import matplotlib as mpl
import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np
import pandas as pd
from statannotations.Annotator import Annotator
tips = pd.read_csv('../Data/tips.csv')
自定义注释¶
Annotator类直接接受seaborn对应绘图函数的参数，最重要的是数据、分组、亚分组及分组顺序相关的参数，即data、x、y、order、hue、hue_order以及子图ax参数。
fig = plt.figure(figsize=(8, 8), dpi=100, layout="constrained")
ax = fig.add_subplot(1, 1, 1)
plot_params = {'data': tips,
'x': 'day',
'y': 'total_bill',
'order': ['Thur', 'Fri', 'Sat', 'Sun'],
'hue': 'day',
'palette': ['#ffa200', '#c8ff00', '#00ffe5', '#ee00ff'],
'width': 0.5,
'linecolor': 'k',
'linewidth': 2,
'ax': ax,
}
pairs = [('Thur', 'Fri'),
('Thur', 'Sat'),
('Thur', 'Sun'),
('Fri', 'Sat'),
('Fri', 'Sun'),
('Sat', 'Sun')]
custom_annotations = [f"pvalue={p:.2e}" for p in [6.477e-01, 1.276e-01, 3.585e-01, 4.690e-02, 2.680e-02, 5.363e-03]]
sns.boxplot(**plot_params)
annotator = Annotator(pairs=pairs, # 比对的组
engine='seaborn', # 绘图引擎，目前仅支持seaborn
plot='boxplot', # 绘图类型
verbose=True, # 是否打印日志
**plot_params # seaborn函数绘图参数
)
annotator.configure(color='r', # 线条颜色
line_height=0.02, # 线条高度
line_width=1, # 线条宽度
loc='inside', # 'inside'绘制在Axes内; 'outside'绘制在Axes外
text_offset=1, # 文本垂直偏移量
fontsize=15, # 字体大小
verbose=True
)
annotator.set_custom_annotations(custom_annotations)
annotator.annotate()
# annotator.annotate_custom_annotations(custom_annotations)
"""
p-value annotation legend:
ns: 5.00e-02 < p <= 1.00e+00
*: 1.00e-02 < p <= 5.00e-02
**: 1.00e-03 < p <= 1.00e-02
***: 1.00e-04 < p <= 1.00e-03
****: p <= 1.00e-04
Thur vs. Fri: pvalue=6.48e-01
Fri vs. Sat: pvalue=4.69e-02
Sat vs. Sun: pvalue=5.36e-03
Thur vs. Sat: pvalue=1.28e-01
Fri vs. Sun: pvalue=2.68e-02
Thur vs. Sun: pvalue=3.58e-01
(<Axes: xlabel='day', ylabel='total_bill'>,
[<statannotations.Annotation.Annotation at 0x1afb5a57980>,
<statannotations.Annotation.Annotation at 0x1afbff4b530>,
<statannotations.Annotation.Annotation at 0x1afc0867830>,
<statannotations.Annotation.Annotation at 0x1afc0831130>,
<statannotations.Annotation.Annotation at 0x1afc0951910>,
<statannotations.Annotation.Annotation at 0x1afc11f8950>])
"""
plt.show()
自动格式化p值¶
p值的格式化有三种形式：'full', 'simple', 'star'。通过text_format参数设置。
-
'full'：注释格式为**'检验方法名 p = 输入p值的格式化形式'**。输入p值的格式化形式通过pvalue_format_string参数设置，默认为'{:.3e}'。 -
'simple'：对于小于或等于某个值（阈值）的p值注释格式为**'p ≤ 设定格式'，而对于大于所设置阈值的最大值的p值注释格式为'p = 输入p值的格式化形式'**。阈值及对应格式通过pvalue_thresholds参数设置，默认为[[1e-5, "1e-5"], [1e-4, "1e-4"], [1e-3, "0.001"], [1e-2, "0.01"], [5e-2, "0.05"]]，一般使用默认值即可。输入p值的格式化形式通过pvalue_format_string参数设置，默认为'{:.2f}'。 -
'star'（默认）：对于小于或等于某个值（阈值）的p值注释格式为**'设定格式'**，而对于大于所设置阈值的最大值的p值则不注释。阈值及对应格式通过pvalue_thresholds参数设置，默认为[[1e-4, "****"], [1e-3, "***"], [1e-2, "**"], [0.05, "*"], [1, "ns"]]，一般使用默认值即可。
fig = plt.figure(figsize=(8, 8), dpi=100, layout="constrained")
ax = fig.add_subplot(1, 1, 1)
plot_params = {'data': tips,
'x': 'day',
'y': 'total_bill',
'order': ['Thur', 'Fri', 'Sat', 'Sun'],
'hue': 'day',
'palette': ['#ffa200', '#c8ff00', '#00ffe5', '#ee00ff'],
'width': 0.5,
'linecolor': 'k',
'linewidth': 2,
'ax': ax,
}
pairs = [('Thur', 'Fri'),
('Thur', 'Sat'),
('Thur', 'Sun'),
('Fri', 'Sat'),
('Fri', 'Sun'),
('Sat', 'Sun')]
pvalues = [6.477e-01, 1.276e-01, 3.585e-01, 4.690e-02, 2.680e-02, 5.363e-03]
