mpileup -B decreases coverage

[grillino]

Hi all!!
I am new in this forum, I am working on Illumina NGS data since recently, and I have some troubles!

I was using this command:

samtools mpileup -C50 -q 25 -f reference.fa file.bam > file.mpileup

when I realised it was producing strange valleys in base quality phred scores in correspondence to mismatches to the reference genome, I now think the BAQ (base alignment quality computation) was responsible for those, so I added the -B option:

samtools mpileup -B -C50 -q 25 -f reference.fa file.bam > file.mpileup

Now the valleys in base quality seem to have disappeared, but I see much lower coverage in different regions of the genome… Of course, the 2 commands were used on the very same .bam file, nothing else should be different apart from the -B option…
I can't see how the -B option can cause the discard of any read, thus the lower coverage I see in my pileup files. Unless in the previous command, the default BAQ computation was disabling some other option, like the -q 25???

Any help is veeeeery welcome!!
Thanks a lot!
grillino

[grillino]

Pleaseee!!

[SNPsaurus]

I also have used the -B option because of how it reduces the quality of all SNPs, it seems. I didn't notice the read loss though (at least, it wasn't obvious enough for me to notice). Do you need to use -C50? Try it without that.

[grillino]

I also have used the -B option because of how it reduces the quality of all SNPs, it seems. I didn't notice the read loss though (at least, it wasn't obvious enough for me to notice). Do you need to use -C50? Try it without that.

I am running the same 2 commands without -C50, I'll see what happens…. The effect of decreased coverage was very evident, but I think not everywhere in the genome… It was localised.

I still can't see how this can be!!

[grillino]

I also have used the -B option because of how it reduces the quality of all SNPs, it seems. I didn't notice the read loss though (at least, it wasn't obvious enough for me to notice). Do you need to use -C50? Try it without that.

Actually eliminating -C50 I see intermediate coverage with respect to the combinations "-B -C50" and "-C50" alone.

At one position I took as example, for different parameter combinations I see this coverage:

"-C50" coverage = 10
"-C50 -B" coverage = 2
"-B" coverage = 5

So the -C50 is behaving reasonably I guess, if I disable it coverage increases. But why coverage decreases with -B option, given everything else is the same, I cannot see.