beautier allows to specify an inference model, from which, together with a DNA alignment, a posterior (holding a distribution of phylogenies and jointly-estimated parameter estimates) can be inferred.
The inference model entails the evolutionary model used in inference, as well as some settings to do the inference.
This vignette shows the options how to specify an inference model, and which options are possible.
Now, we’ll look at the default inference model to get an idea of what an inference model entails.
Creating a default inference model is easy:
inference_model <- create_inference_model()
An inference model is a list, that uses the BEAST2 defaults. Here are the elements of that list:
names(inference_model) #>  "site_model" "clock_model" "tree_prior" #>  "mrca_prior" "mcmc" "beauti_options" #>  "tipdates_filename"
As we can see, an inference model entails these elements, which we will cover below in more detail:
The site model entails how the alignment changes over time. Currently,
beautier supplies a gamma site model. One element of a site model, for DNA/RNA, is the nucleotide substitution model (‘NSM’).
Due to historical reasons,
beautier confuses the site model and NSM:
beautier has no functions with the word ‘nucleotide substitution’ (nor
nsm) `in it. Instead, it is as if these are specific site models.
To see the available site models, use
?create_site_model to see a list, or use:
get_site_model_names() #>  "JC69" "HKY" "TN93" "GTR"
The simplest NSM is the JC69 NSM, which assumes all nucleotides are substituted by one another at the same rate. As an example, to use a gamma site model with the JC69 NSM model in inference:
The clock model entails how the mutation rates differ over the branches.
To see the available site models, use
?create_clock_model to see a list, or use:
get_clock_model_names() #>  "relaxed_log_normal" "strict"
The simplest clock model is the strict clock model, which assumes all branches have the same mutation rate. As an example, to use a strict clock model in inference:
The tree prior is the tree model used in inference. It is called ‘tree prior’ instead of ‘tree model’, as this follow the BEAUti naming. The tree model specifies the branching process of a tree.
To see the available tree models, use
?create_tree_prior to see a list, or use:
get_tree_prior_names() #>  "birth_death" "coalescent_bayesian_skyline" #>  "coalescent_constant_population" "coalescent_exp_population" #>  "yule"
The simplest tree model is the Yule (aka pure-birth) tree model, which assumes that branching events occur at a constant rate, and there are no extinctions. As an example, to use a Yule tree model in inference:
With the MRCA (‘Most Recent Common Ancestor’) prior, one can specify which tips share a common ancestor.
# The alignmet fasta_filename <- get_beautier_path("anthus_aco.fas") # The alignment's ID alignment_id <- get_alignment_id(fasta_filename) # Get the first two taxa's names taxa_names <- get_taxa_names(fasta_filename)[1:2] # Specify that the first two taxa share a common ancestor mrca_prior <- create_mrca_prior( alignment_id = alignment_id, taxa_names = taxa_names ) # Use the MRCA prior in inference inference_model <- create_inference_model( mrca_prior = mrca_prior )
The MCMC (‘Markov Chain Monte Carlo’) specifies how the inference algorithm does its work.
The available MCMC’s can be found using
?create_mcmc and are:
create_mcmc: regular MCMC
create_test_mcmc: shorter regular MCMC, to be used in testing
create_ns_mcmc: MCMC to estimate a marginal likelihood using nested sampling
The BEAUti options entail version-specific options to store an inference model as a BEAST2 XML input file.
The available BEAUti options can be found using
?create_beauti_options and are:
create_beauti_options_v2_4: BEAUti v2.4
create_beauti_options_v2_6: BEAUti v2.6
Using a specific version for an inference:
A tipdates filename is an experimental feature for tip-dating:
inference_model <- create_inference_model( tipdates_filename = get_beautier_path("G_VII_pre2003_dates_4.txt") )
The tipdates filename and the alignment must be compatible. Here is an example: