changing the variant fraction parameters in UG or HC - for RNA editing identification

we are trying to use th ecariant callers (UG or HC) for identifying RNA editing sites in RNA-Seq samples.
As the fraction of the variants (i.e. editing levels) can range from 0% to 100% (as oppose to 0%,50% or 100% when dealing with genomic variants), we are wondering if there is any argument in the UG/HC walkers that can be tuned according to our choice, that determines the variant fraction?
in their paper, Ramaswami et al 2013 (Nature Methods, Vol 10(2) p.128) used "stand_call_conf 0" and "stand_emit_conf 0", and then performed filtering according to their ampirical choices.

does these arguments determine the fraction of variant reads? (we could not understand exactly what these arguments mean)

Will appreciate your help




  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie

    Hi Moran,

    The confidence threshold arguments you are referring to are meant to control what minimum quality score variants must have in order to be emitted to the output VCF file. By setting them to zero, Ramaswami et al. effectively made the caller output every site that showed potential variation. This is one possible way to obtain some of the information you seek, but it is rather crude. Until recently there wasn't really a better way to do this, but there is now (to some extent). With the new GVCF mode of HaplotypeCaller, you can get detailed records for every genomic position with data on all observed alleles, so you're not constrained by allele frequency expectations. We don't yet have detailed recommendations on how to use the GVCF data to assess RNA editing, but I think that is the most promising way to look at it right now.

Sign In or Register to comment.