Question

我遇到了数据帧问题。比方说，我有一个数据框，其中一列包含值（范围0到100000）。一个例子：

                     TCGA.CG.4462
 ENSG00000000003       4.7574661
 ENSG00000000005       0.0000000
 ENSG00000000419       24.1066335
 ENSG00000000457       2.7631012
 ENSG00000000460       0.8928772

我想通过以下5个类别添加一个包含列数据概率的新列：

non_expresed [0]
low_expressed] 0,1]
normal_expressed] 1,10]
high_expressed [10,100]
very_high_expressed＆gt; 100

因此，对于该示例，我想要在新列中添加的值是：

non_expresed：0.2
low_expressed 0.2
normal_expressed：0.4
high_expressed：0.2
very_high_expressed：0.0

所以我的数据框就像这样：

                     TCGA.CG.4462     Prob
 ENSG00000000003       4.7574661      0.4
 ENSG00000000005       0.0000000      0.2
 ENSG00000000419       24.1066335     0.2
 ENSG00000000457       2.7631012      0.4
 ENSG00000000460       0.8928772      0.2

我已经尝试了很多不同的方法，但到目前为止还没有。我认为if条件是解决我的问题的最佳方法，但if条件给出错误，因为条件的长度是> 1。谁能告诉我最新的解决方法呢？

Answer 1

我们可以使用cut来查找间隔并使用所需的概率标记它们。由于概率中存在重复，因此会出现警告消息，可以忽略。请参阅下面的演示：

library(data.table)
cut(df1$TCGA.CG.4462, breaks = c(-Inf, 0, 1, 10, 100, Inf), include.lowest = TRUE)
# [1] (1,10]   [-Inf,0] (10,100] (1,10]   (0,1]   
# Levels: [-Inf,0] (0,1] (1,10] (10,100] (100, Inf]

df1[, prob := as.numeric(as.character(cut(TCGA.CG.4462, 
                                          breaks = c(-Inf, 0, 1, 10, 100, Inf), 
                                          include.lowest = TRUE,
                                          labels = c(0.2, 0.2, 0.4, 0.2, 0.0))))]

# Warning message:
#   In `levels<-`(`*tmp*`, value = if (nl == nL) as.character(labels) 
#                 else paste0(labels,  : duplicated levels in factors are deprecated

df1
#              genes TCGA.CG.4462 prob
# 1: ENSG00000000003    4.7574661  0.4
# 2: ENSG00000000005    0.0000000  0.2
# 3: ENSG00000000419   24.1066335  0.2
# 4: ENSG00000000457    2.7631012  0.4
# 5: ENSG00000000460    0.8928772  0.2

使用基础R（未使用包）

df1 <- within(df1, prob <- as.numeric(as.character(cut(TCGA.CG.4462, 
                                                       breaks = c(-Inf, 0, 1, 10, 100, Inf), 
                                                       include.lowest = TRUE,
                                                       labels = c(0.2, 0.2, 0.4, 0.2, 0.0)))))

数据：

library(data.table) df1 <- fread('ENSG00000000003 4.7574661 ENSG00000000005 0.0000000 ENSG00000000419 24.1066335 ENSG00000000457 2.7631012 ENSG00000000460 0.8928772', header = F) colnames(df1) <- c("genes", "TCGA.CG.4462")

编辑：第三栏：将值1添加到“第三列”

data.table包

df1[, `:=` ( prob = as.numeric(as.character(cut(TCGA.CG.4462, breaks = c(-Inf, 0, 1, 10, 100, Inf), include.lowest = TRUE, labels = c(0.2, 0.2, 0.4, 0.2, 0.0)))), third = 1)]

基础R：

within(df1, c(prob <- as.numeric(as.character(cut(TCGA.CG.4462, breaks = c(-Inf, 0, 1, 10, 100, Inf), include.lowest = TRUE, labels = c(0.2, 0.2, 0.4, 0.2, 0.0)))), third <- 1))

Answer 2

以下是另一个data.table解决方案，该解决方案在非等连接中使用查找表和更新：

library(data.table)
# create lookup table
lookup <- data.table(
  expression = c("non", "low", "normal", "high", "very_high"),
  Prob = c(0.2, 0.2, 0.4, 0.2, 0.0),
  lower = c(-Inf, 0, 10^(0:2))
)
lookup[, upper := shift(lower, type = "lead", fill = Inf)][]

   expression Prob lower upper
1:        non  0.2  -Inf     0
2:        low  0.2     0     1
3:     normal  0.4     1    10
4:       high  0.2    10   100
5:  very_high  0.0   100   Inf

# update in a non-equi join
# note the left open intervals
setDT(DT)[lookup, on = .(TCGA.CG.4462 > lower, TCGA.CG.4462 <= upper), 
   `:=`(expression = expression, Prob = Prob)][]

            row.id TCGA.CG.4462 expression Prob
1: ENSG00000000003    4.7574661     normal  0.4
2: ENSG00000000005    0.0000000        non  0.2
3: ENSG00000000419   24.1066335       high  0.2
4: ENSG00000000457    2.7631012     normal  0.4
5: ENSG00000000460    0.8928772        low  0.2

数据

library(data.table)
DT <- fread(
  "row.id                TCGA.CG.4462
 ENSG00000000003       4.7574661
  ENSG00000000005       0.0000000
  ENSG00000000419       24.1066335
  ENSG00000000457       2.7631012
  ENSG00000000460       0.8928772"
)

如何按类别将数据框colunm的概率值添加到同一数据框中的新列？

2 个答案:

数据