如何在Python中翻转igraph树，解决树形图倒置问题？

问题 #

使用Python的igraph构建树形图时，顶层节点"HoPE"显示在底部，树形结构倒置，需要调整为从上到下展开的样式。

解决方案 #

以下三种方法都可以解决该问题，优先推荐方法1，操作最简便：

方法1：反转Plotly的Y轴 #

直接在Plotly的布局配置中反转Y轴方向，无需修改坐标数据。

修改代码中的fig.update_layout部分，在yaxis配置里添加autorange="reversed"：

fig.update_layout(
    showlegend=True,
    hovermode="closest",
    margin=dict(b=0, l=0, r=0, t=0),
    xaxis=dict(showgrid=False, zeroline=False, showticklabels=False),
    yaxis=dict(showgrid=False, zeroline=False, showticklabels=False, autorange="reversed")
)

方法2：手动翻转坐标值 #

对所有节点和边的Y坐标取反，实现方向翻转。

在获取节点和边的坐标后添加以下代码：

# 翻转所有节点的Y坐标
node_y = [-y for y in node_y]
# 翻转所有边的Y坐标
edge_y = [-y for y in edge_y]

方法3：指定igraph布局根节点并调整方向 #

在生成Reingold-Tilford布局时指定根节点为"HoPE"，再翻转坐标：

# 找到HoPE节点的索引
root_idx = g.vs.find("HoPE").index
# 以HoPE为根生成布局
layt = g.layout("rt", root=root_idx)

# 后续步骤中翻转Y坐标
node_y = [-y for y in node_y]
edge_y = [-y for y in edge_y]

修改后的完整代码（以方法1为例） #

import igraph
import plotly.graph_objects as go

# Create an empty graph
g = igraph.Graph(directed=True)

# Add vertices with labels
g.add_vertex("HoPE")

# add FA node
for a in ['FA']:
    fa_node = f"HoPE_{a}"
    g.add_vertex(fa_node)
    g.add_edge('HoPE', fa_node)

# add exp node
for b in ['exp']:
    exp_node = f"HoPE_{a}_{b}"
    g.add_vertex(exp_node)
    g.add_edge(fa_node, exp_node)

# add TX node
for c in ['TX']:
    tx_node = f"HoPE_{a}_{b}_{c}"
    g.add_vertex(tx_node)
    g.add_edge(exp_node, tx_node)

# Add C nodes
for condition in ["C1", "C2", "C3", "C4"]:
    c_nodes = f"HoPE_FA_exp_TX_{condition}"
    g.add_vertex(c_nodes)
    g.add_edge(tx_node, c_nodes)

    # Add TR nodes
    for i in range(1, 4):  # TR1, TR2, TR3
        tr_node = f"HoPE_{a}_{b}_{c}_{condition}_TR{i}"
        g.add_vertex(tr_node)
        g.add_edge(c_nodes, tr_node)

        # Add 5 AR nodes for each TR node
        for j in range(1, 6):  # AR1, AR2, AR3, AR4, AR5
            ar_node = f"HoPE_{a}_{b}_{c}_{condition}_TR{i}_AR{j}"
            g.add_vertex(ar_node)
            g.add_edge(tr_node, ar_node)

# Convert igraph layout to Plotly format
layt = g.layout("rt")

# Get edge coordinates
edge_x, edge_y = [], []
for edge in g.get_edgelist():
    x0, y0 = layt[edge[0]]
    x1, y1 = layt[edge[1]]
    edge_x.extend([x0, x1, None])
    edge_y.extend([y0, y1, None])

# Get vertex coordinates and labels
node_x, node_y = zip(*[layt[vertex] for vertex in g.vs.indices])
node_labels = g.vs['name']

# Create Plotly trace for edges
edge_trace = go.Scatter(
    x=edge_x,
    y=edge_y,
    line=dict(width=0.5, color="#888"),
    hoverinfo="none",
    mode="lines",
)

# Create Plotly trace for nodes
node_trace = go.Scatter(
    x=node_x,
    y=node_y,
    mode="markers",
    hoverinfo="text",
    marker=dict(showscale=True, colorscale="YlGnBu", size=10),
    text=node_labels,  # Display node labels on hover
)

# Create a figure and add traces
fig = go.Figure(data=[edge_trace, node_trace])

# Customize layout - 添加了autorange="reversed"
fig.update_layout(
    showlegend=True,
    hovermode="closest",
    margin=dict(b=0, l=0, r=0, t=0),
    xaxis=dict(showgrid=False, zeroline=False, showticklabels=False),
    yaxis=dict(showgrid=False, zeroline=False, showticklabels=False, autorange="reversed"),
)

# Show the interactive plot
fig.show()
print(node_labels)

内容的提问来源于stack exchange，提问作者yade acem