R语言ggplot封装函数报错：object not found问题求助

我需要编写一个绘制箱线图的函数，由于拥有多个结构相同的数据集，不想重复编写代码。非函数形式的代码可正常生成箱线图：

resp <- data.frame(sample = colnames(scores), response = metadata$Benefit)

scores %>% 
  t() %>% 
  as.data.frame() %>% 
  rownames_to_column("sample") %>% 
  right_join(resp, .) %>% 
  ggplot(aes(x = response, y = `CD4+ Tem`)) +
  geom_boxplot(aes(fill = response)) +
  scale_fill_manual(values = c('red','lightblue')) +
  geom_jitter(color="black", size=0.8, alpha=0.9) +
  ylim(0, 0.1)

但封装为函数后，无法识别y = celltype参数：

Boxes <- function(scores_data ,meta_data, column, celltype, k) {    ## gg boxplots
  
  resp = data.frame(sample = colnames(scores_data), response = meta_data[[column]])
  scores_data %>% t() %>% as.data.frame() %>% 
    rownames_to_column("sample") %>% 
    right_join(resp, .) %>% 
    ggplot(aes(x = response, y = celltype)) +
    geom_boxplot(aes(fill = response)) +
    scale_fill_manual(values = c('red','lightblue')) +
    geom_jitter(color="black", size=0.8, alpha=0.9) +
    ylim(0, k)
  
}

调用Boxes(scores, metadata, 'Benefit', 'Tgd cells' , 0.2 )时会报错：

Error in FUN(X[[i]], ...) : object 'Tgd cells' not found

样本数据 #

scores样本：

structure(c(0.00175333630321122, 0.0160232697957133, 0.119533527060973, 
3.38920032450346e-20, 0.00577191441771172, 1.43215113814464e-19, 
0.0487323510727856, 0.0872735021313101, 0, 0.0066498486266948, 
0.000940199906083161, 6.29830164958833e-20, 8.70346113353383e-19, 
0.000143134161959237, 2.96379130529708e-18, 2.79878821634228e-19, 
0.0303662173906541, 0, 0.0105316765508914, 0.00150757125433767, 
0, 2.96649645737511e-21, 0.0477634940125643, 0.00470920090451904, 
5.42614855792624e-20, 4.57868960693047e-20, 4.10374452367781e-19, 
0, 0.017933442808207, 0, 0, 0.204175059886713, 0.104955515405957, 
0, 0.0637468109054434, 0.0134166264055911, 0.0290285592482964, 
0.0516962126912615, 0.00613804602680461, 0.0710305792589827, 
0.00778137172189777, 0.000125708674793805, 0.353609047272269, 
1.512434582877e-19, 3.37240641744559e-18, 2.21761495132341e-18, 
0.0107824997605205, 0, 0.0128134041327042, 2.41313695610163e-18, 
0.0730722604536895, 0.041888143774871, 1.48680732530708e-18, 
0.0371111549542145, 0, 0.0147282080390037, 0, 2.57491956764445e-21, 
0.0104516046565067, 0.0433162672733649, 8.15084122983224e-21, 
0.0380473033612007, 0.0145099994894181, 0.014147791345196, 2.71228259849911e-18, 
0.0155558910245464, 2.21377704157198e-18, 2.87358596630636e-19, 
0.0146348072134295, 0, 0.00139414029448817, 0, 0.0914092782821773, 
0.00870970590818875, 0, 0, 3.67839239412613e-19, 0, 0.0273601871652021, 
7.93119212240358e-20, 0, 0.0296730264184565, 0, 0.0215283348860577, 
4.08767873819977e-18, 0.00247702696715108, 2.81027335370524e-18, 
0, 0.0196274656724322, 0, 0.00181974101325591, 0.0209131793129819, 
0.135232905818028, 0.0182668507516641, 2.96556952666363e-18, 
4.90745252242977e-21, 0.0608518971538323, 0, 0, 0), dim = c(10L, 
10L), dimnames = list(c("Adipocytes", "B-cells", "Basophils", 
"CD4+ memory T-cells", "CD4+ naive T-cells", "CD4+ T-cells", 
"CD4+ Tcm", "CD4+ Tem", "CD8+ naive T-cells", "CD8+ T-cells"), 
    c("Pat02", "Pat03", "Pat04", "Pat06", "Pat08", "Pat118", 
    "Pat119", "Pat123", "Pat126", "Pat14")))

