Mutect Allele Counts Vs Mpileup

Question

[Noushin N]

Hi everyone,

I used Mutect to call somatic mutations in my exome sequencing experiment. Just to validate some results, I decided to make mpileup at the positions of somatic mutations. However, I am finding some discrepancies in the number of read counts supporting the alleles at some positions, between mpileup results and what mutect reports in its t_ref_count and t_alt_count columns in the output.

I do realize that some quality filter is applied to the reads. However, I do not seem to find the value mutect uses by default. Turning to my mpileup file, I have tried filtering at different thresholds of mapping quality; but none seem to result in the values reported by mutect.

Here are two example lines:

mpileup:

chr11   3075562 t       29      ..c,cc,,,,C.,c,C,,,.,c,,..,,.   #>!@!!?@@?!?@!?!:>@??!@6>>@;=

mutect:

chrom   start   end     ref     alt     t_ref_count     t_alt_count     n_ref_count     n_alt_count
chr11   3075561 3075562 T       C       20      5       25      0

All the bases reporting "c" seem to have mapping quality of zero, which is the minimum quality at this position. So, any filtering will result in t_alt_count of 0.

Is there anyway to find a compromise between the counts from the two sources?

Thank you!

[Chirag Nepal]

Hi Noushin,

I have a similar question like you, would be nice if you could help.

I have muTect output, and wanted to see the read count for each variant.

contig    position    context    ref_allele    alt_allele    t_ref_count    t_alt_count    t_ref_sum    t_alt_sum    t_ref_max_mapq    t_alt_max_mapq    t_ins_count    t_del_count    normal_best_gt    init_n_lod    n_ref_count    n_alt_count    n_ref_sum    n_alt_sum    judgement
chr1    22328092    CCAxTCA    A    G    22    4    636    103    -2    59    0    0    AA    8.123008    27    0    803    0    KEEP
chr1    117142707    GGCxGAC    G    A    32    12    760    334    124    89    0    0    GG    7.743379    26    0    612    0    KEEP

I see that t_ref_count and t_alt_count do not sum up to t_ref_sum. Same applies for normal.

Could you explain why is that.

Thanks!

[Noushin N]

Hi Chriag,

I came back to biostars today after a long while and noticed this comment. You probably have this figured out already, but I am adding this for future reference.

According to the documentation, t_ref_count and t_alt_count report the read counts supporting each allele, while t_ref_sum, and t_alt_sum correspond to the sum of read quality scores.

Best,
Noushin

Answer 1

[enabieva]

Hi,

having had a similar problem, I just discovered the answer on page 1 of Supplementary Materials of the MuTect paper (among other things it filters on total baq scores of all mismatches in the read). Hope that helps.

[Noushin N]

Thank you so much! That is most likely the reason.