BAli---Phy

Release Notes

See NEWS file for more details, and for older changes.

• Speed & Memory
  • Speed: 0-300% faster.    Memory: 0-300% less memory.
  • Site-compression for fixed-alignment partitions.
• Alignments
  • Allow analyzing 1-sequence and 2-sequence alignments.
  • [3.5] Automatically reconstruct ancestors for summary alignment.
• Large Trees
  • Compute likelihoods for very large trees (likelihood rescaling).
  • New treelength prior: don't prefer huge treelengths for trees with lots of leaves.
• General
  • Add unit tests and likelihood testsuite.
  • [3.4] Allow selecting character sets from a file: sequences.fasta:100-240,300-900
  • [3.4] Allow recovering from initial -Infinity log-likelihood in MCMC.
  • [3.4] Install package manager bali-phy-pkg.
• Models
  • Priors:
    • Explicit priors, including default priors
    • Sample initial branch-lengths from prior.
    • substitution model: -S 'hky85[kappa=2]'
    • insertion/deletion model: -I 'rs07[log_rate~laplace[-4,0.707]]'
    • branch lengths: -B '~iid[num_branches[T],gamma[0.5,div[2,num_branches[T]]]]'
    • partition-specific scale factors: -R '1,3:~gamma[0.5,2]'
  • New syntax:
    • let[m=hky85,mixture[List[m,m,m+Rates.gamma]]] (user-defined variables)
    • [3.3] function[w,gy94[omega=w]] (functions)
  • New models:
    • mixture[List[hky85,jc69]] (rates-across sites mixtures)
    • hky85+multi_rate[beta[2,3]] (rates-across sites with any distribution)
    • [3.3] +fe for equal frequencies.
    • [3.5] +mut_sel or +mut_sel_aa to add selection on a model.
  • RNA stem (16-state) models
    • [3.4] RNA models for fixed alignments. (Preliminary, because you have to pair the letters manually.)
    • [3.4] RNA stem models: RNA.16a, gtr_sym+x2_sym+f, gtr+x2+mut_sel
  • Codon and triplet models
    • [3.3] Improved codon models: gy94, mg94, f1x4, f3x4, gy94_ext, mg94_ext, x3, dNdS, etc.
    • [3.3] dNdS mixtures with mg94 and fMutSel: m3[function[w,fMutSel[omega=w]]]
    • [3.4] gtr+x3+dNdS+mut_sel (syntax for building up codon models piecewise)
  • Heterotachy/Covarion models
    • [3.5] Tuffley-Steel (1998) ON/OFF model.
    • [3.5] Huelsenbeck (2002) ON/OFF + Gamma rates-across-sites model.
    • [3.5] Galtier (2001) Gamma rate-switching model.
    • [3.5] Wang et al (2007) ON/OFF + Gamma rate-switching model.
• Summarization tools
  • Show 2D & 3D topology convergencs figures (MDS).
  • [3.4] Reorganize logged statistics.
  • [3.4] Compute alignment summaries over all runs, not just the first one.
  • [3.5] Rewrite bp-analyze in python.
• Help
  • Help for models, functions, distributions, and commands via bali-phy help topic.
  • [3.3] Rewrite documentation for substitution models and insertion/deletion models.
• Tools
  • [3.4] cut-range: allow reading alignments from multiple files.
  • [3.4] alignment-distances: new tool, add accuracy and recal metrics
  • [3.4] tree-tool: scales trees, prunes trees, computes diameter, etc.
  • [3.4] alignment-thin: clean up options and man pages
  • [3.4] bali-subsample: rename from subsample to avoid conflict with phylobayes.
  • [3.5] extract-ancestors: Extract and name ancestral sequences.
  • [3.5] summarize-ancestors: Construct summary alignment with ancestral sequences.
  • [3.5] tree-tool: add option to remove internal node names.
  • [3.5] Speed up tree drawing in bp-analyze.
• Install
  • [3.4] Transition homebrew package to brewsci/bio.
• Fixes
  • [3.3] Fix incorrect LG model.
  • [3.3] Fix run file for f81 model.
  • [3.3] bp-analyze: parse output files from 3.0-betal.
  • [3.4] Don't crash if the scale is set to a constant (e.g. --scale 1)
  • [3.4] Don't replace ambiguous nucleotides (W,N,etc.) in observed sequences unless given --set infer-ambiguous-observed=true
  • [3.4] Properly handle Newick labels with quotes or underscore.
  • [3.4.1] Fix for sequence names with underscore.
  • [3.4.1] Fix windows binaries.
  • [3.5] Stabilize matrix exponential.
  • [3.5] Fix dirichlet process prior.
  • [3.5] Don't print out Newick trees rooted at a leaf.
  • [3.5] Compute posterior probabilities for MAP and greedy consensus.