如何在多个字符串未列出的列表中找到匹配值？

Question

我有两个数据集：

#df1:
Gene   interactors
ACE      BRCA, HER2
NOS       NA, NA
P53       NA
CDON      TGBP

df2:
Gene   interactors
AGT      NOS, HER2
NPKB     CDON
P70      GPC
IK       TGBP

我希望在df1中找出在df2下被列为相互作用的基因，并在df1中通过与df2中的相互作用物相匹配的相互作用来识别基因。

输出：

Gene   interactors     matched_gene_interactor   matched_interactor_interactor
ACE      BRCA, HER2         FALSE                        TRUE                      
NOS       NA, NA            TRUE                         FALSE
P53       NA                FALSE                        FALSE
CDON      TGBP              TRUE                         TRUE

#ACE has an interactor (HER2) in both df1 and df2
#NOS matches itself as an interactor in df2
#CDON matches itself as an interactor in df2 and as having an interactor (TGBP) in both df1 and df2

我已经能够通过以下方式编写代码以获取matched_gene_interactor列：

df1$matched_gene_interactor <-  df1$Gene %in% unlist(strsplit(df2$interactors, ", "))

但是，我被困在第二个matched_interactor_interactor列上了

我尝试了一些东西，但没有找到如何将它发展到我想要的第二列的地步，例如：

df1interactors <-  unlist(strsplit(df1$interactors, ", "))
df2interactors <-  unlist(strsplit(df2$interactors, ", "))
matched_interactor_interactor <-  df1interactors %in% df2interactors

如何与两个具有未列出字符串拆分的数据集进行匹配？我有生物学背景，所以不知道从哪里开始。

示例输入数据：

df1:

structure(list(Gene = c("ACE", "NOS", "P53", "CDON"), interactors = c("BRCA, HER2", 
"NA, NA", NA, "TGBP")), row.names = c(NA, -4L), class = c("data.table", 
"data.frame"))

df2:

structure(list(Gene = c("AGT", "NPKB", "P70", "IK"), interactors = c("NOS, HER2", 
"CDON", "GPC", "TGBP")), row.names = c(NA, -4L), class = c("data.table", 
"data.frame"))

Answer 1

您可以在逗号上拆分interactors of df2，并对每一行检查是否存在来自df1的interactors值。

temp <- unlist(strsplit(df2$interactors, ', '))
df1$matched_interactor_interactor <- sapply(strsplit(df1$interactors, ', '), 
                                      function(x) any(x %in% temp))

df1
#   Gene interactors matched_gene_interactor matched_interactor_interactor
#1:  ACE  BRCA, HER2                   FALSE                          TRUE
#2:  NOS      NA, NA                    TRUE                         FALSE
#3:  P53        <NA>                   FALSE                         FALSE
#4: CDON        TGBP                    TRUE                          TRUE

如果df2$interactors不是很大，您也可以通过创建动态regex模式来做到这一点，而无需拆分df1$interactors：

grepl(paste0('\\b', temp, '\\b', collapse = '|'), df1$interactors)
#[1]  TRUE FALSE FALSE  TRUE