Inconsistent results with sambamba depth

[moonso]

Hi,

I found a strange problem.
When using sambamba depth region and a bed file with exons from ccds, sambamba gives different results depending on if we use the file with all exons or just one region.
If we look at one region only the file looks like

$ cat small_region.bed
10  46339772    46339841    10-46339773-46339841    0   -   CCDS7215.1  AGAP4
$ sambamba depth region -t 8 sample.bam -L small_region.bed
Processing reference #10 (10)
10  46339772    46339841    10-46339773-46339841    0   -   CCDS7215.1  AGAP4   2   0.768116    ADM1003A2

But when running sambamba on the whole bed file and we look at coverage for the same region:

$ sambamba depth region -t 8 sample.bam -L CCDS.current.bed -o sambamba_regions.csv
$ grep 46339772 sambamba_regions.csv
10  46339772    46339841    10-46339773-46339841    0   -   CCDS7215.1  AGAP4   569 40.0551 ADM1003A2

We see a completely different coverage result...
We inspected the bam file with samtools tview and it looks like the first result is correct, it is very low coverage.
I understand that this might be hard to replicate without the original data but do you have any idea what might cause the problem?

Thanks

[lomereiter]

Thanks, it turned out that unsorted BED files weren't handled correctly.