sambamba depth base result discrepancies

[harperfauni]

Hi,

My name is Harper and I am currently using sambamba 0.6.6 in order to see the coverage for each base in a specific region. I was particularly looking at the gene LOC729737 and the UCSC genomic coordinates for it is 1:134772-140566 (0-based) using a bed file.

The bed file looks like:

1 134772 140556

My command is:

sambamba depth base -z -L bed file bam file | head -n15

The results are:
Processing reference #1 (1)
REF POS COV A C G T DEL REFSKIP SAMPLE
1 134772 8 0 0 0 8 0 0 sample
1 134773 8 0 8 0 0 0 0 sample
1 134774 8 0 0 8 0 0 0 sample
1 134775 8 0 0 0 8 0 0 sample
1 134776 8 0 0 8 0 0 0 sample
1 134777 8 0 0 8 0 0 0 sample
1 134778 8 0 0 8 0 0 0 sample
1 134779 8 0 8 0 0 0 0 sample
1 134780 8 0 0 8 0 0 0 sample
1 134781 8 8 0 0 0 0 0 sample
1 134782 8 0 0 0 8 0 0 sample
1 134783 9 9 0 0 0 0 0 sample
1 134784 10 0 0 0 10 0 0 sample
1 134785 10 0 0 0 10 0 0 sample

Looking at IGV (coordinates/positions translated into 1-based) with the 14 positions above, I saw a read depth of 11 or higher. I was wondering if sambamba was looking at a different region and generated an incorrect coverage value. What would be some explanations as to why I am seeing discrepancies in coverage values when looking at sambamba results and in igv when 0-based/1-based conversion has been taken into consideration? I was also wondering if indel regions could possibly affect sambamba depth base output.

Thanks and have a nice day!

[lomereiter]

Hi,

Reads with duplicate/failed QC flag or mapping quality=0 are excluded by default. You can tweak this using -F flag.

[nathanhaigh]

Hi @harperfauni, IGV will by default display:

• MAPQ 0 alignments
• Secondary and Supplementary alignments
• Vendor QC failed reads
• Duplicate reads

Take a look at IGV's settings, and you might be able to remove from display, the reads which Sambamba is also removing: https://software.broadinstitute.org/software/igv/Preferences#Alignments

0 reads