Sankey风格可视化

import pandas as pd
import plotly.graph_objs as go
import plotly.express as px
import os

print("当前工作目录是：", os.getcwd())

示例数据
data = {
'group': ['Sample1', 'Sample1', 'Sample1', 'Sample2', 'Sample2', 'Sample2'],
'celltype': ['Type1', 'Type2', 'Type3', 'Type1', 'Type2', 'Type3'],
'n': [100, 200, 150, 120, 180, 130]
}
scedata = pd.DataFrame(data)

计算相对频率
celltype_ratio = scedata.groupby(['group', 'celltype'])['n'].sum().reset_index()
celltype_ratio['relative_freq'] = celltype_ratio.groupby('group')['n'].transform(lambda x: x / x.sum())

颜色调色板
mycolor = ['#efb306', '#7db954', '#852f88', '#4e54ac', '#0f8096']

堆叠柱状图函数
def create_stacked_bar(df):
fig = px.bar(df, x='relative_freq', y='group', color='celltype',
color_discrete_sequence=mycolor,
orientation='h',
labels={'relative_freq': '比例', 'group': '样本'},
title='细胞类型分布')
fig.update_layout(
xaxis_title='比例',
yaxis_title='样本',
legend_title='细胞类型'
)
return fig

类Sankey图可视化函数
def create_sankey_like_plot(df):
groups = df['group'].unique()
celltypes = df['celltype'].unique()

traces = []

for i, group in enumerate(groups):
group_data = df[df['group'] == group]

# 柱子
column_bar = go.Bar(
x=[group_data['relative_freq'].sum()],
y=[group],
orientation='h',
marker_color='lightgray',
width=0.5,
name=group
)
traces.append(column_bar)

# 细胞类型分段
bottom = 0
for j, celltype in enumerate(group_data['celltype']):
value = group_data[group_data['celltype'] == celltype]['relative_freq'].values[0]

segment = go.Bar(
x=[value],
y=[group],
orientation='h',
marker_color=mycolor[j % len(mycolor)],
width=0.5,
base=bottom,
name=celltype
)
traces.append(segment)

bottom += value

# 布局
layout = go.Layout(
title='细胞类型分布(Sankey风格)',
xaxis=dict(title='比例'),
yaxis=dict(title='样本'),
barmode='stack',
height=400,
width=800
)

fig = go.Figure(data=traces, layout=layout)
return fig
查看全部
模拟 Alluvial 图风格的调整
def create_alluvial_like_plot(df):
fig = go.Figure()

# 绘制每个细胞类型的流动
for idx, group in enumerate(df['group'].unique()):
group_data = df[df['group'] == group]
bottom = 0
for i, celltype in enumerate(group_data['celltype']):
value = group_data[group_data['celltype'] == celltype]['relative_freq'].values[0]
fig.add_trace(go.Scatter(
x=[value, value],
y=[group, group],
mode='lines+markers',
line=dict(width=5, color=mycolor[i]),
name=celltype
))
fig.update_layout(
title='冲击桑基图（Alluvial风格）',
xaxis_title='比例',
yaxis_title='样本',
showlegend=True,
height=400,
width=800
)

return fig
查看全部
生成图表
stacked_bar = create_stacked_bar(celltype_ratio)
sankey_like = create_sankey_like_plot(celltype_ratio)
alluvial_like = create_alluvial_like_plot(celltype_ratio)

保存为HTML
stacked_bar.write_html("stacked_bar_chart.html")
sankey_like.write_html("sankey_like_plot.html")
alluvial_like.write_html("alluvial_like_plot.html")

print("图表已生成为 stacked_bar_chart.html, sankey_like_plot.html 和 alluvial_like_plot.html")
修改代码，将同类型的的type链接起来，例如这样操作
#冲击桑基图
ggplot(Ratio, aes(x =group, y= relative_freq, fill = celltype,
stratum=celltype, alluvium=celltype)) +
geom_col(width = 0.5, color='black')+
geom_flow(width=0.5,alpha=0.4, knot.pos=0)+
theme_classic() +
labs(x='Sample',y = 'Ratio')+
coord_flip()+
scale_fill_manual(values = mycolor)
，请优化代码实现