Clc genomics, variant caller

Question

[sara.barbosa.salazar]

Can anyone explain me the difference betweent the quality-based or the probablistic variant caller?

Answer 1

[skbrimer]

The quality-based variant caller is base off the read quality at the SNP and the surrounding nucleotides and is based off the Neighborhood Quality Standard algorithm. Which basically means that if your data is good in that "neighborhood", everything but one or two gaps/mismatches and has a overall Q score of X then it feels confident in saying hey this is different.

The probabilistic variant caller is an error correct Bayesian algorithm that works of the mapped reads first. So it looks at all the reads for each position and generates a table of likely-hood ratios of seeing that nucleotide at that position. Then it compares it to your reference and if they are different it gets flagged as a possible SNP.

Either one would work, however the probabilistic might give you a higher false positive rate.

[sara.barbosa.salazar]

I have used both, and what i see is that quality gives me mutch more delections and insertion and probablistic gives me more of altered nucleotides. So i'm thinking that maybe I will cross both results avoiding false positives from the two methods...