Dynamic use of match function - r

Question

I would like to match two data frame based on a certain column. My data frames are attached below
df <- structure(list(Read = structure(1:3, .Label = c("CC", "CG", "GC"
), class = "factor"), index = c(6L, 7L, 10L)), .Names = c("Read",
"index"), row.names = c(NA, -3L), class = "data.frame")
df1 <- structure(list(Ref_base = structure(c(1L, 6L, 4L, 2L, 3L, 4L,
3L, 5L), .Label = c("AT", "CC", "CG", "GC", "GT", "TG"), class = "factor"),
index = c(4L, 15L, 10L, 6L, 7L, 10L, 7L, 12L)), .Names = c("Ref_base",
"index"), row.names = c(NA, -8L), class = "data.frame")
I use match to find the match between the two data frames
match(df$index,df1$index)
and it gives me the correct result 4 5 3 as the index of matches. But i would like to lock down position 4 which is the index of first match and perform the match after 4 or whatever the first index is. I don't want to perform the search beyond the index of first match. For example i am interested to return the indexes as 4,5,6 including repetition if any.

Answer 1

The first solution is basically not more than a loop. It loops through all search elements from df$index and returns the match indices in tmp. The variable search_start is used to let the next search begin from the most recent position. Since search_start was defined outside of the anonymous function in sapply you have to use <<- instead of = or <- to access it. There is also some code for handling NAs (this was missing in the first version of my answer).
match_sapply=function(a,b) {
search_start=1
tmp2=sapply(a,function(x) {
tmp=match(x,b[search_start:nrow(df1)])
search_start<<-search_start+ifelse(is.na(tmp),0,tmp)
tmp
})
#the following line updates all non-NA elements of tmp2 with its cumulative sum
`[<-`(tmp2,!is.na(tmp2),cumsum(tmp2[!is.na(tmp2)]))
}
match_sapply(c(50,df$index,20),df1$index)
#[1] NA 4 5 6 NA
And another version using Recall. This is a recursive approach. Recall calls the function from which it was called (in our case match_recall) again. But you can provide different arguments. The arguments of match_recall are: x the search terms, y target vector, n recursion level (also selects specific element of x), si start index (same as start_index in previous solution). Again, there is some code that handles NAs.
match_recall=function(x,y,n=1,si=1) {
tmp=match(x[n],y[si:length(y)])
tmp1=tmp
if (is.na(tmp1)) tmp1=0
if (length(x)==n) {
return(tmp)
} else {
c(tmp,tmp1+Recall(x,y,n+1,si+tmp1))
}
}
match_recall(c(50,df$index,20),df1$index)
#[1] NA 4 5 6 NA