问题
I am trying to subtract group means from each group's values.
For example:
> x <- data.frame('gene' = c('A','A','A','B','B','C','C','C'),'value' = c(32.3,31,30.5,25,22.1,20.5,21.2,19.8))
> x
gene value
1 A 32.3
2 A 31.0
3 A 30.5
4 B 25.0
5 B 22.1
6 C 20.5
7 C 21.2
8 C 19.8
I can find the group means:
> aggregate(x[,2],list(x$gene),mean)
Group.1 x
1 A 31.26667
2 B 23.55000
3 C 20.50000
How do I subtract the "value" in x by the corresponding group mean? My desire result is as follow:
gene value-group.mean
1 A 1.03333
2 A -0.26667
3 A -0.76667
4 B 1.45
5 B -1.45
6 C 0
7 C 0.7
8 C -0.7
How can I do it in R?
Thanks.
回答1:
Using the library dplyr
, you can do:
library(dplyr)
x %>%
group_by(gene) %>%
mutate_all(funs(.-mean(.)))
# A tibble: 8 x 2
# Groups: gene [3]
gene value
<fct> <dbl>
1 A 1.03
2 A -0.267
3 A -0.767
4 B 1.45
5 B -1.45
6 C 0
7 C 0.700
8 C -0.700
回答2:
A data.table
method:
require(data.table)
setDT(x)[, `value-group.mean` := mean(value), by = gene][, `value-group.mean` := value - `value-group.mean`]
x
# gene value value-group.mean
# 1: A 32.3 1.0333333
# 2: A 31.0 -0.2666667
# 3: A 30.5 -0.7666667
# 4: B 25.0 1.4500000
# 5: B 22.1 -1.4500000
# 6: C 20.5 0.0000000
# 7: C 21.2 0.7000000
# 8: C 19.8 -0.7000000
回答3:
You can use ave()
to achieve it when with base R
, and in that case you don't need to apply aggregate
intermediately:
x$value_group.mean <- with(x,value-ave(value,gene))
such that
> x
gene value value_group.mean
1 A 32.3 1.0333333
2 A 31.0 -0.2666667
3 A 30.5 -0.7666667
4 B 25.0 1.4500000
5 B 22.1 -1.4500000
6 C 20.5 0.0000000
7 C 21.2 0.7000000
8 C 19.8 -0.7000000
来源:https://stackoverflow.com/questions/58948236/subtracting-values-group-wise-by-the-average-of-each-group-in-r