如何绘制含Jaccard指数数值的热图并展示亚型元素与得分？

问题根源 #

你的原代码逻辑是正确的，但和在线工具的差异大概率来自可视化细节缺失（比如未显示Jaccard数值、未标注分组对应的元素集合、保留了聚类树），或是二分矩阵的构建未严格遵循集合逻辑（Jaccard基于元素存在/不存在，而非计数）。

测试数据优化方案 #

以下代码调整了可视化参数，同时确保Jaccard计算的严谨性，结果更贴近在线工具的展示效果：

library(vegan)
library(pheatmap)

# 测试数据
df <- data.frame(tumor_type = c("tumor1", "tumor1", "tumor1", "tumor2", "tumor2", "tumor3", "tumor4", "tumor4"), 
                 genes = c("geneA", "geneB", "geneC", "geneA", "geneD", "geneD", "geneA", "geneD"))

# 构建严格的二分矩阵：行=tumor_type，列=genes，存在为1，不存在为0
binary_matrix <- table(df$tumor_type, df$genes) %>% as.matrix()
binary_matrix[binary_matrix > 0] <- 1  # 转换为集合存在标记，忽略重复

# 计算Jaccard相似度（1 - Jaccard距离）
jaccard_sim <- 1 - as.matrix(vegdist(binary_matrix, method = "jaccard"))

# 准备行注释：展示每个肿瘤亚型对应的基因集合
annot_row <- data.frame(Genes = sapply(rownames(jaccard_sim), function(x) {
  paste(colnames(binary_matrix)[binary_matrix[x,] == 1], collapse = ", ")
}))

# 绘制热图，匹配在线工具的核心特征
pheatmap(jaccard_sim,
         annotation_row = annot_row,
         display_numbers = TRUE,  # 显示单元格的Jaccard得分
         number_format = "%.2f",
         main = "肿瘤亚型Jaccard相似度热图",
         treeheight_row = 0, treeheight_col = 0,  # 关闭聚类树（若在线工具无此结构）
         cellwidth = 80, cellheight = 80)

实际GO Term数据处理方案 #

针对你提供的Comparison-GO Term数据集，按以下步骤绘制符合预期的热图：

library(vegan)
library(pheatmap)

# 实际数据子集
actual_df <- structure(list(Comparison = c("M0_vs_M1", "M0_vs_M1", "M0_vs_M1", 
"M0_vs_M1", "M0_vs_M1", "M0_vs_M1", "M4_vs_M5", "M4_vs_M5", "M4_vs_M5", 
"M4_vs_M5", "M4_vs_M5", "M4_vs_M5"), Term = c("GO:0042742", "GO:0009617", 
"GO:0006959", "GO:0006956", "GO:0006958", "GO:0002455", "GO:0048863", 
"GO:0048864", "GO:0001667", "GO:0048483", "GO:0051129", "GO:0010977"
)), row.names = c(1L, 2L, 3L, 4L, 5L, 6L, 3266L, 3267L, 3268L, 
3269L, 3270L, 3271L), class = "data.frame")

# 构建GO Term二分矩阵
binary_go_matrix <- table(actual_df$Comparison, actual_df$Term) %>% as.matrix()
binary_go_matrix[binary_go_matrix > 0] <- 1

# 计算Jaccard相似度
jaccard_go_sim <- 1 - as.matrix(vegdist(binary_go_matrix, method = "jaccard"))

# 准备行注释：换行展示每个Comparison的GO Term列表
annot_go_row <- data.frame(GO_Terms = sapply(rownames(jaccard_go_sim), function(x) {
  paste(colnames(binary_go_matrix)[binary_go_matrix[x,] == 1], collapse = "\n")
}))

# 绘制热图
pheatmap(jaccard_go_sim,
         annotation_row = annot_go_row,
         display_numbers = TRUE,
         number_format = "%.2f",
         main = "Comparison组GO Term Jaccard相似度热图",
         treeheight_row = 0, treeheight_col = 0,
         cellwidth = 100, cellheight = 100,
         fontsize_row = 10, fontsize_col = 10)

核心调整说明 #

• 二分矩阵修正：将计数转换为1，确保Jaccard计算基于元素集合的存在/不存在（符合Jaccard指数的定义）
• 数值显示：添加display_numbers直接展示相似度数值，和在线工具对齐
• 注释增强：添加行注释展示每个分组对应的元素集合，让热图信息更直观
• 聚类控制：关闭聚类树（若在线工具无聚类结构），避免干扰核心信息展示
• 可视化优化：调整单元格大小、字体大小，提升可读性

内容的提问来源于stack exchange，提问作者PesKchan