Skip to content

Latest commit

 

History

History
74 lines (52 loc) · 1.34 KB

workflow.md

File metadata and controls

74 lines (52 loc) · 1.34 KB

Workflow

All steps are executed in a linux environment as defined in Dockerfile.

  • prepare the data, download original data (use the Makefile)

    make download
make preprocessing
make language-data

  • create character matrix files

    > cd code
> python get_nexus_from_cognate_classes.py
> python get_nexus_from_correspondences.py  
> python get_phylip_from_cognate_classes.py  
> python get_phylip_from_correspondences.py

  • create goldstandard trees from Glottolog

    > python get_glottolog_trees.py

  • create MrBayes scripts and combined nexus files

    > cd code
> julia create_mb_scripts.jl

  • run MrBayes scripts (to be run on a server with at least 100 cores)

    > cd mrbayes
> bash run_mrbayes.sh

  • check convergence of MrBayes runs

    > cd ..
> julia check_convergence.jl

  • extract posterior tree samples

    > Rscript create_posterior_samples.r

  • compute GQD for Bayesian trees

    > julia get_qdists_mb.jl

  • extract $\alpha$-values for Bayesian analysis

    > julia evaluate_alpha.jl

  • run maximum likelihood experiment (script includes execution of RAxML-NG, GQD computation and extraction of $\alpha$-values)

    > cd ml/
> python ml_experiment.py