Sort bed file by chr order and not lexicographically

[gmarco]

I've been trying to sort a bed file with bedtools and bedops.

Both seem to sort the bed file lexicographically.

That means if my input test file is:

input:

chr1 10 30
chr2 30 10
chr3 10 30
chr13 30 40
chr5 10 20

output:

chr1 10 30
chr 13 30 40
chr2 30 10
chr3 10 30
chr5 10 20

and desired output is:

chr1 10 30
chr2 30 10
chr3 10 30
chr5 10 20
chr13 30 40

any ideas how to achieve that?

[Heisman]

I convert the chrX and chrY to chr23 and chr24 with "sed -e 's,chrX,chr23,' -e 's,chrY,chr24,' input.file" and then pipe that into "sort -k 1.4,1n -k 2,2n" and then change the chr23 and chr24 back. This has to be altered if you have other "chr" values (M and all the small contigs).

[gmarco]

Don't you have to also remove "chr" from line start to make the sort work?

[GenoMax]

More here: http://www.biostars.org/p/64687/

[Heisman]

Don't you have to also remove "chr" from line start to make the sort work?

So if you 'sort -k 1.4,1n', the "1.4" tells it to start at the 4th character of the 1st column, so you don't have to remove the "chr". The thread GenoMax linked has other things worth looking at.

Also I typed this above ("sort -k 1.4,1n -k 2,2n"); it should be ("sort -k 1.4,1n -k 2,2n -k 3,3n").

[gmarco]

I convert the chrX and chrY to chr23 and chr24 with "sed -e 's,chrX,chr23,' -e 's,chrY,chr24,' input.file" and then pipe that into "sort -k 1.4,1n -k 2,2n" and then change the chr23 and chr24 back. This has to be altered if you have other "chr" values (M and all the small contigs).

Hello Heisman,

I'm trying sed command with no success:

sed -e 's,chrX,chr23,' -e 's,chrY,chr24' input.bed
sed: -e expression #2, char 12: unterminated `s' command

Fixed missing a comma after chr24 in second sed expression.

[Heisman]

Yeah, sorry about that. I probably make that typo more than any other.