如何使用R有条件地遍历数据帧

Question

由于我是新循环R，我将感谢您的帮助，我的问题。假设我有这样一个数据框架：

Family <- c('mir-1','mir-1','mir-3','mir-4','mir-4','LET-7', 'LET-7','mir-1','mir-4','LET-7')
Species <- c('hsa','chicken','hsa','hsa','chicken','hsa','hsa','chicken','chicken','hsa')
Tissue <- c('blood','liver','blood','blood','liver','skin','skin','skin','liver','nail')
star <- c('1','4','3','4','12','3','7','4','1','5') #numeric
mature <- c('9','6','8','1','7','3','4','2','8','9')  #numeric
df <- data.frame(Family,Species,Tissue,star,mature)

我的输出应该是这样的：

Family_ <- c('mir-1','mir-1','mir-3','mir-4','mir-4','LET-7', 'LET-7','mir-1','mir-4','LET-7')
Species_ <- c('hsa','chicken','hsa','hsa','chicken','hsa','hsa','chicken','chicken','hsa')
Tissue_ <- c('blood','liver','blood','blood','liver','skin','skin','skin','liver','nail')
star <- c('1','4','3','4','12','3','7','4','1','5') #numeric
mature <- c('9','6','8','1','7','3','4','2','8','9')  #numeric
total_count <- c('10','10','11','5','28','17','17','6','28','14')  #numeric
star_total <- c('1','4','3','4','13','10','10','4','13','5')  #numeric
mature_total <- c('9','6','8','1','15','7','7','2','15','9')  #numeric
df_new <- data.frame(Family_,Species_,Tissue_,star,mature,star_total,mature_total,total_count)

我想在each family in each tissue in each species上循环一下。因此，对于第一列中的每个家族，即特定组织和特定物种的(不删除重复行)，我想要计算total_count <- sum (mature) + sum (star)、star_total <- sum (star)、mature_total <- sum (mature) *，并添加一个额外的列*，名为rpm_mature，可以以这种方式计算rpm_mature <- mature_total/total_count*10^6 (这里的输出中不包括这个列)。因此，对于在相似物种的相似组织中有相似家族的行，对这些重复行的计算应该是相同的。也许我描述得不是很好，但如果你看一下输出，那就有意义了。谢谢

Answer 1

下面是一种tidyverse方法--如果它有帮助的话：

library(tidyverse)

df %>%
  mutate_at(c("star", "mature"), as.numeric) %>%
  group_by(Family, Species, Tissue) %>%
  mutate(total_count = sum(mature) + sum(star),
         star_total = sum(star),
         mature_total = sum(mature),
         rpm_mature = mature_total/total_count*10^6)

输出

# A tibble: 10 x 9
# Groups:   Family, Species, Tissue [8]
   Family Species Tissue  star mature total_count star_total mature_total rpm_mature
   <fct>  <fct>   <fct>  <dbl>  <dbl>       <dbl>      <dbl>        <dbl>      <dbl>
 1 mir-1  hsa     blood      1      8           9          1            8    888889.
 2 mir-1  chicken liver      4      5           9          4            5    555556.
 3 mir-3  hsa     blood      3      7          10          3            7    700000 
 4 mir-4  hsa     blood      4      1           5          4            1    200000 
 5 mir-4  chicken liver      2      6          16          3           13    812500 
 6 LET-7  hsa     skin       3      3          16          9            7    437500 
 7 LET-7  hsa     skin       6      4          16          9            7    437500 
 8 mir-1  chicken skin       4      2           6          4            2    333333.
 9 mir-4  chicken liver      1      7          16          3           13    812500 
10 LET-7  hsa     nail       5      8          13          5            8    615385.

编辑

如果您有兴趣开发一种循环的方法，您可以做以下工作以获得相同的结果：

df$star <- as.numeric(df$star)
df$mature <- as.numeric(df$mature)

df <- cbind(df, total_count = NA, star_total = NA, mature_total = NA)

for (Fam in df$Family) {
  for (Spec in df$Species) {
    for (Tiss in df$Tissue) {
      res <- df[df$Family == Fam & df$Species == Spec & df$Tissue == Tiss,]
      if (nrow(res) > 0) {
        res$total_count = sum(res$mature) + sum(res$star)
        res$star_total = sum(res$star)
        res$mature_total = sum(res$mature)
        df[df$Family == Fam & df$Species == Spec & df$Tissue == Tiss,] <- res
      }
    }
  }
}

df$rpm_mature = df$mature_total/df$total_count*10^6

Answer 2

下面是一种方法，我们通过Family, Species, Tissue进行计算：

library(data.table)
setDT(df)
df[,":="(total_count = sum(mature) + sum(star),
         star_total = sum(star),
         mature_total = sum(mature),
         rpm_mature = mature_total/total_count*10^6),.(Family, Species, Tissue)]

print(df)

    Family Species Tissue star mature total_count star_total mature_total rpm_mature
 1:  mir-1     hsa  blood    1      8           9          1            8   888888.9
 2:  mir-1 chicken  liver    4      5           9          4            5   555555.6
 3:  mir-3     hsa  blood    3      7          10          3            7   700000.0
 4:  mir-4     hsa  blood    4      1           5          4            1   200000.0
 5:  mir-4 chicken  liver    2      6           8          3           13  1625000.0
 6:  LET-7     hsa   skin    3      3           6          9            7  1166666.7
 7:  LET-7     hsa   skin    6      4          10          9            7   700000.0
 8:  mir-1 chicken   skin    4      2           6          4            2   333333.3
 9:  mir-4 chicken  liver    1      7           8          3           13  1625000.0
10:  LET-7     hsa   nail    5      8          13          5            8   615384.6