Partial or complete overlap of two genomic ranges

Question

[Jimbou]

Hello,

I need your help. I want to compare two dataframes with genomic features like this:

More precisely, frame 2 compared to frame 1

chr start end name        chr start end   name  
1    2    5   A           1    1    3       AA
2   10    15  B           2    9    16      BB
3   27    30  C           3    28   29      CC

As a result I want this:

1. Completely overlapping:

   chr start end name    chr start end name
   2   10    15   B       2    9    16    BB
   

2. Partial overlap:

   chr start end  name    chr start end   name 
   1    2    5    A        1    1    3    AA
   

3. and within the range:

   chr start end name        chr start end   name 
   3   27    30  C            3    28   29    CC
   

Are the R packages IRanges and GenomicRanges suitable for such analysis? Or do I have to write some > < commands?

Answer 1

[komal.rathi]

Using the GenomicRanges library in R:

library(GenomicRanges)
x1 = read.table(text="chr start end name  
1    2    5   A
2   10    15  B
3   27    30  C",header=T)

x2 = read.table(text="chr start end   name  
1   1    1    3       AA
2   2    9    16      BB
3   3    28   29      CC",header=T)

# Make GRanges object
gr1 = with(x1, GRanges(chr, IRanges(start = start, end = end, names = name)))
gr2 = with(x2, GRanges(chr, IRanges(start = start, end = end, names = name)))

# Completely overlapping
type1 = findOverlaps(query = gr1, subject = gr2, type = "within")
type1.df = data.frame(x1[queryHits(type1),], x2[subjectHits(type1),])
type1.df
  chr start end name chr.1 start.1 end.1 name.1
2   2    10  15    B     2       9    16     BB

# Within range
type3 = findOverlaps(query = gr2, subject = gr1, type = "within")
type3.df = data.frame(x1[subjectHits(type3),], x2[queryHits(type3),])
type3.df
  chr start end name chr.1 start.1 end.1 name.1
3   3    27  30    C     3      28    29     CC

# Partial Overlaps only (no complete overlaps or within range overlaps)
type2 = findOverlaps(query = gr1, subject = gr2, type = 'any')
type2.df = data.frame(x1[queryHits(type2),], x2[subjectHits(type2),])
x = rbind(type1.df, type2.df, type3.df)
type2.df = x[!(duplicated(x) | duplicated(x, fromLast = TRUE)), ]
type2.df
   chr start end name chr.1 start.1 end.1 name.1
21   1     2   5    A     1       1     3     AA

You will get three different data frames as per your question. Based on your example, I am assuming type 1 and type 3 are opposites.

[Jimbou]

Extremely useful add-on

http://faculty.ucr.edu/~tgirke/Documents/R_BioCond/My_R_Scripts/rangeoverlapper.R

[Paul]

How do I find out intergenic locations?

[Jimbou]

check the valr package

[bjwiley23]

Not sure if this is new in IRanges but I tested using type='any' and it also gave me ranges from the query that were completely within the subject.

Answer 2

[Alex Reynolds]

Take a look at BEDOPS bedmap --fraction-map, which allows you to recover map elements (those on your right-hand side) which overlap reference elements (those on your left-hand side) by some fractional value between 0 and 1, inclusive (i.e. between 0 and 100%).

For instance, if your reference and map data sets are sorted BED files called A and B, respectively, then you could do:

$ bedmap --echo --echo-map --fraction-map 1 A B

to get all elements from set B that completely overlap an element in set A.

These tools can run from the command line or within R, via system() calls.

Answer 3

[EagleEye]

You could also try GenGen: http://www.openbioinformatics.org/gengen/tutorial_scan_region.html