Bedtools merge -d option help

Question

[jantu008]

Hi, I'm using Bedtools for R (bedr) in Ubuntu and I am trying to merge genomic regions that are within 100 bp of one another. For example, I have a data frame (df) with 3274 obs of 3 variables and the first rows look like:

df

chr	start	end
chr1	1214882	1214884
chr1	1214942	1214944
chr1	1215030	1215032
chr1	1215036	1215038

and when I merge using:

> df1 <- bedr.merge.region(df.sorted, distance = 100, number = TRUE, check.zero.based = TRUE, check.chr = TRUE, check.valid = TRUE, check.sort = TRUE)

I get the data frame (df1):

df1

chr	start	end	V4
chr1	1214882	1215038	4

My goal is to merge all coordinates that are within 100 bp (distance=100) so from data frame df it should've merged the 2 first rows together and then the 2 last together since there's less than 100 bp between the start of row 1 (on df) and end of row 2 (on df), not the 4 (as if shows in df1), since that gives a distance of 156 bp (1215038 - 1214882 = 156),

Any help as to why the parameter "distance = 100" is not merging only regions within 100 bp and it merges regions at 156 bp? The goal is to be able to design probes for wet lab to capture regions of interest but our probes are limited to 100 bp so I want to see how many probes of 100 bp I would need to build to capture all regions and what would their coordinates be.

Thank you Joana

To clarify, my expected output from the code is (df2)

df2

chr	start	end	V4
chr1	1214882	1214944	2
chr1	1215030	1215038	2

[Alex Reynolds]

If you need to merge one element at a time, you might use a Python or awk script to read in one element at a time, storing the first element. Progressively merge when subsequent element's end positions are within X bases of the first element's start position. When that distance test fails, then you print the range of the merged elements, and reset the "first" element. Repeat this test as you iterate through the rest of the elements. This is a fairly basic scripting exercise.

[jantu008]

Hi Alex,

Thank you for your reply. I figured it could be something simple to write even though my experience in Python or R is very basic. So I figured if this function existed already or there was a parameter for the 'merge' function in Bedtools that I could use for this purpose, I could just plug my data and get the probes for the wet lab portion, and then revisit the issue as a scripting exercise for my Python course ... Thank you!

Answer 1

[ATpoint]

The "wrong" result is correct. Merging intervals 1 and 2 creates the new interval 1214882 to 1214944 and 1214994 is 36bp away from the start of interval 3, triggering the merge again, and so it goes for interval 4, resulting in one final interval.

[jantu008]

Hi ATpoint,

I see what is doing now ... any idea how can I prevent it from doing this? I don't see any parameter on the code ... Thank you! Joana

[ATpoint]

It will come down to a custom script as Alex Reynolds suggests. Try something out, and come back if you get stuck (but try something yourself first ;-)