各组显著上下调差异表达基因统计及可视化简化方案需求

问题需求 #

在存储差异表达值的data.frame中，按组统计显著上调/下调基因数量（显著性由FDR≤0.05和fold-change阈值定义），并生成可视化图；额外要求在图中区分不同FC水平（如0.5、1、2、4、>4），同时简化当前过于繁琐的实现流程。

示例数据 #

# 创建差异表达数据框
gene_creator <- paste("gene", 1:1000, sep = "")
genes <- sample(gene_creator, 100)

dex_A <- data.frame(
  gene = genes,
  group = "group_A",
  logFC = sample(-5:5, replace = TRUE, size = 100),
  FDR = sample(c(0.01, 1), replace = TRUE, size = 100)
)

dex_B <- data.frame(
  gene = genes,
  group = "group_B",
  logFC = sample(-5:5, replace = TRUE, size = 100),
  FDR = sample(c(0.01, 1), replace = TRUE, size = 100)
)

dex_C <- data.frame(
  gene = genes,
  group = "group_C",
  logFC = sample(-5:5, replace = TRUE, size = 100),
  FDR = sample(c(0.01, 1), replace = TRUE, size = 100)
)

dex_D <- data.frame(
  gene = genes,
  group = "group_D",
  logFC = sample(-5:5, replace = TRUE, size = 100),
  FDR = sample(c(0.01, 1), replace = TRUE, size = 100)
)

dex_df <- rbind(dex_A, dex_B, dex_C, dex_D)

简化实现方案 #

利用tidyverse的流式语法，一次性完成基因分类、统计与可视化，无需拆分上下调数据再合并：

library(tidyverse)

# 1. 数据处理：筛选显著基因 + 标记方向与FC水平 + 统计数量
dex_summary <- dex_df %>%
  # 过滤显著基因（FDR阈值可按需调整）
  filter(FDR <= 0.05) %>%
  # 标记上调/下调方向
  mutate(direction = case_when(
    logFC > 0 ~ "上调",
    logFC < 0 ~ "下调",
    TRUE ~ "无差异" # 实际会被FDR过滤，仅作兜底
  )) %>%
  # 标记FC水平区间（可按需修改区间阈值）
  mutate(fc_level = case_when(
    abs(logFC) <= 0.5 ~ "0-0.5",
    abs(logFC) <= 1 ~ "0.5-1",
    abs(logFC) <= 2 ~ "1-2",
    abs(logFC) <= 4 ~ "2-4",
    abs(logFC) > 4 ~ ">4"
  )) %>%
  # 按组、方向、FC水平统计基因数
  group_by(group, direction, fc_level) %>%
  summarise(n = n(), .groups = "drop") %>%
  # 下调基因数设为负数，方便绘图时向下展示
  mutate(count = if_else(direction == "下调", -n, n))

# 2. 可视化：堆叠柱状图展示各组各FC水平的上下调基因数
ggplot(dex_summary, aes(x = group, y = count, fill = fc_level)) +
  geom_col(position = "stack") +
  # 在柱子中间添加基因数量标签
  geom_text(aes(label = abs(n)), position = position_stack(vjust = 0.5), size = 3) +
  # 设置填充色方案
  scale_fill_brewer(palette = "Set1") +
  # 自定义标题与坐标轴标签
  labs(
    title = "各组差异表达基因数量（按FC水平划分）",
    x = "分组",
    y = "基因数量",
    fill = "FC区间"
  ) +
  # 添加水平线区分上下调区域
  geom_hline(yintercept = 0, linetype = "solid", color = "black", linewidth = 0.8) +
  theme_minimal()

方案优势 #

1. 流程简化：仅需一次数据流转，无需拆分上下调统计再合并，代码更简洁易维护；
2. 扩展性强：修改FC区间阈值仅需调整case_when的条件，新增分组或统计维度也无需大幅改动；
3. 信息完整：同时展示上调/下调方向与FC水平，满足额外需求的同时，可视化结果更直观。

内容的提问来源于stack exchange，提问作者Sebastian Hesse