如何按组在R中的数据帧中执行KS.test

我想在R中对下面的示例数据帧“数据”执行两个示例Kolmogorov-Smirnov（KS）测试：

要在两个单独的列之间执行KS测试，代码如下：

> ks.test(data$Protein1, data$Protein2, data=data)

    Two-sample Kolmogorov-Smirnov test

data:  data$Protein1 and data$Protein2
D = 0.42308, p-value = 0.01905
alternative hypothesis: two-sided

Warning message:
In ks.test(data$Protein1, data$Protein2, data = data) :
  cannot compute exact p-value with ties

>ks.test（数据$Protein1，数据$Protein2，数据=数据）
两样本Kolmogorov-Smirnov检验
数据：数据$Protein1和数据$Protein2
D=0.42308，p值=0.01905
替代假设：双边
警告信息：
在ks.测试中（数据$Protein1，数据$Protein2，数据=数据）：
无法使用关系计算精确的p值

但是，我希望每个列和每个组都这样做。例如，对t.test或wilcox.test执行此操作很容易，因为您可以将其编码为 t、 test（y1，y2）或t.test（y~x）#其中y是数字，x是二进制因子 但是当涉及到二进制因子时，没有用于ks.test的代码。 有人能帮忙吗

最后，我想对所有蛋白质的整个数据帧进行此操作，正如我可以成功地对t测试所做的那样，但我想对ks.test进行如下操作：

t_tests <- lapply(
  data[, -1], # apply the function to every variable *other than* the first one (group)
  function(x) { t.test(x ~ HealthGroups, data = data) }
)

---

t_测试这里有一个非常简单的方法。这使用了一个循环，在R圈中通常是不赞成的。然而，它非常简单且不言自明，这对新用户来说是一个加号，在这种情况下，循环速度太慢没有问题。（请注意，如果愿意，您可以使用
lappy（）
，但这仍然是一个循环，只是外部看起来不同。）

只需使用相同的变量创建两个新的子集数据帧。然后在数据帧上循环调用
ks.test
。输出对用户不是很友好，它只会说
j
——因此我添加了一个调用来打印被测试变量的名称

# I am assuming the original data frame is called d
dc <- d[d$Group=="Control",]
dp <- d[d$Group=="Patient",]
for(j in 1:8){  
  writeLines(names(dc)[j])
  print(ks.test(dc[,j], dp[,j]))  
}
# Protein1
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.30769, p-value = 0.5882
# alternative hypothesis: two-sided
# 
# Protein2
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.23077, p-value = 0.8793
# alternative hypothesis: two-sided
# 
# Protein3
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.23077, p-value = 0.8793
# alternative hypothesis: two-sided
# 
# Protein4
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.46154, p-value = 0.1254
# alternative hypothesis: two-sided
# 
# Protein5
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.23077, p-value = 0.8793
# alternative hypothesis: two-sided
# 
# Protein6
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.15385, p-value = 0.9992
# alternative hypothesis: two-sided
# 
# Protein7
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.38462, p-value = 0.2999
# alternative hypothesis: two-sided
# 
# Protein8
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.46154, p-value = 0.1265
# alternative hypothesis: two-sided
# 
# Warning messages:
# 1: In ks.test(dc[, j], dp[, j]) : cannot compute exact p-value with ties
# 2: In ks.test(dc[, j], dp[, j]) : cannot compute exact p-value with ties
# 3: In ks.test(dc[, j], dp[, j]) : cannot compute exact p-value with ties
# 4: In ks.test(dc[, j], dp[, j]) : cannot compute exact p-value with ties

#我假设原始数据帧被称为d
非常感谢你。这真的很有帮助。我只是想问你能不能帮我把它转换成矩阵。使用我上面提供的t-test代码，它生成了一个列表，我可以取消列表并提供colname和rowname。在本例中，结果以注释行的形式出现，而不是列表。那么是否可以让我的最终结果在行中显示为：Protein 1、2、3等，并且colname应该包含相应的列：D和p-value。@Letin，请查看文档：在值部分，它告诉您函数输出的内容。如果您将输出分配给一个变量，而不是打印它，您可以提取统计信息（D）和p值，并将它们分配给列表、矩阵或数据框等（您可能最喜欢后者，因为您可以有一个带有蛋白质名称的变量。谢谢@gung提供的此信息。我将尝试此方法。
# I am assuming the original data frame is called d
dc <- d[d$Group=="Control",]
dp <- d[d$Group=="Patient",]
for(j in 1:8){  
  writeLines(names(dc)[j])
  print(ks.test(dc[,j], dp[,j]))  
}
# Protein1
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.30769, p-value = 0.5882
# alternative hypothesis: two-sided
# 
# Protein2
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.23077, p-value = 0.8793
# alternative hypothesis: two-sided
# 
# Protein3
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.23077, p-value = 0.8793
# alternative hypothesis: two-sided
# 
# Protein4
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.46154, p-value = 0.1254
# alternative hypothesis: two-sided
# 
# Protein5
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.23077, p-value = 0.8793
# alternative hypothesis: two-sided
# 
# Protein6
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.15385, p-value = 0.9992
# alternative hypothesis: two-sided
# 
# Protein7
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.38462, p-value = 0.2999
# alternative hypothesis: two-sided
# 
# Protein8
# 
#   Two-sample Kolmogorov-Smirnov test
# 
# data:  dc[, j] and dp[, j]
# D = 0.46154, p-value = 0.1265
# alternative hypothesis: two-sided
# 
# Warning messages:
# 1: In ks.test(dc[, j], dp[, j]) : cannot compute exact p-value with ties
# 2: In ks.test(dc[, j], dp[, j]) : cannot compute exact p-value with ties
# 3: In ks.test(dc[, j], dp[, j]) : cannot compute exact p-value with ties
# 4: In ks.test(dc[, j], dp[, j]) : cannot compute exact p-value with ties

dk <- data.frame(Protein=character(8), D=numeric(8), p=numeric(8), stringsAsFactors=F)
for(j in 1:8){  
  k <- ks.test(dc[,j], dp[,j])
  dk$Protein[j] <- names(dc)[j]
  dk$D[j]       <- k$statistic
  dk$p[j]       <- k$p.value
}
dk
#    Protein         D         p
# 1 Protein1 0.3076923 0.5881961
# 2 Protein2 0.2307692 0.8793244
# 3 Protein3 0.2307692 0.8793244
# 4 Protein4 0.4615385 0.1253895
# 5 Protein5 0.2307692 0.8793244
# 6 Protein6 0.1538462 0.9992124
# 7 Protein7 0.3846154 0.2999202
# 8 Protein8 0.4615385 0.1264877