samtools mpileup giving QUAL of 0

Question

[blur]

Hello all,

An innocuous mpileup gives QUAL of 0 for all bases. 0 shouldn't even be a legal QUAL value (as it is the result of an exponent, according to VCF format documentation). This is the command we're running:

samtools mpileup --VCF -r ChrA_C_glabrata_CBS138:1-100,000 --output-tags DP,DP4 -u -f CBS138.fasta 12.sorted.bam > 12.mpileup.txt

A sample output line:

ChrA_C_glabrata_CBS138    337    .    G    A,C,T    0    .    DP=31;I16=4,11,7,8,530,19116,501,17779,6,6,2,2,298,6778,286,6564;QS=0.5,0.4,0.0666667,0.0333333;VDB=0.523807;SGB=-0.688148;RPB=0.768403;MQB=0.461075;MQSB=0.999074;BQB=0.952765;MQ0F=0.741935    PL:DP:DP4    0,7,3,30,26,5,36,31,9,5:30:4,11,7,8

Thanks!

[blur]

Another example, that might explain the diploid assignment:

Total count: 95
A      : 0
C      : 67  (71%,     33+,   34- )
G      : 0
T      : 28  (29%,     12+,   16- )
N      : 0

[Devon Ryan]

As I said previously, samtools assumes you have a diploid organism. Sites like the one above suggest that either you don't have a haploid organism, you have multimappers, you have incorrect alignments, or you have notable genetic heterogeneity.

Answer 1

[Devon Ryan]

0 is certainly legal, since 10^0 = 1, meaning that there's a 100% chance that the call is wrong. This is unsurprising given that all four possible base changes are apparently present at similar depths and samtools assumes a diploid genome.

[blur]

my genome is haploid, and all the SNPs I got look like the line above...

Also, my DP doesn't match the read depth - I opened these files in IGV beforehand, and for example, if IGV has 94 reads, the DP is marked for the same position as 31....

mpileup problem?

[Devon Ryan]

There's some filtering that happens before the DP metric is made. This is unlikely to be an mpileup problem per se but rather either a usage problem or simply due to having a bunch of multimappers (otherwise, the depths of the alternate sequences makes no sense). Are you trying to use this for some sort of pooled input?

[blur]

no... this is what worries me...

[Devon Ryan]

What do things look like in IGV in this area?