Entering edit mode
5.7 years ago
erfan741
▴
80
I have called a set of indels from paired end sequencing data using pindel, but the reported allele fractions are not accurate. I was womndering if there is a tool available for calculating reference and alternative allele counts for known variants (indels and dups with known break points and known alternative allele) from paired end seq data?
Thanks
if your final output from sequencing is vcf, you can use bcftools stats for indels. Before that probably, it is a good idea to left align and collapse VCF for indels. Refer to https://samtools.github.io/bcftools/bcftools.html#stats
my output is VCF but i want to extract the new metrics from the bam file. The estimates in the VCF file is not accurate. BCFtools reports stats based on what is available in the VCF file.
if you are not satisfied by variant callers such as PINDEL, you can use other variant callers such as DINDEL, platypus, GATK haplotype caller. They all support bam input, as I understand. By any chance, did you do base recalibration and variant recalibration using known indels in variant calling pipeline?
Those other callers do not detect the indels that I want. Pindel was the only one that superseded all in detecting all of my known variants, but it doesnt report the correct AF. I need a tool that give me the correct AF of a known variant. What variant calibration method do you recommend? Even if I do variant callibration (which is something I had planned to do), I will need a method to give me the real allele fractions.
I guess AF is coming from AC and AN. Do you see any discrepancies in them too?
Yes that s the source of discripancy. For a variant which has a known AF of 0.5, pindel reports the AC an AN of 200 and 1000 sort of thing!
Are there any other alleles at the same coordinates?
Most often not. rarely