Counting read depth using bedtools

[Kaas]

I have sequenced a cell line exposed to UV and would like to know if any genes have been deleted compared to the ancestor. I extracted the CDS regions from the annotation file to annotation.bed and ran coverageBed in order to find the read depth at any given exon

coverageBed -abam DXB11.bam -b annotation.bed > depth.txt

the output for a particular domain was
NW_003614442.1 464809 465646 158 837 837 1.0000000
So 100% of the region 464809-465646 had a depth of 158 and that entire region of 837 bp was that depth, correct?

That that is very high as the theoretical depth should be 35. So i looked into the depth at every position of the genome
/BEDTools-Version-2.16.2/genomeCoverageBed -d -ibam DXB11.bam > DXB11.coverage

and looked into the same region (464809-465646) and done this way it had a median depth of 18 = much more realistic.
Are you able to see what i did wrong or maybe advice me another way of more easily getting to a median depth of each exon in the genome from a bam file?

[rnaeye]

Hi Kaas,
I think 158 is the number of features that overlapped the interval, not necessarily fold coverage. I would recommend that you read The bedtools manual.

Default Output:
After each entry in B, reports:
1) The number of features in A that overlapped the B interval.
2) The number of bases in B that had non-zero coverage.
3) The length of the entry in B.
4) The fraction of bases in B that had non-zero coverage.
You may want to try -d or -hist options.

Hope this helps.

[Kaas]

Hi rnaeye

Thank you for you answer. I tried going through the description for genomeCoverageBed (http://bedtools.readthedocs.org/en/l.../coverage.html) and for genomecov (http://bedtools.readthedocs.org/en/l...genomecov.html) but had a hard time translating their bioinformatic terms into what conclusion i can make from my own data based on the results.

The number of features in A that overlapped the B interval = number of reads that are identified in the exon region i specify. But then you would expect at least some kind of correlation between the length of a given region and the depth coverageBed gives, right? because i do not see any correlation. That is the reason why I find this a bit fishy.

ok, i will use -d and extract the median from there

[rnaeye]

Hi,
Try to google following search them "The BEDTools manual PDF"
You can download a PDF version of user manual. I think it explains better. I guess you should calculate coverage per base and conclude it from there. have fun, best.

[shuoguo]

read the help of bedtools coverage
i think you can use -hist or -d option