multilocus *BEAST analysis and clock rate questions

[Herman Mays]

I'm attempting a reanalysis of an analysis done in BEAST v1.7 and can not replicate the results.

Basically we have a relatively small genus-level phylogenetic analysis with 5 nuclear introns and 3 mtDNA coding genes and we want to produce some rough estimates of divergence time at each of the nodes in the species tree. A very standard analysis using *BEAST. We have an prior estimate of a divergence rate of around 2.1% divergence per site per Ma for one of the mtDNA genes (cytb for birds) and we are attempting to measure the rest of the loci relative to that rate. This means that in the clock tab we leave the rates set to 1.0 for all the genes except for cytb and set that one to 0.0105 (1/2 the divergence rate). Do we then click 'estimate' for the known cytb partition and leave the rest unchecked or visa versa, or do we check all and specify some narrow bounds for the known cytb rate? (I've seen the latter suggested on the beast-users Google Group before by Alexei Drummond) We have some prior clock rate tests that led us to set relaxed lognormal clocks for the mtDNA loci and one of the introns and strict clocks for the rest of the introns and I've stuck to that in my attempts at a reanalysis. Also the three mtDNA loci are linked to the same tree.

I can run the analysis in BEAST2 but the node heights are always substantially off from the original analysis in v1.7 (I'm assuming based on what I described above this will return heights in Ma). I've tried to recreate the same analysis in BEAST v1.8 and BEAUTi fails to create a proper xml file everytime (something about the .taxa file already exists). I would like to recreate the same result in BEAST2 if possible? Or at least see that I can get close to our original v1.7 analysis divergence times in a BEAST2 analysis.

This is all precursor to re-doing our anlaysis with codon partitions for the mtDNA genes (suggested by a reviewer) but I was hoping to recreate the same, or close, result in either v1.8 or preferably BEAST2 before making this change.

I've attached the original xml file from the v1.7 analysis and the original species tree from that analysis. I'm a newbie to BEAST so any help/advice would be greatly appreciated!!!!

Thanks!
Herman