Model (JC+gaps, K2P+gaps, F81+gaps and F84+gaps)
|
M |
F84 |
Estimate initial character frequencies from data
|
E
|
YES
|
Estimate transition/transversion ratio from data
|
R
|
YES
|
Frequencies of the three topologies
|
P
|
0.333333 0.333333 0.333333
|
Difficulty of changing trees
|
D
|
0.9
|
Read in initial hidden state sequence?
|
J
|
YES
|
This uses the F84 (Felsenstein 84) model of nucleotide
substitution, estimates both the
initial stationary frequencies and the transition-transversion
ratio from the data, sets the initial recombination
parameter lambda to 0.9,
selects a uniform a priori
distribution on the tree topologies,
and reads in the initial state
sequence from
mosaic.in.
Length of the burn-in period
|
B
|
1000000
|
Length of the sampling period
|
.
|
1000000
|
Number of points to return
|
N
|
1000
|
Thinning interval
|
I
|
1000
|
Tuning interval (for proposal distributions)
|
C
|
100
|
Update lambda with Gibbs sampling
|
W
|
YES
|
Annealing scheme for lambda
|
Q
|
PAR
|
Update stationary frequencies in MCMC algorithm
|
U
|
YES
|
Update transition-transversion ratio in MCMC algorithm
|
A
|
YES
|
Branch length in initial trees
|
O
|
0.1
|
This uses an equilibration (or burn-in) period
of 1,000,000 MCMC steps, after which
1000 configurations
are sampled in intervals of 1000 MCMC steps,
amounting to 1,000,000 MCMC steps altogether.
These values are over-cautious and can possibly
be considerably reduced.
During equilibration, the parameters of the proposal
distribution are adjusted in intervals of 100 MCMC steps.
We sample the transition-transversion ratio,
the nucleotide
stationary frequencies and the recombination parameter
lambda from the posterior distribution,
and apply the
PAR version of simulated annealing.
The initial branch length should be set to the
average value of the global maximum likelihood tree,
0.1 in this case.
For further information on the parameters, click
here.