From: Ana Rodrigues Grant <anapcr@um...> - 2013-02-18 15:32:25
I am struggling to interpret the output of the trio calling mode and
hoping someone can shed light.
I used the following command:
samtools mpileup -DSuf bwa/hg19.fa Sample_14827_dd.bam
Sample_14825_dd.bam Sample_14826_dd.bam | bcftools view -vcgT trioauto
-s trio_config.txt -> trio_calls.vcf
The kinds of variants I would be most interested in extracting from
the vcf are those where: the son has a variant, and the mother and
father are either carriers or homozygous reference.
Looking at the GT field for each of 3 samples, I find a lot of such
variants. I am struggling to decide which to consider confident calls.
Does anyone have some combination of UGT/CGT/CLR/GT values that they
use to select high confidence variants of interest?
When doing tumor-normal pairs, I've been using the CLR value with a
cutoff of 60 to decide which variants with different genotypes to
trust. However, in the trio vcf file, I see calls with the genotypes
that I am looking for that don't have a CLR associated with it. For
chr18 77703299 . C G 151.0 .
GT:PL:DP:SP:GQ 1/1:149,101,89:7:0:9 0/1:135,0,210:36:12:99
In this case, the quality of the son's genotype is not great, but I
don't understand why no CLR is reported? This call also doesn't have
In cases where the CLR is reported, what exactly is the CLR reflecting
in trio data? Is it the likelihood that UGT and CGT are different? Or
the likelihood that the predicted sample genotypes are different?
Thank you for any help you can provide,
Ana Rodrigues Grant, Ph.D.
Department of Computational Medicine & Bioinformatics / Bioinformatics Core
University of Michigan Medical School
Ann Arbor, MI
Get latest updates about Open Source Projects, Conferences and News.