如何在ggplot2函数中正确传递混合参数：直接列名+循环列名

我编写了以下R代码用于生成单个或多个ggplot图表：

#list of columns names in my dataframe:
metals_to_analys <- c("51 V ug/g","52 Cr ug/g","75 As ug/g","27 Al ug/g","55 Mn ug/g","60 Ni ug/g","63 Cu ug/g","66 Zn ug/g","95 Mo ug/g","111 Cd ug/g","88 Sr ug/g","47 Ti ug/g")

#general dynamic list to design the function:
design <- list(
  geom_boxplot(aes(fill=treatment), na.rm=TRUE),
  geom_jitter(aes(shape=season), na.rm=TRUE),
  stat_n_text(size=3),
  scale_fill_manual(values=c(control="palegreen4",`external control`="palegreen3",`capture: external control`="palegreen",pollution="tomato3")),
  theme(plot.title = element_text(size = 20, face = "bold"), legend.position = "bottom"),
  stat_compare_means(method = "t.test", label = "p.format", size = 5, na.rm=TRUE)
  )

#the function to create a single ggplot and pass the inputs as variables
general_plot <- function(df,x.var,y.var,fill.var){
  x.var <- enquo(x.var)
  y.var <- enquo(y.var)
  fill.var <- enquo(fill.var)
  ggplot(data=df, aes(x=!!x.var, y=!!y.var, fill=!!fill.var))
}

#the "final" function to run the plot function inside a loop of col.names:
plot_generator <- function(df,x.var,fill.var){
  for (metal in metals_to_analys){
  print(general_plot(df=df,x.var=x.var, !!sym(metal), fill.var=fill.var)+design)
  }
}

直接运行以下循环调用general_plot函数可得到预期输出：

#my attempt to check the "general_plot" function:
for (metal in metals_to_analys){
  print(general_plot(df=subset_AC_with_breeding_col,x.var=treatment, !!sym(metal), fill.var=treatment)+design)
  }

但调用plot_generator(subset_AC_with_breeding_col,treatment,treatment)时出现如下错误：

Error in `geom_boxplot()`:
! Problem while computing aesthetics.
ℹ Error occurred in the 1st layer.
Caused by error:
! object 'treatment' not found
Backtrace:
  1. global plot_generator(subset_AC_with_breeding_col, treatment, treatment)
  3. ggplot2:::print.ggplot(...)
  5. ggplot2:::ggplot_build.ggplot(x)
  6. ggplot2:::by_layer(...)
 13. ggplot2 (local) f(l = layers[[i]], d = data[[i]])
 14. l$compute_aesthetics(d, plot)
 15. ggplot2 (local) compute_aesthetics(..., self = self)
 16. ggplot2:::scales_add_defaults(...)
 17. base::lapply(aesthetics[new_aesthetics], eval_tidy, data = data)
 18. rlang (local) FUN(X[[i]], ...)

我尝试了以下方法均未解决问题：

• 传入字符串列名：plot_generator(subset_AC_with_breeding_col,"treatment","treatment")，但X轴刻度丢失，轴标题变为“x.var”；
• 传入向量：plot_generator(subset_AC_with_breeding_col,subset_AC_with_breeding_col$treatment,subset_AC_with_breeding_col$treatment)，但X轴标题显示为“x.var”而非“treatment”；
• 将design列表中geom_boxplot(aes(fill=treatment))的“treatment”替换为“x.var”，但该列表在直接循环中可正常工作。

希望了解如何顺畅传递这类混合参数（直接列名+循环获取的列名），为何函数输出与直接循环输出存在差异。

错误根源 #

直接循环能运行，是因为design里的aes(fill=treatment)在全局环境中解析，此时treatment是数据框的列名，能被正确识别。但封装到plot_generator函数后，函数运行在局部环境，design里硬编码的treatment无法在函数环境中找到对应的对象，导致报错。同时，plot_generator中直接传递x.var和fill.var给general_plot时，enquo会捕获x.var这个符号本身，而非传入的treatment列名。

修改后的代码 #

1. 将design改为动态生成函数，让fill参数能接收传入的变量，避免硬编码列名；
2. 调整plot_generator函数，先捕获传入的变量引用，再传递给general_plot，确保变量能在函数环境中正确解析。

# 保持金属列名列表不变
metals_to_analys <- c("51 V ug/g","52 Cr ug/g","75 As ug/g","27 Al ug/g","55 Mn ug/g","60 Ni ug/g","63 Cu ug/g","66 Zn ug/g","95 Mo ug/g","111 Cd ug/g","88 Sr ug/g","47 Ti ug/g")

# 将design改为生成函数，接收fill变量参数
get_design <- function(fill_var) {
  fill_var <- enquo(fill_var)
  list(
    geom_boxplot(aes(fill=!!fill_var), na.rm=TRUE),
    geom_jitter(aes(shape=season), na.rm=TRUE),
    stat_n_text(size=3),
    scale_fill_manual(values=c(control="palegreen4",`external control`="palegreen3",`capture: external control`="palegreen",pollution="tomato3")),
    theme(plot.title = element_text(size = 20, face = "bold"), legend.position = "bottom"),
    stat_compare_means(method = "t.test", label = "p.format", size = 5, na.rm=TRUE)
  )
}

# general_plot函数保持不变
general_plot <- function(df,x.var,y.var,fill.var){
  x.var <- enquo(x.var)
  y.var <- enquo(y.var)
  fill.var <- enquo(fill.var)
  ggplot(data=df, aes(x=!!x.var, y=!!y.var, fill=!!fill.var))
}

# 修改后的plot_generator函数
plot_generator <- function(df,x.var,fill.var){
  # 捕获传入的变量引用
  x_var <- enquo(x.var)
  fill_var <- enquo(fill.var)
  
  for (metal in metals_to_analys){
    # 动态生成design，传入fill变量
    design <- get_design(!!fill_var)
    # 传递捕获后的变量给general_plot
    p <- general_plot(df=df, x.var=!!x_var, y.var=!!sym(metal), fill.var=!!fill_var) + design
    # 添加标题显示当前金属名称
    p <- p + ggtitle(metal)
    print(p)
  }
}

调用方式 #

现在直接运行以下代码即可正常生成所有图表：

plot_generator(subset_AC_with_breeding_col, treatment, treatment)

差异原因解释 #

• 直接循环时：所有代码在全局环境运行，treatment作为数据框列名能被aes正确解析，!!sym(metal)也能正确映射到对应的金属列；
• 函数封装后：函数有独立的局部环境，硬编码的treatment无法在局部环境中找到，必须通过enquo和!!（解引用）将变量从全局环境传递到函数局部环境，同时动态生成design确保美学映射能正确识别变量。

内容的提问来源于stack exchange，提问作者Maya Eldar