Extracting Regions of Zero Coverage

[drea11]

I am looking for a quick and simple way to identify regions of zero coverage in a sequencing set. My ultimate goal is to try and determine whether there might be genomic features which contribute to this.

I've used both GATK and Samtools depth to identify the depth of coverage at each base position across my genome. What I'd like is to have the base positions for each REGION which are not covered (ie. 2307-2412). I have depth of coverage on many samples, and would like to have a quick way of getting this information across all of them. Are there any functions out there which could do this? Am I better off just sticking with some sort of bash command?

Below is a sample of what the output of GATK is (depthofcoverage) as an example.

Locus Total_Depth Average_Depth_sample Depth_for_s1 Depth_for_s2 Depth_for_s3
genome:1 283 94.33 82 111 90
genome:2 284 94.67 82 111 91
genome:3 285 95.00 82 112 91

[dpryan]

I imagine that tabix indexing/fetching would be useful here. That's the basis for things like indexing VCF files.

[drea11]

Thanks for the speedy reply. Could you perhaps elaborate on your answer a bit? I'm not familiar with tabix, and the online documentation seems somewhat unspecific.

[piet]

I'd like is to have the base positions for each REGION which are not covered (ie. 2307-2412)

Maybe the 'genomecov' subcommand from 'bedtools' with option '-bga' is useful for you.

http://bedtools.readthedocs.org/en/latest/content/tools/genomecov.html

[dpryan]

Thanks for the speedy reply. Could you perhaps elaborate on your answer a bit? I'm not familiar with tabix, and the online documentation seems somewhat unspecific.

If we modify your file slightly so that the contig and position are in different columns and replace spaces with tabs:

Locus	Pos	Total_Depth	Average_Depth_sample	Depth_for_s1	Depth_for_s2	Depth_for_s3
genome	1	283	94.33	82	111	90
genome	2	284	94.67	82	111	91
genome	3	285	95.00	82	112	91

Then we can:

bgzip foo.tab
tabix -b 2 -e 2 -f -s 1 -S 1 foo.tab.gz #This will index the file
tabix foo.tab.gz genome:2-3

The last command will output the last 2 lines. Of course if you have a LOT of lines and multiple contigs/chromosomes/genomes, then this is more useful than the small example. The biggest benefit to this method is that you can randomly ask for ranges without needing to process everyline until you get to that range (if you've ever dealt with indexed VCF or BAM files, it's the same underlying principle). This is actually how IGV and many other tools randomly access VCF and BED files.

Note that the bgzip utility is really just a special version of gzip, so you can view the resulting files with zcat or even just gunzip them.