R語言ggplot2繪制熱圖展示GO富集分析結果的是怎樣的

# R語言ggplot2繪制熱圖展示GO富集分析結果的是怎樣的

## 摘要
基因本體論(GO)富集分析是生物信息學中解讀高通量數據的核心方法。本文詳細介紹如何使用R語言中的ggplot2包將GO富集結果轉化為直觀的熱圖可視化，包括數據預處理、圖形定制和結果解讀的全流程。通過完整的代碼示例和參數解析，幫助研究者掌握專業級GO熱圖的繪制技巧。

## 1. GO富集分析與可視化概述

### 1.1 GO富集分析原理
基因本體論(Gene Ontology, GO)通過三個層次描述基因功能：
- 分子功能(Molecular Function)
- 生物過程(Biological Process) 
- 細胞組分(Cellular Component)

富集分析通過統計檢驗識別在差異表達基因中顯著過表征的GO term，常用方法包括：
- 超幾何檢驗
- Fisher精確檢驗
- GSEA算法

### 1.2 可視化需求
原始富集結果通常包含：
- Term名稱
- P值/q值
- 富集因子
- 基因數量

熱圖通過顏色和大小雙重編碼可同時展示：
- 顯著性水平(-log10(p-value))
- 富集程度(基因比例)
- 術語間層次關系

## 2. 數據準備與預處理

### 2.1 示例數據加載
```r
# 模擬GO富集結果
go_terms <- data.frame(
  ID = c("GO:0008152", "GO:0009987", "GO:0002376", 
         "GO:0006955", "GO:0006950"),
  Description = c("metabolic process", "cellular process", 
                 "immune system", "immune response", 
                 "response to stress"),
  GeneRatio = c(120/1000, 85/1000, 45/1000, 30/1000, 25/1000),
  BgRatio = c(500/10000, 600/10000, 200/10000, 150/10000, 100/10000),
  pvalue = c(1e-12, 1e-8, 1e-5, 0.001, 0.01),
  p.adjust = c(1e-10, 1e-6, 1e-4, 0.0005, 0.005),
  qvalue = c(1e-10, 1e-6, 1e-4, 0.0004, 0.004),
  Count = c(120, 85, 45, 30, 25),
  Category = c("BP", "BP", "BP", "BP", "BP")
)

2.2 數據轉換關鍵步驟

library(dplyr)

plot_data <- go_terms %>%
  mutate(
    log_p = -log10(pvalue),  # 轉換p值
    GeneRatio_num = sapply(strsplit(as.character(GeneRatio), "/"), 
                          function(x) as.numeric(x[1])/as.numeric(x[2])),
    Description = factor(Description, levels = rev(unique(Description)))

3. 基礎熱圖繪制

3.1 最小代碼示例

library(ggplot2)

ggplot(plot_data, aes(x = Category, y = Description)) +
  geom_tile(aes(fill = log_p), color = "white") +
  scale_fill_gradient(low = "blue", high = "red") +
  theme_minimal()

3.2 核心圖層解析

• geom_tile(): 創建熱圖矩陣
• aes(fill): 顏色映射變量
• color: 格子邊框顏色
• scale_fill_gradient(): 連續顏色標度

4. 高級定制技巧

4.1 多維度編碼

ggplot(plot_data, aes(x = Category, y = Description)) +
  geom_point(aes(size = Count, color = log_p)) +
  scale_color_gradientn(colors = c("blue", "yellow", "red")) +
  scale_size(range = c(3, 10)) +
  theme_bw(base_size = 12) +
  labs(x = "", y = "", 
       color = "-log10(p-value)", 
       size = "Gene Count")

4.2 分面展示

# 當有多個比較組時
plot_data$Group <- rep(c("Treatment", "Control"), each = 3)[1:5]

ggplot(plot_data, aes(x = Group, y = Description)) +
  geom_tile(aes(fill = log_p)) +
  facet_grid(. ~ Category, scales = "free") +
  scale_fill_viridis_c(option = "magma")

4.3 文本標簽優化

ggplot(plot_data, aes(x = Category, y = Description)) +
  geom_tile(aes(fill = log_p), alpha = 0.8) +
  geom_text(aes(label = sprintf("%.1f", log_p)), 
            color = "white", size = 3) +
  scale_fill_distiller(palette = "Spectral") +
  theme(
    axis.text.x = element_text(angle = 45, hjust = 1),
    plot.title = element_text(hjust = 0.5))

5. 配色方案選擇

5.1 常用配色方案

5.2 自定義配色

my_palette <- colorRampPalette(c("#2E86AB", "#F24236"))(10)

ggplot(plot_data) +
  geom_tile(aes(x = Category, y = Description, fill = log_p)) +
  scale_fill_gradientn(colors = my_palette)

6. 結果解讀與導出

6.1 圖形元素解讀

• X軸：通常為分組或GO類別
• Y軸：GO term描述
• 顏色：-log10(p-value)表示顯著性
• 點大小：富集基因數量

6.2 圖形導出

ggsave("GO_heatmap.pdf", 
       width = 10, height = 6, 
       dpi = 300, device = cairo_pdf)

# 高分辨率TIFF格式
ggsave("GO_heatmap.tiff", 
       compression = "lzw", 
       units = "in", width = 8, height = 5)

7. 完整案例代碼

library(ggplot2)
library(dplyr)

# 數據準備
data <- clusterProfiler::enrichGO(...) %>% 
  as.data.frame() %>%
  filter(p.adjust < 0.05) %>%
  arrange(pvalue) %>%
  head(20) %>%
  mutate(
    log_p = -log10(p.adjust),
    Description = stringr::str_wrap(Description, width = 40))

# 高級熱圖
ggplot(data, aes(x = GeneRatio_num, y = reorder(Description, log_p))) +
  geom_point(aes(size = Count, color = log_p)) +
  scale_color_gradientn(
    colors = rev(RColorBrewer::brewer.pal(11, "Spectral")),
    limits = c(0, max(data$log_p))) +
  scale_size_continuous(range = c(3, 8)) +
  facet_grid(ONTOLOGY ~ ., scales = "free", space = "free") +
  labs(x = "Gene Ratio", y = "", 
       color = "-log10(adj.p)", 
       size = "Gene Count",
       title = "GO Enrichment Analysis") +
  theme_classic(base_size = 12) +
  theme(
    strip.background = element_rect(fill = "grey90"),
    panel.spacing = unit(0.2, "lines"),
    axis.text.y = element_text(lineheight = 0.8))

8. 常見問題解決

8.1 長文本處理

plot_data %>%
  mutate(Description = stringr::str_wrap(Description, width = 30)) %>%
  ggplot(aes(...)) + ...

8.2 過密標簽

theme(axis.text.y = element_text(size = 8))
scale_y_discrete(labels = function(x) substr(x, 1, 20))

8.3 缺失值處理

scale_fill_gradient(na.value = "gray90")

9. 擴展應用

9.1 結合KEGG結果

bind_rows(
  mutate(go_data, Type = "GO"),
  mutate(kegg_data, Type = "KEGG")) %>%
  ggplot(aes(x = Type, ...)) + ...

9.2 動態交互熱圖

## 參考文獻
1. Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer, 2016.
2. Yu G. et al. clusterProfiler: an R package for comparing biological themes. Bioinformatics, 2012.
3. RStudio ggplot2 Cheat Sheet