Bed file coordinate sorting

[lethalfang]

Apparently, I don't quite understand how a bed file should be sorted. I can't quite figure out what's wrong.

bedops --ec --everything myData.bed > /dev/null
May use bedops --help for more help.

Error: in oncosnp.bed
Bed file not properly sorted by first column.
See row: 771

cat myData.bed | awk 'NR>=769' | head
9	138338321	138339032
9	139327439	140167730
10	69083	3368595
10	367632	393272
10	448018	448786
10	703003	763714
10	998193	999290
10	1084691	1123868
10	1368946	1369689
10	1500525	1503838

Any idea? How does the ordering rule work for bed files?

A related question: when I used bedtools and use the "-sorted" flag, how should that bed file be sorted?

Thanks in advance.

[dpryan]

You treated them as numbers, so they're in numeric order. However, they need to be in lexicographic order, since "10" as a word comes before "9" as a word (the normal order would be something like: 1, 10, 11 ... 19, 2, 3, ...). This will make more sense if you have chromosomes X and Y and/or MT (or are using UCSC chromosome names).

[dariober]

A related question: when I used bedtools and use the "-sorted" flag, how should that bed file be sorted?

Sort by chromosome (lexicograhic) then by position. Something like this will do

sort -k1,1 -k2,2n -k3,3n a.bed > asort.bed

(In fact I suspect that for -sorted option to work, what matters is that chromosomes are in the same order in both files).

By the way, if records within chromosomes are already sorted by position and so you just want to sort chromosomes, use the --stable option in sort. It will save quite some time for big files.

[AlexReynolds]

You can use BEDOPS sort-bed to sort BED files:

http://bedops.readthedocs.org/en/latest/content/reference/file-management/sorting/sort-bed.html

GNU sort is not as fast as sort-bed (even with various tricks, like --parallel and so forth) and it's easy to get the options wrong.