v0.9.2

Release notes:

• Wrap N and K values in Val in count_kmers method to improve type stability and unnecessary allocations. See BioJulia/BioMarkovChains.jl#25

v0.9.1

Release notes:

• Explicitly index the values array of a KmerArray in count_kmers! methods where performance is critical.

v0.9.0

Release notes:

• Rename KmerVector to KmerArray, with K dimensions, with each dimension having size N, instead of being a vector of length N^K.
• Make KmerArray{N, K} a subtype of StaticArrays.StaticArray{NTuple{K, N}}
• Change indexing of KmerArray to be offset by 1. Linear indexing is done with indices 0 through N^K-1, and cartesian indexing is done on each axis with values ranging from 0 through N-1.
• Remove AbstractKmerArray types
• Remove KmerColumns, KmerRows. Columns or rows of a matrix can now simply be wrapped by KmerArray to count K-mers of sequences. E.g. count_kmers!.(KmerArray{4, 2}.(eachcol(zeros(4^2, n))), sequences)
• K-mers of arbitrary sequences can now be counted by simply defining the method axis_index(::KmerArray, ::eltype(sequence))

v0.8.1

Release notes:

• Changed the count_kmers methods for Vectors of sequences to work with any AbstractVector.
• Removed CUDA from weakdeps (was accidentally left in after the extensions and compat entries had been removed in v0.8.0).

v0.8.0

Release notes:

• Moved functionality for k-mer counting on GPU to the gpu branch, as the code haven't been properly tested, and can not be tested with the GitHub CI.
  • Removed compat entry for CUDA
  • Removed the BioSeqCUDAExt extension
  • Removed the CUDAExt extension
• Changed the meaning of the D type parameter such that KmerColumns means KmerVectors{1} and KmerRows means KmerVectors{2}. This change is purely semantic and does not change any structure or functionality.
• Removed get_S and get_k functions.
• Added usage examples to README.md

v0.7.0

Release notes:

• count_kmers! method for amino acid sequences using a non-ambiguous subset (first 20) of BioSequences.AminoAcidAlphabet.

v0.6.1

Release notes

• Added a method for doing k-mer counting of nucleotide sequences with BitVectors, where the value of each bin ends up representing whether that k-mer occurred in the sequence an odd or even number of times.
• Removed T <: Real type requirement
• Tweaked some docstrings
• Added the "return" keyword at the end of all non-one-liner functions
• Slightly better (although incomplete) documentation

v0.6.0

Release notes:

• Removed "Count" from type names. See #28

v0.5.1

Release notes:

• Now supporting k-mer counting of LongSubSeq-type sequences.
• Now supporting k-mer counting of RNA sequences (DNA method was generalized to nucleotide)

v0.5.0

Release notes:

• Overhauled type hierarchy
• Changed args and kwargs of constructors and count_kmers methods.
  • count_kmers now takes a sequence/vector of sequences, and type parameters as arguments.
  • count_kmers! now takes a KmerCountVector/KmerCountVectors and a sequence/vector of sequences.
  • Alphabet size S can be specified if the alphabet_size function is not defined for a particular sequence type.
  • Default count_kmers! method takes a RefValue{Vector{<:Integer}}, and is meant for testing purposes.
  • count_kmers has been generalized.