BCFTools missing mutation at 8%

[sowalsky]

I am having difficulty figuring out why bcftools is not calling my SNP. When I run samtools mpileup, the position is being "piled up" properly but it is not being interpreted as a SNP. It is a position read at extremely high depth (100,000x).

Here is how I am invoking the pileup/mutation calling:

samtools mpileup -d10000000 -BRuf ~/hg19.fa alignment.bam | bcftools view -vcg ->snps.vcf

I have tried varying the -t and -p options in bcftools view without success.

If I omit -v from bcftools, I see that bcftools is not regarding the position as a variant. Here is the output from that line:

chrX 66941716 . G . 283 . DP=113309;AF1=0;AC1=0;DP4=54138,47121,4400,3713;MQ=25;FQ=-282;PV4=0.18,1,1,1 PL 0

So as you should see there are significant reads at the position that should support the SNP. Does anyone have any suggestions?

[sowalsky]

Just a follow-up. By using:

samtools mpileup -d10000000 -BRuf ~/hg19.fa alignment.bam | bcftools view -p 2 -vcg ->snps.vcf

it forced bcftools to consider my mutation but it gave it a really poor quality score. It found:

chrX 66941716 . G A -0 . DP=113309;AF1=0;AC1=0;DP4=54138,47121,4400,3713;MQ=25;FQ=-282;PV4=0.18,1,1,1 GT:PL:GQ 0/0:0,255,255:99

The original bam file was from RNA-seq of a highly-heterogeneous population. Is there a way to control for this, so bcftools would accept this as a genuine mutation and not score it so poorly?