biGaussianNull: biGaussianNull
In BANFF: Bayesian Network Feature Finder

Compute a prior null distribution without given any prior biological knowledge. Methods used to estimate this prior include: "biGaussian","biGaussianWith0Cutoff","biGaussianMean0","biGaussianModeSplit" etc.

1 2	biGaussianNull(rstat, null.quantile = c(0.25, 0.75), method = c("biGaussianMean0", "biGaussianModeSplit"))

`rstat`	The observed test statistics.
`null.quantile`	Data quantiles, pre-fixed so that we could reduce the impact of extreme values for estimating prior null distribution. Default value is c(0.25, 0.75).
`method`	A char. The methods we used to do estimation: either "biGaussian"– EM algorithm for mixtures of two univariate normals, or "biGaussianWith0Cutoff"– assume all negative test statistics forms one normal and all positive test statistics forms the other one normal. And proportion parameter is proportional to a number of observations each class, or "biGaussianMean0"– null is formed by two half normals. "biGaussianModeSplit"– split data from median value, then flip each part to the other side to estimate normal distribution.