Color Palettes

Palette Types

Type	Use Case	Example
Sequential	Continuous data (expression, coverage)	viridis, Blues
Diverging	Centered data (fold change, z-score)	RdBu, coolwarm
Qualitative	Categories (clusters, conditions)	Set1, tab10

viridis (Colorblind-Safe)

library(viridis)

# Continuous scale
ggplot(df, aes(x, y, color = value)) +
    geom_point() +
    scale_color_viridis_c()

# Discrete scale
ggplot(df, aes(x, y, color = group)) +
    geom_point() +
    scale_color_viridis_d()

# Options: viridis, magma, plasma, inferno, cividis, turbo
scale_color_viridis_c(option = 'magma')

import matplotlib.pyplot as plt

plt.scatter(x, y, c=values, cmap='viridis')
# Options: viridis, magma, plasma, inferno, cividis

RColorBrewer (R)

library(RColorBrewer)

# View all palettes
display.brewer.all()

# Sequential
scale_fill_brewer(palette = 'Blues')
scale_color_distiller(palette = 'YlOrRd', direction = 1)

# Diverging
scale_fill_brewer(palette = 'RdBu')
scale_color_gradient2(low = '#4DBBD5', mid = 'white', high = '#E64B35', midpoint = 0)

# Qualitative
scale_color_brewer(palette = 'Set1')
scale_fill_brewer(palette = 'Dark2')

# Get colors directly
brewer.pal(n = 5, name = 'Set1')

matplotlib/seaborn (Python)

import matplotlib.pyplot as plt
import seaborn as sns

# Sequential
plt.scatter(x, y, c=values, cmap='Blues')

# Diverging
plt.scatter(x, y, c=values, cmap='RdBu_r', vmin=-2, vmax=2)

# Qualitative
palette = sns.color_palette('Set1', n_colors=5)
sns.scatterplot(x=x, y=y, hue=group, palette='Set1')

# Custom palette
custom_palette = {'Control': '#4DBBD5', 'Treatment': '#E64B35'}
sns.scatterplot(x=x, y=y, hue=group, palette=custom_palette)

Scientific Journal Palettes

library(ggsci)

# Nature Publishing Group
scale_color_npg()
scale_fill_npg()

# AAAS Science
scale_color_aaas()

# Lancet
scale_color_lancet()

# JAMA
scale_color_jama()

# JCO
scale_color_jco()

Custom Palettes

# Define custom colors
my_colors <- c(
    'Control' = '#4DBBD5',
    'Treatment' = '#E64B35',
    'Vehicle' = '#00A087'
)

scale_color_manual(values = my_colors)
scale_fill_manual(values = my_colors)

# Create gradient
colorRampPalette(c('blue', 'white', 'red'))(100)

from matplotlib.colors import LinearSegmentedColormap

colors = ['#4DBBD5', 'white', '#E64B35']
cmap = LinearSegmentedColormap.from_list('custom_diverging', colors)
plt.imshow(data, cmap=cmap)

Heatmap Colors

library(circlize)

# For ComplexHeatmap
col_fun <- colorRamp2(c(-2, 0, 2), c('#4DBBD5', 'white', '#E64B35'))

# For pheatmap
pheatmap(mat, color = colorRampPalette(rev(brewer.pal(9, 'RdBu')))(100))

import seaborn as sns

sns.heatmap(data, cmap='RdBu_r', center=0, vmin=-2, vmax=2)

Colorblind Simulation

library(colorspace)

# Check if palette is colorblind safe
demoplot(rainbow(5), type = 'map')
demoplot(viridis(5), type = 'map')

# Simulate colorblindness
cvd_colors <- deutan(c('#E64B35', '#4DBBD5', '#00A087'))  # deuteranopia
cvd_colors <- protan(c('#E64B35', '#4DBBD5', '#00A087'))  # protanopia

Recommended Palettes

Data Type	Recommended	Avoid
Expression heatmap	RdBu (diverging)	Rainbow
Categories (<8)	Set1, Dark2, npg	Too many colors
Categories (>8)	tab20, Paired	Qualitative sets
Continuous	viridis, plasma	Jet, rainbow
p-values	viridis (reversed)	Red-green

Transparency

# Add alpha
scale_color_manual(values = alpha(c('#E64B35', '#4DBBD5'), 0.7))

# In geom
geom_point(alpha = 0.6)

# Add alpha to hex
def add_alpha(hex_color, alpha):
    return hex_color + format(int(alpha * 255), '02x')

color_with_alpha = add_alpha('#E64B35', 0.7)

# In scatter
plt.scatter(x, y, c='#E64B35', alpha=0.7)

Extract Colors from Palette

# Get discrete colors
pal <- brewer.pal(8, 'Set1')
pal[1:3]  # First 3 colors

# Interpolate more colors
colorRampPalette(brewer.pal(8, 'Set1'))(20)

import seaborn as sns

palette = sns.color_palette('Set1', n_colors=8)
palette[:3]  # First 3 colors

# As hex
palette.as_hex()

Related Skills

• data-visualization/ggplot2-fundamentals - Apply colors
• data-visualization/heatmaps-clustering - Heatmap colors
• data-visualization/specialized-omics-plots - Plot styling