-An example template that utilizes a skygrid coalescent prior as well as the new classes and likelihood discussed
-above can be downloaded from here along with a
-nexus tree containing
-1,000 SARS-CoV-2 taxa whose genomes were downloaded from GISAID and COG-UK.
-An acknowledgement table for the genomes from GISAID is available
-here
-
+
+
+We should now see our tree listed as partition.
+
+
+
+From here most of the set up follows a standard BEAST analysis.
+
+Under the Tips pannel, we can parse tip dates from these taxa labels.
+
+
+
+In this initial analysis we are not including a phylogeographic component, but we could add such an analysis by
+importing trait data in the Traits pannel. Since we are just scaling a pre-estimated divergence tree into time
+we do not need a Sites model, and can skip right to the Clocks panel.
+
+In this case we are using a strict clock, but this approach is compatible with any non-hamiltonian clock model.
+The gradients needed for HMC operators have not yet been implemented for this data type.
+
+
+
+The magic happens in the next panel - Trees. Here we select both the tree prior and Thorney tree model.
+For this analysis we will use the bayesian skygrid, with a cut off of three quarters of year, and nearly weekly gridpoints (39 parameters).
+
+On the bottom half of the panel we select `ThorneyBEAST` as the 'Tree as data model', and set the scaler to 29903 - the length of the alignment used to build the tree.
+The number of mutation on each branch will be calculated by multiplying branchlengths in the input tree by this number and rounding to the nearest whole number.
+
+
+
+We can then adjust priors an operators as we see fit in the Priors and Operators panel.
+
+*NB: the default ctmc reference prior for evolutionary rate can add some computational burden on large trees.
+If a gamma(0.001,1000) distribution is appropriate, it may provide some computational speed ups on large datasets.*
+
+
+
+## Operators
+
+One of the advantages of this approach over fixed tree approach is that we are able to sample polytomy resolutions
+from the posterior. Standard tree operators are used to do this, but they are constrained to respect the clades present
+in the input tree. If the input tree is fully resolved these operators should be removed.
+
+
+
-The xml can be generated with
-```bash
-beastgen -date_order -1 -date_prefix "|" -date_precision -D "outputStem=1K-SARS-CoV-2" SG-thorney.template 1K_SARS-CoV-2.nexus 1K_SARS-CoV-2_SG-thorney.xml
-```
-
-The xml file can also be downloaded directly from here
-
+We can then generate and run the xml as usual.
-The template, `SG-thorney.template`, is small and easy to edit. It can also be used to generate similar analysis on much
-larger datsets without the tedious task of editing a large xml. The operator weights are experimental and may need to be
-adjusted.
+
-A brief introduction to BEASTGen can be found [here](http://beast.community/beastgen). The Apache FreeMarker template
-engine is quite powerful, which makes BEASTGen a useful tool for managing large datasets and analyses.
## References
Zuckerkandl E , Pauling L. 1962. Molecular disease, evolution, and genic heterogeneity. In: Kasha M , Pullman B, editors. Horizons Biochem. New York: Academic Press. p. 189–222