moire 3.7.0

New Features

• run_mcmc(seed =) sets the sampler seed. NULL draws from the current R RNG a value that still fits after per-chain offsets. Independent MCMCs are spaced by the number of parallel-tempering replicas. The seed used is returned as $seed (printed if the run errors), with offsets as $chain_seeds. Replay needs the same sampler settings; num_cores does not affect the draws.

Other Changes

• Every exported function now has runnable examples and documents its return value, in preparation for CRAN submission.
• Corrected the swapped epsilon_pos / epsilon_neg descriptions in simulate_observed_allele().
• parallel is now declared in Imports; markdown was dropped from Suggests as nothing used it.
• Parallel tempering temperature adaptation now interpolates temperature as a function of cumulative communication barrier with a monotone cubic Hermite scheme (src/monotone_interpolator.h), replacing the bundled third-party spline.h and its root solving. Adapted ladders differ slightly from earlier versions; if no barrier has been observed yet the update is skipped.

moire 3.6.0

New Features

• Richer Initialization failure diagnostics in run_mcmc()
  • Failures now raise a classed moire_initialization_failure condition with structured $diagnostics (Failure-locus counts, classification, examples)
  • Stop message distinguishes a Consistent failure cause from a Hard starting set and includes short rule-based guidance
  • Inspect with tryCatch(run_mcmc(...), error = function(e) e$diagnostics)

Bug Fixes

• Treat non-finite genotyping log-likelihood (NaN or ±Inf) as an Ill-conditioned start during Initialization retries (previously only NaN)
• Do not initialize sample COI below the maximum number of alleles already observed in that sample
• Fix latent-genotype false-positive bound used when sampling starting genotypes (max_fp now follows 1 ≤ |latent| ≤ COI instead of fn / 2)
• Stabilize transmission coverage probability for large latent genotypes
  • Adaptive prob_any_missing: float inclusion-exclusion for small k, exact closed form when n == k, and EGF product for large k
  • Guard log(1 - pam) against float inclusion-exclusion overshoot

Other Changes

• Raised R dependency to R (>= 4.1.0) (package already uses |>)

moire 3.5.0

New Features

• Implemented maximum runtime functionality (#34)
  • The user can now specify a maximum runtime for the MCMC. After the maximum runtime is reached, the MCMC will stop and the current state of the MCMC will be returned.
  • Enabled by setting max_runtime = {time in minutes} in run_mcmc()

Bug Fixes

• Fixed handling of NA values in allele frequency vector
  • NA values are now replaced with 0
  • Warning message added to alert users of potential MCMC chain issues or loci lacking diversity (#20)
• Fixed memory crash in prob_any_missing() function (#38)

Other Changes

• Modified data loading to ungroup input data when loading long form (#37)

• Updated citation and manuscript information

• Added OpenMP detection utility

• Implement logging of individual sample log likelihoods

  • May be useful for diagnosing issues with particular samples, such as those that come from different populations

moire 3.4.0

New features

• Implemented logging of latent genotypes (#14)
• Added new datasets: namibia_data and regional_allele_frequencies (#27)

Bug fixes and improvements

• Fixed handling of NA values in allele frequency vector (#20)
  • The code now replaces NA values with 0 and displays a warning message
• Updated vignette with code examples using the new datasets (#27)
• Various documentation updates and minor cleanup

Internal changes

• Modified src/Makevars to include ENABLE_PROFILER flag and link with -lprofiler (#24)
• Created new file src/profiler.cpp for profiling functions implementation (#24)
• Added start_profiler() and stop_profiler() functions for performance profiling (#24)

moire 3.3.2

Minor bugfix that corrects an issue with temperature gradient tuning when using parallel tempering.

moire 3.3.1

Minor bugfix that corrects an issue with parameter logging when not using parallel tempering.

moire 3.3.0

This is a minor revision that greatly improves the speed of the MCMC computationally, various bug fixes, and improvements to numerical stability when starting the MCMC.

moire 3.2.0

This is a minor revision that fixes bugs in the adaptive temperature gradient approach and changes the default priors on false positive and false negative rates.

• New default priors
• Fixed bug in adaptive temperature gradient approach
• Numerical stability improvements
• New summarization functions
  • calculate_med_allele_freqs() calculates median allele frequencies
  • plot_chain_swaps() creates diagnostic plots for chain swaps when using parallel tempering

moire 3.1.0

This is a minor revision that introduces new functionality to improve mixing of the MCMC. This also updates the required version of R to 4.0.0. and C++ to C++17.

• Added option to specify prior on within-host relatedness
• Added adaptive approach to tune temperature gradient used during parallel tempering
• Various bugfixes and improvements

moire 3.0.0

This is a major revision to moire, introducing a simplified API and functionality to infer within host relatedness and effective MOI. This release also introduces a parallel tempering based approach that leverages OpenMP, greatly improving mixing of the MCMC.

moire 2.2.0

• Added support for running multiple chains simultaneously, then pooling output
• fix bug with missing data
• various other bugfixes and improvements

moire 2.1.0

• Implemented a new error model that removes sensitivity to total number of alleles at a locus
• Removed option to marginalize out latent genotypes below some complexity threshold
• bug fixes, documentation improvements

moire 2.0.1

• Minor bugfix when sampling that caused computational slow down

moire 2.0.0

• Underlying model no longer uses pseudo marginal MH algorithm. Instead, model is augmented with latent genotypes above some user defined complexity level which are then sampled. Loci below the user defined complexity level have the latent state fully marginalized out.
• Removed dependency on GSL
• various bugfixes and speedups

moire 1.1.1

• fixed overflow bug that would dramatically increase computational costs
• fixed bug in sampling complexity of infection where accepted updates weren’t being recorded in some cases

moire 1.1.0

• Added several new common functions for analyzing data
• Added functions to import data in common formats to the format required
• Made errors an independent parameter across samples
• Changed error model to no longer depend on underlying number of strains contributing alleles
• Removed multiple chain implementations. If multiple chains are desired, use multiprocessing
• Added progress bar for duration of MCMC

moire 1.0.0

• Initial release of moire.