bedops -u does not work properly

Question

[Ming Tommy Tang]

bedops -u is not sorting the bed files

bedops --version
bedops
  citation: http://bioinformatics.oxfordjournals.org/content/28/14/1919.abstract
  version:  2.4.14
  authors:  Shane Neph & Scott Kuehn

#dummy file:
 cat my.bed 
chr1    1   5   peak_1
chr1    2   6   peak_2
chr1    3   7   peak_3
chr1    2   4   peak_4
chr1    8   10  peak_5
chr1    3   6   peak_6

#test bedops -u 

bedops -u my.bed my.bed 
chr1    1   5   peak_1
chr1    1   5   peak_1
chr1    2   6   peak_2
chr1    2   6   peak_2
chr1    3   7   peak_3
chr1    2   4   peak_4
chr1    3   7   peak_3
chr1    2   4   peak_4
chr1    8   10  peak_5
chr1    3   6   peak_6
chr1    8   10  peak_5
chr1    3   6   peak_6

correct output should be:

sort-bed my.bed my.bed 
chr1    1   5   peak_1
chr1    1   5   peak_1
chr1    2   4   peak_4
chr1    2   4   peak_4
chr1    2   6   peak_2
chr1    2   6   peak_2
chr1    3   6   peak_6
chr1    3   6   peak_6
chr1    3   7   peak_3
chr1    3   7   peak_3
chr1    8   10  peak_5
chr1    8   10  peak_5

please fix this, my bedops -u *bed | bedmap —count —echo —delimit ‘\t’ - is failing... according to http://bedops.readthedocs.io/en/latest/content/usage-examples/multiple-inputs.html
bedmap should complain that if the input is not sorted rather than just let it go silently.

Thanks,
Tommy

[Sinji]

I think you have to include the --ec flag for error checking no?

[Ming Tommy Tang]

yes, you are right. -ec is for checking error. but bedops -u should not expect sorted bed files, right?

[Sinji]

I think the requirements for all bedops operations including -u are that the input be sorted prior to being run, however, i've never gotten any errors when giving un-sorted files to bedops, I don't think the -u flag changes that, at least not in my experience. I always sort every file I give bedops just for this reason.

[Ming Tommy Tang]

Now it makes sense. every input bed file should be sorted. with -u it does not give me errors, but just missing many overlapping entries. bedops is a very good tool BTW:)

[Sinji]

Great, and yes, bedops is among my most used tools, very powerful!

[Ming Tommy Tang]

it is under-appreciated compared with bedtools, another good one :)

[Ming Tommy Tang]

even version: 2.4.30 (typical) is having the same issue.

[Alex Reynolds]

Just give bedops -u sorted data and it will work fine.

Everyone should definitely update to 2.4.30 where possible, which includes a number of major speed enhancements and fixes across the board.

Answer 1

[Alex Reynolds]

You need to provide sorted inputs to bedops -u. It doesn't replace sort-bed!

 $ sort-bed my.bed > my.sorted.bed
 $ bedops -u my.sorted.bed my.sorted.bed > answer.bed

The file answer.bed is sorted correctly. You don't need to sort answer.bed. This is because the inputs are sorted, which allows bedops -u to do a merge sort under the hood.

Or if you like process substitutions, you can skip making intermediate files:

$ bedops -u <(sort-bed my.bed) <(sort-bed my.bed) > answer.bed

Those <(sort-bed my.bed) blocks are short-lived equivalents to my.sorted.bed.

I think --ec should tell you if your inputs are not sorted. Hope this helps and thanks for using BEDOPS!

[Ming Tommy Tang]

Thanks for the detailed answer. I used <() quite a bit as well :) learning to use more BEDOPS.

[1769mkc]

To clarify my doubt ..so when i have say replicates from same condition then i have to use this -u, --everything and when I want to find overlap between lets say different biological condition I have to use this one -m, --merge. Or I m doing it opposite way?