将两个 data.tables 乘以一个具有动态公差的数值变量

Question

我有两个 data.tables，我想通过一个数字变量（双精度）加入它们。然而，数值变量存在不确定性的缺陷。因此，我必须允许一个特定的公差，该公差因变量而异。

在下面的示例中，"mz" 是我想加入 DT1 和 DT2 的变量。根据变量 iso_mz 计算的公差：iso_mz * 5e-6。

DT1 <- data.table(mz = c(433.231512451172, 451.091953822545, 454.347605202415, 490.167234693255, 518.225894504123), 
Var1 = c(433.231018066406, 451.091430664062, 454.347015380859, 490.166381835938, 518.22509765625), 
Var2 = c(433.232147216797, 451.092559814453, 454.34814453125, 490.168273925781, 518.2265625))


DT2 <- data.table(iso_mz = c(451.0900, 490.1651, 518.2281, 433.2335), 
comp = c("m1", "m2", "m3", "m4"))

如果我不必使用公差，我会使用 data.table 包的 "on=.()" 功能。我尝试改编中的代码，但由于某种原因我无法进入运行,..

我的示例所需的输出是：

Output <- data.table(
iso_mz = c(433.2335, 451.0900, 490.1651, 518.2281), 
comp = c("m4", "m1", "m2", "m3"),
mz = c(433.231512451172, 451.091953822545, 490.167234693255, 518.225894504123), 
Var1 = c(433.231018066406, 451.091430664062, 490.166381835938, 518.22509765625), 
Var2 = c(433.232147216797, 451.092559814453, 490.168273925781, 518.2265625))

提前致谢！

Answer 1

这是一种使用 data.table 中的 foverlaps() 的方法。

tolerance = 5e-6
#create ranges to join on
DT1[, `:=`(min = mz - mz * tolerance, 
           max = mz + mz * tolerance) ]
DT2[, `:=`(min = iso_mz - iso_mz * tolerance, 
           max = iso_mz + iso_mz * tolerance) ]
#set keys
setkey(DT1, min, max )
setkey(DT2, min, max )
#perform overlap join, order, remove min-max columns
ans <- setorder( foverlaps( DT2, DT1 ), mz)[, `:=`(min=NULL,max=NULL,i.min=NULL,i.max=NULL)][]


# mz     Var1     Var2   iso_mz comp
# 1: 433.2315 433.2310 433.2321 433.2335   m4
# 2: 451.0920 451.0914 451.0926 451.0900   m1
# 3: 490.1672 490.1664 490.1683 490.1651   m2
# 4: 518.2259 518.2251 518.2266 518.2281   m3


#check
all.equal( setcolorder(ans, names(Output)), Output )
[1] TRUE

将两个 data.tables 乘以一个具有动态公差的数值变量

Join two data.tables by one numeric variable with dynamic tolerance

r

dataframe

merge

data.table

jointable