sns.boxplot(**plot_params)
annotator = Annotator(pairs=pairs, **plot_params)
annotator.configure(fontsize=15,
text_format='simple',
pvalue_thresholds=[[1e-5, "1e-5"], [1e-4, "1e-4"], [1e-3, "0.001"], [1e-2, "0.01"], [5e-2, "0.05"]],
pvalue_format_string='{:.2f}',
alpha=0.05, # 显著性阈值
hide_non_significant=False, # 隐藏无显著性的组间注释
p_capitalized=True, # p字母是否大写
p_separators=(" ", " "), # 等号前后间隔符号：'both'(True)==(" ", " "), 'after'==("", " "), 'none'(False)==("", "")
test_short_name='M.W.W' # 计算p值所使用的检验方法名称
)
annotator.set_pvalues(pvalues)
annotator.annotate()
# annotator.set_pvalues_and_annotate(pvalues)
"""
Thur vs. Fri: Custom statistical test, P_val:6.477e-01
Fri vs. Sat: Custom statistical test, P_val:4.690e-02
Sat vs. Sun: Custom statistical test, P_val:5.363e-03
Thur vs. Sat: Custom statistical test, P_val:1.276e-01
Fri vs. Sun: Custom statistical test, P_val:2.680e-02
Thur vs. Sun: Custom statistical test, P_val:3.585e-01
(<Axes: xlabel='day', ylabel='total_bill'>,
[<statannotations.Annotation.Annotation at 0x2651bdb1280>,
<statannotations.Annotation.Annotation at 0x2651b4d98b0>,
<statannotations.Annotation.Annotation at 0x2651b4cd7f0>,
<statannotations.Annotation.Annotation at 0x2651b4cd190>,
<statannotations.Annotation.Annotation at 0x2651bda3cb0>,
<statannotations.Annotation.Annotation at 0x2651bda37a0>])
"""
plt.show()
带亚组（hue）的注释¶
fig = plt.figure(figsize=(8, 8), dpi=100, layout="constrained")
ax = fig.add_subplot(1, 1, 1)
plot_params = {'data': tips,
'x': 'sex',
'y': 'total_bill',
'order': ['Female', 'Male'],
'hue': 'day',
'hue_order': ['Thur', 'Fri', 'Sat', 'Sun'],
'palette': ['#ffa200', '#c8ff00', '#00ffe5', '#ee00ff'],
'width': 0.5,
'linecolor': 'k',
'linewidth': 2,
'ax': ax,
}
pairs = [[('Female', 'Thur'), ('Female', 'Fri')],
[('Female', 'Thur'), ('Female', 'Sat')],
[('Female', 'Thur'), ('Female', 'Sun')],
[('Female', 'Fri'), ('Female', 'Sat')],
[('Female', 'Fri'), ('Female', 'Sun')],
[('Female', 'Sat'), ('Female', 'Sun')],
[('Male', 'Thur'), ('Male', 'Fri')],
[('Male', 'Thur'), ('Male', 'Sat')],
[('Male', 'Thur'), ('Male', 'Sun')],
[('Male', 'Fri'), ('Male', 'Sat')],
[('Male', 'Fri'), ('Male', 'Sun')],
[('Male', 'Sat'), ('Male', 'Sun')]]
pvalues = [5.287e-01, 6.087e-02, 1.059e-01, 8.600e-02, 4.768e-02, 9.372e-01, 9.378e-01, 4.740e-01, 1.131e-01, 7.916e-01, 3.587e-01, 3.211e-01]
sns.boxplot(**plot_params)
annotator = Annotator(pairs=pairs, **plot_params)
annotator.configure(fontsize=15, text_format='simple', test_short_name='M.W.W')
annotator.set_pvalues_and_annotate(pvalues)
"""
Female_Thur vs. Female_Fri: Custom statistical test, P_val:5.287e-01
Female_Fri vs. Female_Sat: Custom statistical test, P_val:8.600e-02
Female_Sat vs. Female_Sun: Custom statistical test, P_val:9.372e-01
Male_Thur vs. Male_Fri: Custom statistical test, P_val:9.378e-01
Male_Fri vs. Male_Sat: Custom statistical test, P_val:7.916e-01
Male_Sat vs. Male_Sun: Custom statistical test, P_val:3.211e-01
Female_Thur vs. Female_Sat: Custom statistical test, P_val:6.087e-02
Female_Fri vs. Female_Sun: Custom statistical test, P_val:4.768e-02
Male_Thur vs. Male_Sat: Custom statistical test, P_val:4.740e-01
Male_Fri vs. Male_Sun: Custom statistical test, P_val:3.587e-01
