EvolutionModels.jl

A Julia package implementing continuous-time Markov models for molecular sequence evolution. Provides both parametric nucleotide models (JC69, HKY85, GTR) and empirical amino acid models (WAG, LG) with simulation and likelihood computation capabilities.

⚠️ Note: This package is currently under development and extension (loaded with FASTX and Optim) is experimental. Use with caution in production environments.

Features

• Nucleotide substitution models: JC69, HKY85, GTR
• Amino acid substitution models: WAG, LG
• Sequence evolution simulation
• Likelihood computation
• Distance matrix computation with:
  • Analytical formulas for JC69, K2P, HKY85 (fast, matches common software)
  • ML estimation for GTR and other models (more accurate but slower)
• Integration with BioSequences.jl
• FASTA support via FASTX.jl extension

Installation

using Pkg
Pkg.add("EvolutionModels")

Basic Usage

using EvolutionModels
using BioSequences

# Create JC69 model
model = create_model(JC69Model, 0.1)

# Simulate evolution
seq = dna"ATCG"
evolved = evolve_sequence(model, seq, 1.0)

# Compute likelihood
logL = sequence_likelihood(model, seq, evolved, 1.0)

Computing Distances from FASTA Alignment

The package can compute pairwise evolutionary distances from multiple sequence alignments in FASTA format. This functionality requires both FASTX.jl and Optim.jl:

using EvolutionModels
using FASTX
using Optim

# Read alignment
seqs = read_alignment("alignment.fasta")

# Create model and compute distances
model = create_model(JC69Model, 0.1)

# Compute distances using default method (analytical formula for supported models)
result = compute_distances(model, seqs)

# Or explicitly specify the method:
result_analytical = compute_distances(model, seqs, method=:analytical)  # Fast, matches FastTree
result_ml = compute_distances(model, seqs, method=:ml)                 # Slower but more accurate
result_auto = compute_distances(model, seqs, method=:auto)            # Choose best method for model

# Print the distance matrix
print_distance_matrix(result)

The resulting distance matrix contains evolutionary distances computed using either:

• Analytical formulas (for JC69, K2P, HKY85) that match common phylogenetic software
• Maximum likelihood estimation (for GTR and other models) that may be more accurate for highly diverged sequences

These distances can be used for:

• Phylogenetic tree reconstruction
• Sequence clustering
• Evolutionary rate estimation

License

MIT