I am doing analysis of exom data and the major objective is the identification of variants. I want to know how the Poisson model is applied in calculating the p-value of the variants. It would be nice if someone would explain it with a example as theoretically I do understand its formula.
I have read somewhere that given the number of reads supporting a variant, Poisson distribution is applied to estimate the probability of observing that the number of reads supporting the variants are true and not by sequencing error.
To my understanding tools such as the GenomeAnalysis Toolkit (GATK) use a binomial, not a Poisson, distribution to compute the probability that a position contains a variant. The basic idea is to compute the probability that, given X reads covering the position of which Y differ from the reference, what's the probability that the position contains a heterozygous or homozygous (non-reference) variant?
Note that the when the number of trials (reads in this example) is large and the probability of success (observed non-reference bases) is small, the Poisson can be used to approximate the binomial, but I don't believe this approximation would be appropriate in calling variants from sequencing data.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology