next up previous contents
Next: analyser Up: mcmcphase Previous: PERTURBATION Block for MIXED   Contents

Proposals priority

Each cycle mcmcphase perturbs one branch length and a topology change is tried every ten cycles (see the Bayesian phylogenetics section). Each cycle PHASE will also try to modify the substitution model. Let us consider the following example with the MIXED model and three substitution models,
{PERTURBATION}
#PERTURBATION OF THE TREE :
...

#PERTURBATION OF THE MODEL :
Model 1 priority = 8
Model 2 priority = 24
Model 3 priority = 7

Average rates, proposal priority = 1
Average rates, initial step = .3
Average rates, proposal minimum acceptance rate = .15
Average rates, proposal maximum acceptance rate = .20

{PERTURBATION1}
Frequencies, proposal priority = 1
Rate ratios, proposal priority = 2
Gamma parameter, proposal priority = 1
{$\backslash$PERTURBATION1}
{PERTURBATION2}
Frequencies, proposal priority = 3
Rate ratios, proposal priority = 4
Gamma parameter, proposal priority = 1
Invariant parameter, proposal priority = 1
{$\backslash$PERTURBATION2}
{PERTURBATION3}
Frequencies, proposal priority = 4
Rate ratios, proposal priority = 2
Gamma parameter, proposal priority = 1
{$\backslash$PERTURBATION3}
{$\backslash$PERTURBATION}

The mcmcphase program will modify the average substitution rate of the second model with probability given in 1.1a; the probability of modifying the average substitution rate of the third model is the same. The average substitution rate of the first model is the reference, it is never modified and remains equal to $1.0$. With probability given in 1.1b, mcmcphase will modify the parameters of the substitution model $i$ (i.e., any parameters but the average substitution rate). The priorities inside the corresponding PERTURBATIONi block are used, the figures inside each PERTURBATIONi block do not have any effect outside the block.


\begin{oldsubeqnarray}P(average\ rate) & = & \frac{Average\ rates,\ proposal\ pr...
...tes,\ proposal\ priority + \sum_{i=1}^{N}Model\ i \ priority\end{oldsubeqnarray}

In our example, if the second model is selected for the next modification, we define $priority_{total}$:

\begin{displaymath}
priority_{total} = priority_{frequencies} + priority_{gamma}...
...y_{invariant} + number_{rates\ ratios}*priority_{rate\ ratios}
\end{displaymath}

and we modify either all the frequencies (probability 1.2a), or one of the rate ratios (probability 1.2b each), or the invariant parameter (probability 1.2c), or the gamma shape parameter (probability 1.2d).

\begin{oldsubeqnarray}P &=& \frac{priority_{frequencies}}{priority_{total}}\slab...
...\frac{priority_{gamma}}{priority_{total}}\slabel{eqn2:gamma}\end{oldsubeqnarray}


next up previous contents
Next: analyser Up: mcmcphase Previous: PERTURBATION Block for MIXED   Contents
Gowri-Shankar Vivek 2003-04-24