Apr 28, 2013

Bayesian phylogenetic analysis of Grammitidaceae

Materials for reconstruction of Grammitidaceae phylogeny (part of my current research). 
First attempt with cpDNA (rbcL, atpB, trnL-F) for 217 samples from GenBank using Bayesian inference. 
Tools: MrBayes 3.2.1 x64 + BEAGLE, jModelTest 2.1.3, Geneious 6.0.6 Trial, FigTree 1.4.0, Tracer 1.5, (+ BEAST for molecular clock tests, not included to this post).
Options: models selected with Akaike information criteria (atpB = TVM+I, rbcL = TrN+I, trnL-F = K80+I); 15·10⁶  generations in Markov chain Monte Carlo sampling, 10% burn-in, 27000 resulting trees, tree selection criteria Posterior Probability > 0.5.
Resulting tree looks like nice, but there are some basal polytomy and many of basal nodes are weakly supported. Reconstruction for New World genera looks compartible with published papers, but topology for genera from Tropical Asia seems to be an atrifact of poor sampling (according to M.Sun. comments). Not so bad for first attempt, after all.
Some pictures below.

Consensus tree.


Unrooted graph and some Tracer output





Allignment in Genious 6, using Muscle algorithm and manual ajustment.

P.S. Another consensus tree, afeter some severe problems in trnL-F allingment were resolved.


No comments:

Post a Comment