Hi,

I did parallel searching overlapped regions from two GRanges objects, and I did Vectorization over them, so I got CompressedIntegerList objects (a.k.a, non-overlapped regions indicates integer(0) in the list).

Now, I need to sum up two vectors (a.k.a, sum of vector indicates overall overlapped regions number for each). However, I did keep only one (the one with lowest pvalue if multiple intersection found) and I have new CompressedIntegerList objects, but it has NAs index now ( NAs values indicates non-overlapped regions).

I did sum over two vector, but R consider length of regions with NAs index as 1, while R also consider length of region with interger(0) as 0. So, I got wrong results if I used regions with NAs index for vector sum(though I kept only one regiosn with lowest pvalue, but it has NAs index, this gives wrong results for vector sum) .

I tried every possible solution from statckoverflow and I could not find solution yet. How can I fix this issue? Thank you

For the sake of clarity of problem, I have this reproducible example:

# set up

a <- GRanges(
  seqnames=Rle(c("chr1", "chr2", "chr3", "chr4"), c(3, 2, 1, 2)),
  ranges=IRanges(seq(1, by=9, len=8), seq(7, by=9, len=8)),
  rangeName=letters[seq(1:8)], score=sample(1:20, 8, replace = FALSE))
b <- GRanges(
  seqnames=Rle(c("chr1", "chr2", "chr3","chr4"), c(4, 1, 2, 1)),
  ranges=IRanges(seq(2, by=6, len=8), seq(4, by=6, len=8)),
  rangeName=letters[seq(1:8)], score=sample(1:20, 8, replace = FALSE))
c <- GRanges(
  seqnames=Rle(c("chr1", "chr2", "chr3","chr4"), c(2, 3, 1,2)),
  ranges=IRanges(seq(6, by=7, len=8), seq(8, by=7, len=8)),
  rangeName=letters[seq(1:8)], score=sample(1:20, 8, replace = FALSE))

# pre-step : pvalue conversion for a, b,c

a$pvalue <- 10^(score(a)/-1)
a$score <- NULL
b$pvalue <- 10^(score(b)/-1)
b$score <- NULL
c$pvalue <- 10^(score(c)/-1)
c$score <- NULL

​#find overlap 

ov_ab <- as(findOverlaps(a, b), "List")
ov_ac <- as(findOverlaps(a, c), "List")

#keep only one regions with lowest pvalue from multiple intersection

b_idx <- as(which.min(extractList(b$pvalue, ov_ab)), "List")
c_idx <- as(which.min(extractList(c$pvalue, ov_ac)), "List")

# count total overlapped number : (all NA must be replaced with integer(0) to proceed vector sum)

b_idx <- c(1,2,1,1,NA, NA, NA, NA)
c_idx <- c(1,1, NA, 2, NA, NA, 1, NA)

# my desired output (I did manually, because I tried to replace NA with integer(0) but I always get error) : 

a.len <- as(c(1,1,1,1,1,1,1,1), "List")     
b_one.len <- as(c(1,1,1,1, 0, 0, 0, 0), "List")           
c_one.len <- as(c(1,1,0, 1, 0, 0,1, 0), "List")             
totNumOvlap <-  a.len + b_one.len + c_one.len

# this is my desire output  (vector sum is very crucial for accuracy of overall result) : 

totNumOvlp <- c(3,3,2,3,1,1,2,1)

Objective: I must replace NAs values into integer(0) in compressedIntegerList objects, then I can able to do vector sum over them. 

How can I address this problem ? Is there any alternative way to solve the issue ? Thanks a lot

Best regards:

Jurat