Bedtools wrongly indicates zero coverage

Question

I have a BAM file produced using BWA-MEM and I need to calculate the coverage of each exon in this sequencing. I tried using this command:

bedtools coverage -a exome_sorted.bed -b exome_alignment.bam -sorted

And the result was the following:

chr1    12595   12802   ref|DDX11L1,ref|NR_046018,ens|ENST00000518655,ens|ENST00000450305,ens|ENST00000456328,ens|ENST00000515242  .       ref|DDX11L1     0       0       207     0.0000000
chr1    13163   13658   ref|DDX11L1,ref|NR_046018,ens|ENST00000518655,ens|ENST00000450305,ens|ENST00000456328,ens|ENST00000515242  .       ref|DDX11L1     0       0       495     0.0000000
chr1    14620   15015   ref|WASH7P,ref|NR_024540,ens|ENST00000538476,ens|ENST00000438504,ens|ENST00000488147,ens|ENST00000541675,ens|ENST00000423562       .       ref|WASH7P      0       0       395     0.0000000
...

That is, zero coverage for all exons. But there are reads mapping to the exons, as I can see by opening my BAM file on IGV and looking at some exonic regions:

I have already tried sorting the BED file using sort -k1,1 -k2,2n (as recommended on the Bedtools documentation) and using bedtools sort with the -faidx option, but neither option solved my issue.

if I skip the -sorted option, the memory usage becomes way too high.

I have also checked if the chromosome names are the same in the BAM file and in the BED file, using samtools view -H. Both use chr1,chr2,...,chr22,chrX,chrY, although BAM has chrM and stuff like chrUn_gl000214 which are not on the BED.

Thanks!

EDIT:

@ATpoint pointed out that all visible reads in this region have MAPQ0. I looked at another exon that did not have this issue, but Bedtools still outputs zero coverage. IGV picture below:

Answer 1

If this is your output :

chr1    12595   12802   ref|DDX11L1,ref|NR_046018,ens|ENST00000518655,ens|ENST00000450305,ens|ENST00000456328,ens|ENST00000515242  .       ref|DDX11L1     0       0       207     0.0000000

Then I guess your bedfile looks like this (the first 6 fields) :

• chrom : chr1
• FeatureStart : 12595
• FeatureEnd : 12802
• name : ref|DDX11L1,ref|NR_046018,ens|ENST00000518655,ens|ENST00000450305,ens|ENST00000456328,ens|ENST00000515242
• score : .
• strand : ref|DDX11L1

The issue I see here is that you are giving the wrong information to the strand field of your bedfile. It should be either +, either - or . if you don't want to consider the strand information.

Hope it helps !