如何在R中绘制仅以疾病为分区的circlize弦图？

要实现仅以疾病为分区的弦图，核心是先将基因×疾病的原始矩阵转换为疾病×疾病的关联矩阵，再传入chordDiagram()函数。以下是具体步骤：

1. 加载所需工具包 #

library(circlize)
library(tidyr)
library(dplyr)

2. 转换原始数据为长格式 #

先筛选出有药物记录的基因-疾病对，便于后续构建疾病间关联：

# 将矩阵转为长格式，仅保留药物数>0的记录
df_long <- as.data.frame(df) %>%
  rownames_to_column("gene") %>%
  pivot_longer(cols = -gene, names_to = "disease", values_to = "drug_count") %>%
  filter(drug_count > 0)

3. 构建疾病间的关联权重矩阵 #

我们需要计算每对疾病之间，由共享基因带来的药物数量总和（可根据需求调整权重计算方式，比如用乘积替代求和）：

# 生成每个基因对应的疾病两两组合
disease_pairs <- df_long %>%
  group_by(gene) %>%
  summarize(diseases = list(disease)) %>%
  filter(lengths(diseases) >= 2) %>%  # 仅保留关联至少两种疾病的基因
  mutate(pairs = map(diseases, ~ combn(.x, 2, simplify = FALSE))) %>%
  unnest(pairs) %>%
  mutate(disease1 = sapply(pairs, `[`, 1),
         disease2 = sapply(pairs, `[`, 2)) %>%
  select(-diseases, -pairs)

# 计算每对疾病的总权重（此处用基因在两个疾病中的药物数之和累加）
pair_weights <- disease_pairs %>%
  left_join(df_long, by = c("gene", "disease1" = "disease")) %>%
  rename(drug1 = drug_count) %>%
  left_join(df_long, by = c("gene", "disease2" = "disease")) %>%
  rename(drug2 = drug_count) %>%
  mutate(total = drug1 + drug2) %>%
  group_by(disease1, disease2) %>%
  summarize(weight = sum(total), .groups = "drop")

# 构建对称的疾病关联矩阵
disease_list <- unique(df_long$disease)
disease_matrix <- matrix(0, 
                         nrow = length(disease_list), 
                         ncol = length(disease_list),
                         dimnames = list(disease_list, disease_list))

# 填充矩阵
for(i in 1:nrow(pair_weights)) {
  d1 <- pair_weights$disease1[i]
  d2 <- pair_weights$disease2[i]
  disease_matrix[d1, d2] <- pair_weights$weight[i]
  disease_matrix[d2, d1] <- pair_weights$weight[i]
}

4. 绘制仅含疾病分区的弦图 #

传入疾病关联矩阵，设置仅显示疾病分区：

# 绘制弦图，仅保留网格轨道
chordDiagram(disease_matrix, 
             annotationTrack = "grid", 
             preAllocateTracks = list(track.height = 0.1))

# 添加疾病标签（优化显示方向）
circos.trackPlotRegion(track.index = 1, 
                       panel.fun = function(x, y) {
                         circos.text(CELL_META$xcenter, CELL_META$ylim[1], 
                                     CELL_META$sector.index, 
                                     facing = "clockwise", 
                                     niceFacing = TRUE, 
                                     adj = c(0, 0.5))
                       }, 
                       bg.border = NA)

关键说明 #

直接用原始的基因×疾病矩阵绘制时，chordDiagram()会将行（基因）和列（疾病）都视为独立分区，因此会同时显示两者。转换为疾病×疾病矩阵后，行和列都是疾病类别，最终弦图就只会保留疾病分区。

内容的提问来源于stack exchange，提问作者DN1