How do you preserve hue associations across multiple plots in seaborn?

With the following code as an example, how can I preserve the column-colour associations that you see in Plot 1 in the plot below it, Plot 2?

Is there a way to instruct sns to 'skip' a hue at specific points? I am looking for a simple workaround that I can control finely, and ideally without manually assigning hex codes.

To be clear, in Plot 2 I would like 'C' to appear green and 'E' to appear purple.

Any advice is appreciated, thanks.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

np.random.seed(42)

df_1 = pd.DataFrame(np.random.rand(5, 5), columns=['A', 'B', 'C', 'D', 'E'])

df_2 = df_1.drop(['B', 'D'], axis=1)

— Plot 1 —

sns.boxplot(data=df_1,
            orient="h")
plt.show()

— Plot 2 —

sns.boxplot(data=df_2,
            orient="h")
plt.show()

Pass a dictionary to boxplot's palette:

# all unique keys
keys = df_1.columns.union(df_2.columns).unique()
# or for more dataframes:
# set().union(*(df.columns for df in [df_1, df_2, df_3, df_4]))
# ['A', 'B', 'C', 'D', 'E']

# create key: color dictionary
colors = dict(zip(keys, sns.color_palette('Set1')))

# plot 1
sns.boxplot(data=df_1, orient='h', palette=colors)
# plot 2
sns.boxplot(data=df_2, orient='h', palette=colors)

NB. you can replace sns.color_palette('Set1') by a list of colors (e.g. ['#e41a1c', '#377eb8', '#4daf4a', '#984ea3', '#ff7f00']).

If you use seaborn's palettes, keep in mind that some are limited in the number of colors. To cycle over them use:

import itertools

colors = dict(zip(keys, itertools.cycle(sns.color_palette('Set1'))))

Output:

sns.palplot(colors.values())