如何将二分图转换为自定义颜色刻度的单元格矩阵（R语言）

需求说明 #

我用R的bipartite包生成了二分图，但希望转换成论文常用的带颜色单元格的矩阵图——这种形式更适合大数据集的读取与可视化。要求：

• 基于交互值为单元格上色
• 支持自定义颜色刻度范围

附原二分图代码与数据集：

原二分图代码 #

plotweb(data.mat,
        method="normal",
        text.rot = 90,
        labsize = 1,
        y.width.low = 0.05,
        col.high = "#66A61E",
        col.interaction = "#E6AB02",
        col.low = "#E7298A",
        bor.col.interaction =  NA,
        arrow="none",
        empty=TRUE,
        plot.axes=FALSE,
        ybig=1)

数据集 #

data.mat <- structure(c(7336L, 1640L, 26293L, 111533L, 71255L, 161705L, 104046L, 
169975L, 395822L, 153999L, 6780L, 10866L, 131162L, 239110L, 130786L, 
88922L, 46725L, 99169L, 109625L, 203947L, 2258L, 550L, 47109L, 
1813L, 362753L, 14451L, 17025L, 30251L, 308L, 34412L, 11L, 0L, 
56988L, 1360L, 4L, 5L, 107L, 31L, 56L, 868L, 31775L, 31L, 327L, 
19L, 180L, 4284L, 77L, 27L, 0L, 854L, 5L, 143L, 4622L, 264L, 
7945L, 51L, 14545L, 32L, 8L, 126L, 66L, 0L, 6L, 14130L, 5L, 9L, 
165L, 1017L, 5L, 115L, 65L, 5L, 9572L, 77L, 820L, 165L, 521L, 
788L, 17L, 538L), dim = c(10L, 8L), dimnames = list(c("Arctium lappa", 
"Cichorium intybus", "Circium arvense", "Circium vulgare", "Hypericum perforatum", 
"Leucanthemum vulgare", "Plantago lanceolata", "Trifolium pratense", 
"Trifolium repens", "Tripleurospermum inodorum"), c("Cladosporium", 
"Alternaria", "Aureobasidium", "X....1", "Golovinomyces", "Podosphaera", 
"Ampelomyces", "X....2")))

解决方案 #

方法1：用ggplot2实现高度自定义的矩阵热图 #

ggplot2可完全控制颜色刻度、标签与布局，适合论文级可视化：

library(ggplot2)
library(tidyr)

# 1. 将矩阵转换为长格式数据框
data_df <- as.data.frame(data.mat) %>%
  tibble::rownames_to_column("Plant") %>%
  pivot_longer(cols = -Plant, names_to = "Fungus", values_to = "Interaction")

# 2. 绘制热图，自定义颜色刻度
ggplot(data_df, aes(x = Fungus, y = Plant, fill = Interaction)) +
  geom_tile(color = "white") + # 单元格白色边框
  # 自定义颜色渐变，手动设置刻度范围
  scale_fill_gradient2(low = "#E7298A", mid = "#E6AB02", high = "#66A61E",
                       midpoint = median(data_df$Interaction),
                       limits = c(0, max(data_df$Interaction)), # 自定义刻度上下限
                       name = "交互值") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1), # X轴标签旋转90度
    axis.title = element_text(size = 12),
    legend.position = "right"
  ) +
  labs(x = "真菌", y = "植物")

方法2：用pheatmap快速生成矩阵热图 #

pheatmap包专门用于热图绘制，操作更快捷：

library(pheatmap)

# 直接绘制，自定义颜色与刻度
pheatmap(data.mat,
         color = colorRampPalette(c("#E7298A", "#E6AB02", "#66A61E"))(100),
         breaks = seq(0, max(data.mat), length.out = 101), # 自定义刻度范围
         border_color = "white",
         angle_col = 90, # X轴标签旋转90度
         main = "植物-真菌交互矩阵热图",
         legend_title = "交互值"
)

关键说明 #

• 两种方法均支持自定义颜色刻度范围：ggplot2通过scale_fill_gradient2的limits参数，pheatmap通过breaks参数设置
• 颜色沿用原代码配色：#E7298A（低交互值）、#E6AB02（中间值）、#66A61E（高交互值）
• 矩阵热图比二分图更适合展示大数据集的数值分布，便于快速定位高/低交互值的组合

内容的提问来源于stack exchange，提问作者Katherine Chau