# User:R. Eric Collins/MBL/Bayes

From OpenWetWare

< User:R. Eric Collins | MBL

## Bayesian Phylogenetics

- probabilities are applied to intervals, not points
- curve is probability density, not probability

- gamma distribution
- Gamma(1,1) = Exponential(1)
- Gamma(1,2) = Exponential(1/2) in some cases
- sometimes papers use different definition of gamma, where b = 1/b

- dirichlet process
- use simplified (e.g. clustered at 97% similarity) tree as dirichlet prior for treespace?

## MrBayes lab

- mrbayes only recognizes bifurcating tree
- "not converged diagnostics"

- database of alignment "types" e.g. those that work well, don't work well, easy, hard
- what makes a dataset harder to converge than another? more tree islands...