`mpileup --max-depth 0`?

[schorlton-bugseq]

Hi! We were wondering what mpileup --max-depth 0 does? The samtools docs are pretty clear:

However, there is no mention of this setting in the bcftools docs:

The peculiar thing is that with bcftools (version 1.22-13-g458f97a1), setting --max-depth 0 paradoxically gives us DP ~70 with a BAM of >10,000X depth, whereas inspecting the BAM or setting --max-depth 1000000 gives the correct DP.

It also appears that others are using bcftools mpileup --max-depth 0, e.g. https://github.com/nf-core/viralrecon/blob/3731dd3a32a67a2648ea22c2bd980c224abdaee2/conf/modules_illumina.config#L405

Thanks!

[pd3]

In bcftools the value of --max-depth 0 is passed directly to htslib via bam_mplp_set_maxcnt() and is, in this case, really set to 0.

But why do you get DP~70? I am guessing it is because of the way the depth counting works, I remember it is not a hard limit and probably has to do with how many reads start at the same position.

Since the current behavior is meaningless really, I changed it now to follow that of samtools mpileup, making --max-depth 0 use the maximum possible value, i.e. unlimited depth.