Model | Type I error rate |
---|
Binomial | 4.9% |
NB: p=D N A | 3.5% |
PG: Ï•=D N A | 3.8% |
PG: q=1/2 | 3.2% |
- Allele-specific counts from DNA and RNA-seq data were simulated with no bias and no allelic imbalance for three replicates each of 10,000 exonic regions and analyzed using the binomial exact test, the random bias parameter NB and PG models that use DNA controls and the PG model with fixed bias parameter, q = 1/2. Even when there is no bias, the Bayesian models have better performance than a binomial exact test.