Female_Thur vs. Female_Sun: Custom statistical test, P_val:1.059e-01
Male_Thur vs. Male_Sun: Custom statistical test, P_val:1.131e-01
(<Axes: xlabel='sex', ylabel='total_bill'>,
[<statannotations.Annotation.Annotation at 0x265190c3530>,
<statannotations.Annotation.Annotation at 0x2651b4b6480>,
<statannotations.Annotation.Annotation at 0x2651bd71490>,
<statannotations.Annotation.Annotation at 0x2651be640b0>,
<statannotations.Annotation.Annotation at 0x2651be65d30>,
<statannotations.Annotation.Annotation at 0x2651cf6d190>,
<statannotations.Annotation.Annotation at 0x2651d79c9b0>,
<statannotations.Annotation.Annotation at 0x2651a8cc0e0>,
<statannotations.Annotation.Annotation at 0x2651cf1af30>,
<statannotations.Annotation.Annotation at 0x2651aa2e090>,
<statannotations.Annotation.Annotation at 0x2651a99c080>,
<statannotations.Annotation.Annotation at 0x2651923ef60>])
"""
plt.show()
自动应用假设检验¶
支持的检验方法包括：
"Brunner-Munzel"
-
Mann-Whitney：
"Mann-Whitney"、"Mann-Whitney-gt"、"Mann-Whitney-1s" -
t-test：
"t-test_ind"、"t-test_paired" -
Welch's t-test：
"t-test_welch" -
Levene test：
"Levene" -
Wilcoxon test：
"Wilcoxon" -
Kruskal-Wallis test：
"Kruskal"
fig = plt.figure(figsize=(8, 8), dpi=100, layout="constrained")
ax = fig.add_subplot(1, 1, 1)
plot_params = {'data': tips,
'x': 'sex',
'y': 'total_bill',
'order': ['Female', 'Male'],
'hue': 'day',
'hue_order': ['Thur', 'Fri', 'Sat', 'Sun'],
'palette': ['#ffa200', '#c8ff00', '#00ffe5', '#ee00ff'],
'width': 0.5,
'linecolor': 'k',
'linewidth': 2,
'ax': ax,
}
pairs = [[('Female', 'Thur'), ('Female', 'Fri')],
[('Female', 'Thur'), ('Female', 'Sat')],
[('Female', 'Thur'), ('Female', 'Sun')],
[('Female', 'Fri'), ('Female', 'Sat')],
[('Female', 'Fri'), ('Female', 'Sun')],
[('Female', 'Sat'), ('Female', 'Sun')],
[('Male', 'Thur'), ('Male', 'Fri')],
[('Male', 'Thur'), ('Male', 'Sat')],
[('Male', 'Thur'), ('Male', 'Sun')],
[('Male', 'Fri'), ('Male', 'Sat')],
[('Male', 'Fri'), ('Male', 'Sun')],
[('Male', 'Sat'), ('Male', 'Sun')],]
sns.boxplot(**plot_params)
annotator = Annotator(pairs=pairs, **plot_params)
annotator.configure(fontsize=15, text_format='simple',
test='Mann-Whitney',
show_test_name=True
)
annotator.apply_test() # 可以传入并设置scipy对应假设检验函数的对应参数
annotator.annotate()
# annotator.apply_and_annotate()
annotator.new_plot(pairs=[('Female', 'Male')], data=tips, x='sex', y='total_bill', order=['Female', 'Male'], ax=ax)
annotator.apply_and_annotate()
"""
Female_Thur vs. Female_Fri: Mann-Whitney-Wilcoxon test two-sided, P_val:5.287e-01 U_stat=1.645e+02
Female_Fri vs. Female_Sat: Mann-Whitney-Wilcoxon test two-sided, P_val:8.600e-02 U_stat=7.700e+01
Female_Sat vs. Female_Sun: Mann-Whitney-Wilcoxon test two-sided, P_val:9.372e-01 U_stat=2.480e+02
Male_Thur vs. Male_Fri: Mann-Whitney-Wilcoxon test two-sided, P_val:9.378e-01 U_stat=1.470e+02
Male_Fri vs. Male_Sat: Mann-Whitney-Wilcoxon test two-sided, P_val:7.916e-01 U_stat=2.790e+02
Male_Sat vs. Male_Sun: Mann-Whitney-Wilcoxon test two-sided, P_val:3.211e-01 U_stat=1.528e+03
Female_Thur vs. Female_Sat: Mann-Whitney-Wilcoxon test two-sided, P_val:6.087e-02 U_stat=3.210e+02
Female_Fri vs. Female_Sun: Mann-Whitney-Wilcoxon test two-sided, P_val:4.768e-02 U_stat=4.200e+01
Male_Thur vs. Male_Sat: Mann-Whitney-Wilcoxon test two-sided, P_val:4.740e-01 U_stat=8.020e+02
