Hi sjneph, thanks for your reply. Did not realize it could be that easy. I can extend your code so that it also allows for windows with a step size. And I will have a look at the bedops forum, last time I checked it did not look very active so that's why I asked here :-)
With bedtools you can make genomic windows from a genome file or a bed file
input.bed
chr1 1000000 1500000
chr3 500000 900000
[prompt]$ bedtools makewindows -b input.bed -w 250000
chr1 1000000 1250000
chr1 1250000 1500000
chr3 500000 750000
chr3 750000 900000
Does the bedops suite provide a similar way to create genomic windows?
2 answers
Hi Isran,
Here is a pretty straightforward script that breaks up regions into fixed-sized chunks, using a standard awk 'program'. I put it inside of a simple c-shell script. You wouldn't have to do that though. If you name this script, say, as windowed-chunks, then you'd invoke as:
[prompt]$ ./windowed-chunks 250000 input.bed
#!/bin/csh -ef
if ( $#argv != 2 ) then
printf "Expect a window size and an input BED file!\n"
exit -1
endif
awk -v winsize=$1 \
'{ \
i = $2; \
while ( i < $3 ) { \
j = i+winsize; \
if ( j > $3 ) { j = $3; } \
print $1"\t"i"\t"j; \
i = j; \
} \
}' $2
exit 0
You could embed this awk statement into a bash or sh script instead if you wanted, or even a simpler awk script (this would allow you to get rid of all of the '\' characters I used). If you wanted to make this script a bit more robust, you could add checks that the window size given isn't a negative number or string, and other little things like that. If you have questions regarding BEDOPS, feel free to join the forum at http://bedops.uwencode.org/forum/ - we'd be happy to help.
You bet. The forum has been very quiet for a while now. We monitor it regularly, but there is very little activity as you pointed out. I'd like to see that change. I guess I'll float around out here too. This is my first post, so I'm happy it was helpful.
Version 2.4.3 of the bedops tool now offers --chop and --stagger functions to melt input regions into non-overlapping, adjacent or overlapping windows.
For instance, given input.bed:
chr1 1000000 1500000
chr3 500000 900000
You can use --chop 250000 to get the desired adjacent windows:
$ bedops --chop 250000 input.bed
chr1 1000000 1250000
chr1 1250000 1500000
chr3 500000 750000
chr3 750000 900000
Use of --stagger will step the window from the initial start position by the specified amount, keeping within bounds. Here's an example that makes overlapping windows:
$ bedops --chop 250000 --stagger 100000 input.bed
chr1 1000000 1250000
chr1 1100000 1350000
chr1 1200000 1450000
chr1 1300000 1500000
chr1 1400000 1500000
chr3 500000 750000
chr3 600000 850000
chr3 700000 900000
chr3 800000 900000
The stagger value can be larger than the chop to create non-overlapping windows, where all output elements fall within the bounds of the input elements:
$ bedops --chop 50000 --stagger 100000 input.bed
chr1 1000000 1050000
chr1 1100000 1150000
chr1 1200000 1250000
chr1 1300000 1350000
chr1 1400000 1450000
chr3 500000 550000
chr3 600000 650000
chr3 700000 750000
chr3 800000 850000
Log in to answer this question.
Does it matter? The two tools complement each other pretty well. Why not just use the windows created by
bedtoolswithbedops?hi aaronquinlan, i am trying to minimize the amount of dependencies of our pipelines. but it seems like we will continue to use both