Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project follows Semantic Versioning.

[Unreleased]

Added an opt-in RL/GFlowNet research starting path: docs/rl-research-start.md plus runnable examples/rl/ toy scaffold. No Rust API, dependencies, model weights, or default behavior changed.

v1 beam-search optimizer with configurable CAI, GC%, U%, and repeat-penalty objective.
CLI flags for --strategy, --beam-width, --target-gc, --target-u, and objective weights.
GC%, U%, and repeat-penalty metrics in stdout output.
Public RP mini-benchmark in examples/benchmark/ with four protein FASTA targets.
Golden CLI tests and beam-search fidelity tests.
JOSS paper scaffold and code of conduct.

Replaced the LinearDesign-derived bundled codon usage table with a Kazusa Codon Usage Database Homo sapiens export normalized from raw counts.
Documented CAI table dependency and codon table provenance.
--lambda now emits an explicit compatibility warning when supplied.

Greedy highest-frequency codon optimizer for protein-faithful human mRNA sequences.
Codon Adaptation Index (CAI) computation using the Sharp and Li method.
Protein sequence validation and RNA translation fidelity verification.
Protein FASTA input and pure mRNA FASTA output.
LinearDesign-style stdout for human-readable benchmarking logs.
Human codon usage frequency table bundled under data/.
Unit and integration tests for parser, optimizer, CAI, output formatting, and CLI behavior.
MIT license.