Kalign 3.5.1

Bugfix release following v3.5.0. All new features from 3.5.0 are included — this release fixes issues found during the initial release.

What's new in 3.5 (since 3.4)

• Three alignment modes: default (balanced), fast (same speed as v3.4), precise (ensemble, highest accuracy)
• Ensemble alignment (--precise or --ensemble N): multiple runs with varied parameters, POAR consensus
• Per-column confidence scores from ensemble mode
• Alignment refinement (--refine): iterative realignment for improved accuracy
• PFASUM substitution matrices (--type pfasum, pfasum43, pfasum60)
• Variable Scoring Matrix (VSM): adaptive scoring based on sequence divergence
• Python package on PyPI: pip install kalign-python

Fixes in 3.5.1

• Fix memory leak in build_tree_from_pairwise (affected realign and ensemble modes)
• Move seaborn from core dependency to optional (pip install kalign-python[analysis])
• Benchmark workflow handles unavailable datasets gracefully
• Code formatting fixes

Install

Python

pip install kalign-python

CMake (from source)

mkdir build && cd build
cmake ..
make
make test
sudo make install

Zig (cross-compilation)

zig build

Homebrew (macOS)

brew install kalign

Quick start

Command line

kalign input.fasta -o aligned.fasta

# Fast mode (same as v3.4)
kalign --fast input.fasta -o aligned.fasta

# Precise mode (ensemble, highest accuracy)
kalign --precise input.fasta -o aligned.fasta

Python

import kalign

# Default mode
aligned = kalign.align(["MKTAYIAK...", "MKTAYIAKQ..."])

# Precise mode (ensemble)
aligned = kalign.align(sequences, mode="precise")