Skip to content

Releases: MuteJester/LZGraphs

v3.1.0

02 Jun 17:22

Choose a tag to compare

v3.0.2

04 Apr 12:17

Choose a tag to compare

Full Changelog: v3.0.1...v3.0.2

v3.0.1

29 Mar 14:44

Choose a tag to compare

Full Changelog: v3.0.0...v3.0.1

LZGraphs v3.0.0 — Complete C Rewrite

26 Mar 14:03

Choose a tag to compare

LZGraphs v3.0.0

Major release: the entire library has been rewritten as a C core with CPython bindings.

Highlights

  • Unified LZGraph class — replaces AAPLZGraph, NDPLZGraph, NaiveLZGraph with a single class using variant='aap'|'ndp'|'naive'
  • C performance — graph construction, simulation, and analytics run at native speed (80ms for 5K sequences, ~5K simulations/sec)
  • LZ76 dictionary constraints — per-walk dynamic dictionary ensures every simulated sequence is a valid LZ76 decomposition
  • Sentinel model@/$ sentinels eliminate stop probabilities and end-of-sequence ambiguity

New Capabilities

  • lzpgen() — exact generation probabilities (single or batch)
  • simulate() — LZ-constrained sequence generation with optional V/J gene constraints
  • Hill diversity numbers — D(0), D(1), D(2), and arbitrary orders via exact-probability importance sampling
  • Occupancy predictions — predicted richness, overlap, and sharing spectra
  • Graph algebraunion (|), intersection (&), difference (-), weighted_merge
  • Bayesian posteriorsposterior() for personalized repertoire models
  • ML feature extraction — reference-aligned vectors, mass profiles, 15-element stats vectors
  • CLI tool (lzg) — 10 subcommands for terminal-based analysis
  • Binary .lzg format — fast serialization with CRC-32C checksums

Documentation

  • 39-page documentation site with tutorials, concept guides, and API reference
  • Empirical comparison with IGoR, OLGA, and SONIA
  • Performance benchmarks with real data

Breaking Changes

  • AAPLZGraph(data)LZGraph(sequences, variant='aap')
  • walk_probability(seq)lzpgen(seq)
  • random_walk() / genomic_random_walk()simulate(n)
  • get_posterior()posterior()
  • pandas is no longer a dependency
  • Serialization uses .lzg binary format (not pickle)

See the migration guide for details.

Install

pip install LZGraphs==3.0.0

Full documentation: https://MuteJester.github.io/LZGraphs/

v2.3.1

07 Mar 15:29

Choose a tag to compare

Release v2.3.1

v2.3.0

07 Mar 14:57

Choose a tag to compare

Release v2.3.0

v2.2.0

26 Feb 16:15

Choose a tag to compare

Release v2.2.0

v2.1.2

12 Feb 08:01

Choose a tag to compare

Release v2.1.2

v2.1.1

09 Feb 12:56

Choose a tag to compare

Release v2.1.1

v2.1.0

08 Feb 13:23

Choose a tag to compare

Release v2.1.0