metadata样本：

structure(list(Benefit = c("NoResponse", "NoResponse", "Response", 
"NoResponse", "NoResponse", "NoResponse", "NoResponse", "NoResponse", 
"Response", "NoResponse"), age_start = c(42, 61, 71, 33, 73, 
43, 61, 50, 77, 48), gct = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1), RECIST = c("SD", 
"PD", "PR", "PD", "PD", "PD", "PD", "SD", "PR", "PD"), overall_survival = c(1632, 
100, 987, 161, 140, 313, 808, 853, 641, 163), progression_free = c(538, 
76, 646, 77, 73, 120, 108, 544, 187, 84), primary = c("occult", 
"skin", "skin", "skin", "mucosal", "skin", "skin", "occult", 
"skin", "skin"), group = c("response", "nonresponse", "response", 
"nonresponse", "nonresponse", "nonresponse", "long-survival", 
"response", "response", "nonresponse"), histology = c("Unknown or NA", 
"Acrolentigenous", "Nodular", "Superficial spreading", "Nodular", 
"Unknown or NA", "Unknown or NA", "Unknown or NA", "Nodular", 
"Acrolentigenous"), stage = c("Stage 4", "Stage 4", "Stage 4", 
"Stage 4", "Stage 4", "Stage 4", "Stage 3", "Stage 4", "Stage 4", 
"Stage 4")), row.names = c("Pat02", "Pat03", "Pat04", "Pat06", 
"Pat08", "Pat118", "Pat119", "Pat123", "Pat126", "Pat14"), class = "data.frame")

ggplot的aes()使用非标准求值，直接写y = celltype时，它会尝试在数据框中寻找名为celltype的列，而不是将celltype变量中的字符串（比如'Tgd cells'）当作列名，这就导致了找不到对应对象的错误。

方法1：使用.data代词（推荐） #

利用tidyverse提供的.data代词，通过字符串引用数据框中的列：

Boxes <- function(scores_data ,meta_data, column, celltype, k) {    ## gg boxplots
  
  resp = data.frame(sample = colnames(scores_data), response = meta_data[[column]])
  scores_data %>% t() %>% as.data.frame() %>% 
    rownames_to_column("sample") %>% 
    right_join(resp, .) %>% 
    ggplot(aes(x = response, y = .data[[celltype]])) +  # 修改此处
    geom_boxplot(aes(fill = response)) +
    scale_fill_manual(values = c('red','lightblue')) +
    geom_jitter(color="black", size=0.8, alpha=0.9) +
    ylim(0, k)
  
}

方法2：使用符号求值 #

将字符串转换为符号，再用!!强制求值：

Boxes <- function(scores_data ,meta_data, column, celltype, k) {    ## gg boxplots
  
  resp = data.frame(sample = colnames(scores_data), response = meta_data[[column]])
  scores_data %>% t() %>% as.data.frame() %>% 
    rownames_to_column("sample") %>% 
    right_join(resp, .) %>% 
    ggplot(aes(x = response, y = !!sym(celltype))) +  # 修改此处
    geom_boxplot(aes(fill = response)) +
    scale_fill_manual(values = c('red','lightblue')) +
    geom_jitter(color="black", size=0.8, alpha=0.9) +
    ylim(0, k)
  
}

测试调用 #

注意样本数据中没有'Tgd cells'列，可用现有列测试：

Boxes(scores, metadata, 'Benefit', 'CD4+ Tem' , 0.1 )

内容的提问来源于stack exchange，提问作者Programming Noob