Claude vs DeepSeek：相关性分析与“星号”标记热图的高效绘制

# 创建模拟数据
set.seed(123) # 保证结果可重复
matrix1 <- matrix(rnorm(100*5), nrow=100, ncol=5) # 100个样本, 5个变量
colnames(matrix1) <- paste0("VarX", 1:5)
matrix2 <- matrix(rnorm(100*8), nrow=100, ncol=8) # 100个样本, 8个变量
colnames(matrix2) <- paste0("VarY", 1:8)

# 或者使用内置数据集
# library(dplyr)
# data_subset <- select(mtcars, mpg, cyl, disp, hp, wt, qsec)

# 加载所需的R包
library(psych)
library(ComplexHeatmap)
library(circlize)  # 用于自定义热图颜色

# 读取数据
otu_data <- read.csv("OTU.csv", row.names = 1, check.names = FALSE)
env_data <- read.csv("envs.csv", row.names = 1, check.names = FALSE)

# 转置OTU数据使样本在行，OTU在列
otu_data <- t(otu_data)

# 确保两个数据框中的样本顺序一致
# 假设两个文件中的样本ID相同
common_samples <- intersect(rownames(otu_data), rownames(env_data))
otu_data <- otu_data[common_samples, ]
env_data <- env_data[common_samples, ]

# 使用psych包进行相关性分析
cor_result <- corr.test(env_data, otu_data, method = "pearson", adjust = "fdr")

# 提取相关性系数矩阵和p值矩阵
cor_matrix <- cor_result$r# 相关性系数矩阵
p_matrix <- cor_result$p    # p值矩阵

# 设置热图颜色
col_fun <- colorRamp2(c(-1, 0, 1), c("#9370DB", "white", "#FF7F24"))

# 创建热图
heatmap <- Heatmap(
  cor_matrix,
  name = "correlation", 
  col = col_fun,
  rect_gp = gpar(col = "white", lwd = 1.2),  # 单元格描边为白色，粗细为1.2
  cluster_rows = TRUE,    # 对行进行聚类
  cluster_columns = TRUE, # 对列进行聚类
  show_row_names = TRUE,  
  show_column_names = TRUE,
  row_names_gp = gpar(fontsize = 7),    # 行名文本大小为7
  column_names_gp = gpar(fontsize = 7), # 列名文本大小为7
  width = ncol(cor_matrix) * unit(0.7, "cm"),  # 设置单元格宽度为0.7cm
  height = nrow(cor_matrix) * unit(0.7, "cm"), # 设置单元格高度为0.7cm
  row_dend_width = unit(1.5, "cm"),    # 行聚类树高度为1.5cm
  column_dend_height = unit(1.5, "cm"), # 列聚类树高度为1.5cm
  heatmap_legend_param = list(title_gp = gpar(fontsize = 7), 
                              labels_gp = gpar(fontsize = 7)),
  cell_fun = function(j, i, x, y, width, height, fill) {
    # 根据p值添加显著性标记
    p_value <- p_matrix[i, j]
    if (p_value < 0.01) {
      grid.text("**", x, y, gp = gpar(fontsize = 10, col = "black"))
    } elseif (p_value < 0.05) {
      grid.text("*", x, y, gp = gpar(fontsize = 10, col = "black"))
    }
  }
)
heatmap

# 绘制热图
pdf("OTU_ENV_Correlation_Heatmap.pdf", width = 10, height = 8)
draw(heatmap)
dev.off()

# 加载必要的包
library(psych)
library(ComplexHeatmap)
library(circlize)

# 读取数据
otu <- read.csv("OTU.csv", row.names = 1)
envs <- read.csv("envs.csv", row.names = 1)

# 转置OTU数据（样本为行，OTU为列）
otu_t <- t(otu)

# 按样本名称排序以对齐数据
otu_t <- otu_t[order(rownames(otu_t)), ]
envs <- envs[order(rownames(envs)), ]

# 验证样本顺序一致
stopifnot(all(rownames(otu_t) == rownames(envs)))

# 计算Pearson相关系数及p值
cor_result <- corr.test(otu_t, envs, method = "pearson")
cor_matrix <- cor_result$r
p_matrix <- cor_result$p

# 转置矩阵，使行=环境因子，列=OTU
cor_matrix_transposed <- t(cor_matrix)
p_matrix_transposed <- t(p_matrix)  # 同步转置p值矩阵

# 设置颜色映射
col_fun <- colorRamp2(c(-1, 0, 1), c("#9370DB", "white", "#FF7F24"))

# 定义显著性标记函数
add_sig_symbol <- function(j, i, x, y, width, height, fill) {
  p_value <- p_matrix_transposed[i, j]
if (p_value < 0.01) {
    grid.text("**", x, y, gp = gpar(fontsize = 10, col = "black"))  # 黑色双星号
  } elseif (p_value < 0.05) {
    grid.text("*", x, y, gp = gpar(fontsize = 10, col = "black"))   # 黑色单星号
  }
}

# 绘制热图
ht <- Heatmap(
  matrix = cor_matrix_transposed,
  name = "correlation",
  col = col_fun,
  rect_gp = gpar(col = "white", lwd = 1.2),
  width = unit(ncol(cor_matrix_transposed) * 0.7, "cm"),
  height = unit(nrow(cor_matrix_transposed) * 0.7, "cm"),
  row_dend_width = unit(1.5, "cm"),
  column_dend_height = unit(1.5, "cm"),
  row_names_gp = gpar(fontsize = 7),
  column_names_gp = gpar(fontsize = 7),
  heatmap_legend_param = list(
    title_gp = gpar(fontsize = 7),
    labels_gp = gpar(fontsize = 7)
  ),
  cell_fun = add_sig_symbol  # 添加显著性标记
)

# 显示图形
draw(ht)
pdf("Correlation_Heatmap.pdf", width = 10, height = 8)
draw(ht)
dev.off()