Venn for 4+ Sets

Demonstrate why UpSet plots are preferred for many sets.

# Create 5 sets
np.random.seed(400)
five_sets = {
    'Set 1': set(np.random.randint(1, 100, 50)),
    'Set 2': set(np.random.randint(30, 130, 55)),
    'Set 3': set(np.random.randint(60, 140, 45)),
    'Set 4': set(np.random.randint(20, 110, 48)),
    'Set 5': set(np.random.randint(40, 120, 52))
}

# Create side-by-side comparison
fig = plt.figure(figsize=(14, 6))

# Venn diagram (left)
ax1 = plt.subplot(1, 2, 1)
pp.venn(
    sets=list(five_sets.values()),
    labels=list(five_sets.keys()),
    colors=pp.color_palette('pastel', n_colors=5),
    ax=ax1
)
ax1.set_title('5-Way Venn Diagram', fontsize=12, fontweight='bold')

# UpSet plot (right) - need to handle this differently since upsetplot returns multiple axes
plt.subplot(1, 2, 2)
# For this comparison, we'll create a separate figure for the UpSet plot
plt.text(0.5, 0.5, 'UpSet plot shown separately\n(see next figure)',
         ha='center', va='center', fontsize=12)
plt.axis('off')
plt.title('UpSet Plot (Better for 5+ Sets)', fontsize=12, fontweight='bold')

plt.tight_layout()
plt.show()

# Show the UpSet plot
fig, axes = pp.upsetplot(
    data=five_sets,
    sort_by='size',
    title='Same Data as UpSet Plot (Clearer Visualization)',
    color='#5D83C3',
    alpha=0.3,
    show_counts=20,
)
plt.show()

• 
•