Skip to contents

Changelog

Source: NEWS.md

BioGA 0.99.18

Major changes

  • Rewrote core algorithm as a correct NSGA-II implementation: strict domination, full non-dominated front construction, crowding distance, and NSGA-II environmental selection for replacement (new src/nsga2.h).
  • Added a hybrid representation via representation = c("real", "binary"). Real mode minimises squared error (f1) and mean absolute value (f2); binary mode minimises the negated mean ANOVA F-statistic of selected genes (f1) and the fraction of selected genes (f2) using class labels.
  • Added idiomatic R wrappers bioga() and bioga_fitness().
  • crossover_cpp now generates both complementary SBX children and handles a single parent without hanging.
  • mutation_cpp uses degree-based network damping (bounded, non-negative) for the real representation and bit-flip mutation for the binary representation.
  • replacement_cpp now performs NSGA-II environmental selection; num_to_replace = 0 returns the population unchanged.
  • bioga_main_cpp returns population, fitness, history, best, best_fitness, pareto_population, and pareto_fitness, and accepts a labels argument for supervised objectives.

Documentation

  • Corrected scientific statements in the vignettes (removed unverified claims about tools, benchmarks, databases, and convergence guarantees; aligned formulae with the implementation).
  • Lowered Depends to R (>= 4.3).

Testing

  • Rewrote the test suite for the new API and behaviours.

BioGA 0.99.17

Added

  • Demo vignette for BioGA.
  • Mathematical background vignette for BioGA.

Updated

  • Vignettes with new examples and explanations.
  • selection_cpp for improved performance and bug fixes.

BioGA 0.99.16

Testing improvements

  • Tests for crossover_cpp: new parameters (crossover_rate, eta_c), SBX offspring range validation, edge cases.
  • Tests for selection_cpp: multi-objective fitness matrix support, parent identity verification, single-individual edge case.
  • Tests for mutation_cpp: new parameters (iteration, max_iterations, network), zero-rate preservation, network constraint extremes.
  • Tests for replacement_cpp: new signature with fitness matrices, elite preservation, zero-replacement edge case.
  • Tests for initialize_population_cpp: seed reproducibility, clusters parameter, minimal input edge case.
  • Tests for bioga_main_cpp: output structure, dimension verification, optional parameters, minimal input.
  • Package integration test: export verification, full workflow execution.

BioGA 0.99.15

Testing improvements

  • Updated evaluate_fitness_cpp tests for multi-objective matrix output.
  • Added non-negative fitness checks.
  • Added minimal input edge case.
  • Removed redundant tryCatch.

BioGA 0.99.14

New feature

  • Main GA loop (bioga_main_cpp) integrating all components with multi-objective optimization and parallelization.

BioGA 0.99.13

Improvements

  • Biologically informed initialization using gene clusters.
  • Reproducible random seed handling.

BioGA 0.99.12

Maintenance

  • Code cleanup and indentation in replacement.cpp.

BioGA 0.99.11

Improvements

  • Elitism to preserve best individuals.
  • Diversity-based replacement to avoid premature convergence.
  • Multi-objective fitness matrix support.

BioGA 0.99.10

Improvements

  • Adaptive mutation rate based on iteration progress.
  • Optional gene network constraints for biologically relevant mutations.

BioGA 0.99.9

Improvements

  • NSGA-II non-dominated sorting for multi-objective selection.
  • Tournament selection for diversity.

BioGA 0.99.8

Improvements

  • Simulated binary crossover (SBX) for better exploration.
  • Adaptive crossover rate based on population diversity.
  • Configurable crossover rate and distribution index.

BioGA 0.99.7

Improvements

  • Multi-objective fitness evaluation (expression difference and sparsity).
  • Parallelized computation via RcppParallel.
  • Flexible objective weighting via weights parameter.

BioGA 0.99.6

Maintenance

  • Added authors and date to vignettes.
  • Set minimum R version to 4.4.

BioGA 0.99.4

Maintenance

  • Updates requested by Bioconductor peer review.

BioGA 0.99.2

Bug fixes

  • Fixed Git tracking of .Rproj system file.

BioGA 0.99.0

Initial release

  • Added NEWS.md to track changes.