Male_Fri vs. Male_Sun: Mann-Whitney-Wilcoxon test two-sided, P_val:3.587e-01 U_stat=2.365e+02
Female_Thur vs. Female_Sun: Mann-Whitney-Wilcoxon test two-sided, P_val:1.059e-01 U_stat=2.075e+02
Male_Thur vs. Male_Sun: Mann-Whitney-Wilcoxon test two-sided, P_val:1.131e-01 U_stat=6.895e+02
Female vs. Male: Mann-Whitney-Wilcoxon test two-sided, P_val:2.135e-02 U_stat=5.614e+03
(<Axes: xlabel='sex', ylabel='total_bill'>,
[<statannotations.Annotation.Annotation at 0x26521f06480>])
"""
plt.show()
多重检验p值校正¶
支持的校正方法包括：
-
Bonferroni ("bonf")
-
Benjamini-Hochberg ("BH")
-
Holm-Bonferroni ("HB")
-
Benjamini-Yekutieli ("BY")
fig = plt.figure(figsize=(8, 8), dpi=100, layout="constrained")
ax = fig.add_subplot(1, 1, 1)
plot_params = {'data': tips,
'x': 'sex',
'y': 'total_bill',
'order': ['Female', 'Male'],
'hue': 'day',
'hue_order': ['Thur', 'Fri', 'Sat', 'Sun'],
'palette': ['#ffa200', '#c8ff00', '#00ffe5', '#ee00ff'],
'width': 0.5,
'linecolor': 'k',
'linewidth': 2,
'ax': ax,
}
pairs = [[('Female', 'Thur'), ('Female', 'Fri')],
[('Female', 'Thur'), ('Female', 'Sat')],
[('Female', 'Thur'), ('Female', 'Sun')],
[('Female', 'Fri'), ('Female', 'Sat')],
[('Female', 'Fri'), ('Female', 'Sun')],
[('Female', 'Sat'), ('Female', 'Sun')],
[('Male', 'Thur'), ('Male', 'Fri')],
[('Male', 'Thur'), ('Male', 'Sat')],
[('Male', 'Thur'), ('Male', 'Sun')],
[('Male', 'Fri'), ('Male', 'Sat')],
[('Male', 'Fri'), ('Male', 'Sun')],
[('Male', 'Sat'), ('Male', 'Sun')]]
sns.boxplot(**plot_params)
annotator = Annotator(pairs=pairs, **plot_params)
annotator.configure(fontsize=15, text_format='simple',
test='Mann-Whitney',
show_test_name=True,
comparisons_correction="bonferroni"
)
annotator.apply_and_annotate()
annotator.new_plot(pairs=[('Female', 'Male')], data=tips, x='sex', y='total_bill', order=['Female', 'Male'], ax=ax)
annotator.apply_and_annotate()
"""
Female_Thur vs. Female_Fri: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=1.645e+02
Female_Fri vs. Female_Sat: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=7.700e+01
Female_Sat vs. Female_Sun: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=2.480e+02
Male_Thur vs. Male_Fri: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=1.470e+02
Male_Fri vs. Male_Sat: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=2.790e+02
Male_Sat vs. Male_Sun: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=1.528e+03
Female_Thur vs. Female_Sat: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:7.305e-01 U_stat=3.210e+02
Female_Fri vs. Female_Sun: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:5.721e-01 U_stat=4.200e+01
Male_Thur vs. Male_Sat: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=8.020e+02
Male_Fri vs. Male_Sun: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=2.365e+02
Female_Thur vs. Female_Sun: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=2.075e+02
Male_Thur vs. Male_Sun: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:1.000e+00 U_stat=6.895e+02
Female vs. Male: Mann-Whitney-Wilcoxon test two-sided with Bonferroni correction, P_val:2.135e-02 U_stat=5.614e+03
(<Axes: xlabel='sex', ylabel='total_bill'>,
[<statannotations.Annotation.Annotation at 0x26523cd4260>])
"""
plt.show()
