i cant sort bed file. Error in bedtools merge.

Question

[unique379]

Dear all,

I have bed file and try to do some work using bedtools merge. To do this first I have sorted my bed file and then used bedtools merge. But got error..

Error: Sorted input specified, but the file ......

So, I open my file to see possible error then I realized that used sort -k 1,1 -k 2,2n command does not worked properly.... I mean some line has sorted and some of them are not...

chr1 149928394 149928450 chr1_163309 2 + 5
chr1 165786346 165786432 chr1_186360 2 - 30
chr11 66240033 66240092 chr11_24483 1 + 3
chr11 100970875 100970934 chr11_26610 1 + 2
chr11 122156196 122156259 chr11_38105 1 - 5
chr1 168141086 168141168 chr1_186480 3 - 747
chr1 182391712 182391791 chr1_187506 2 - 8
chr12 1805175 1805235 chr12_39518 1 + 3
chr12 21518670 21518726 chr12_41343 1 + 4
chr12 24743093 24743136 chr12_51637 1 - 9
chr12 30513872 30513960 chr12_52133 2 - 64
chr1 231586879 231586960 chr1_170106 2 + 19
chr1 242310594 242310652 chr1_192505 16 - 33
chr12 9025545 9025632 chr12_40322 8 + 19
chr12 52416011 52416060 chr12_53993 1 - 9
chr12 93282380 93282488 chr12_57473 16 - 29
chr12 94834227 94834317 chr12_47071 2 + 53
chr12 95308393 95308445 chr12_47159 3 + 12
chr13 20880735 20880793 chr13_65648 10 - 25
chr13 21148757 21148795 chr13_61611 1 + 339

Please help.

Note1: sort version (GNU coreutils) 8.22, OS = centOS

Note2: Same file I used in different system RedHat with sort version 5.97, its work perfectly. No problem was found by bedtools.

Thanks

[dariober]

Try also sort -k1,1 -k2,2n -k3,3n. But anyway the lines you post do not seem to be the output of sort, regardless of GNU version and OS. I suspect there is something missing here, maybe you are looking at the file before sorting or the column delimiter is not tab... Maybe have a look at the file with cat -vet file.bed, you should see tab as ^I and end-of-lines as $.

[unique379]

Thanks but its sorted bed.. As I said and I checked by cat -vet and found tab as ^I and end-of-lines as $.

By the way this sort -k1,1 -k2,2n -k3,3n also does not work for me.

[michael.ante]

If you already use the Bedtools, why not using its sortBed function?

sortBed -i infile.bed > outfile.bed

Anyway, your example looks a bit suspicious: it has 7 columns, but usually it should have 6 or 12 (see here).

Answer 1

[Pierre Lindenbaum]

Try specifying LC_ALL and the delimiter .

LC_ALL=C sort -t '(insert-tab)' -k1,1 k2,2n input.bed > out.bed

[unique379]

Thanks Pierre for your hint....

I used

LC_ALL=C sort -t$'\t' -k1,1 -k2,2n

instead

LC_ALL=C sort -t '(insert-tab)' -k1,1 k2,2n

and this time worked.

Answer 2

[Alex Reynolds]

Consider using BEDOPS sort-bed:

sort-bed in.bed > out.bed

Generally works faster than Unix sortat sorting BED files, and it supports arbitrary column numbers and other potential use cases that are problematic for other tools.