Zoe v0.0.28

Release Notes for v0.0.28 - 2026-04-29

Added

• Similar to the OrFail trait for Result, a Fail trait has been introduced to handle errors directly
• Added *_mapped versions for many string/byte search functions, which search a lazily-mapped haystack
• Added as_map accessor to convert ByteIndexMap to ByteMap
• Nucleotides and AminoAcids (and view types) can now be converted into Vec, slices, and Cow using From
• NucleotidesView, AminoAcidsView, and QualityScoresView now have inherent slice methods offering a longer lifetime of the returned view

Changed

• LocalProfiles and SharedProfiles now can hold either a borrowed or an owned sequence, enabling sequences and profiles to be stored in the same struct without becoming self-referential
• CigletIterator now yields empty ciglets when a CIGAR string contains an explicit increment that is 0. A missing increment field ends the iterator early
• find_mapped_match_simd better handles needles with a single repeating byte, but now only supports lane counts greater than 2

Fixed

• Fixed incorrect ranges returned by Alignment::to_reverse and Alignment::make_reverse
• Methods in StatesSequence and StatesSequenceMut now properly skip empty Ciglets (with an increment of 0)
• Pre-alignment filter sneaky_snake now properly handles inputs with length differences greater than provided threshold.
• Subsequent calls to next_back in CigletIterator now properly return None after invalid state is reached
• Equality comparisons between Cigar and AlignmentStates now properly return false for an overflowing increment
• Fixes the potential of overflow in hamming

Zoe v0.0.27

Release Notes for v0.0.27 - 2026-04-15

Added

• Added ByteMap for defining and representing maps from u8 values to other u8 values (e.g., for sanitization/recoding), as well as many pre-defined maps
• Added ByteValidator for defining alphabets or subsets of valid bytes
• Added SanitizeBase extension trait for u8 to perform DNA validation, recoding, and refinement on a single byte
• Added find_start_codon as well as a modified string search routine supporting lazy transformations of the haystack (find_mapped_match_simd)
• Added to_aa_iter_exact and to_aa_iter_exact_with for amino acid translation that does not include a partial codon for codons with less than 3 bases at the end of a sequence
• Added find_next_aa and find_next_aa_in_frame to RangeSearch
• Added find_byte to ByteSubstring (for searching strings and for use with RangeSearch)
• Added as_mut_vec to AlignmentStates, Nucleotides, and AminoAcids to enable custom editing
• Added push, starts_with, and ends_with to Nucleotides and AminoAcids
• Implemented Extend for Nucleotides and AminoAcids
• Added AcgtNoGapsUc strategy for retaining/recoding DNA

Changed

• Renamed from_filename to from_path for Zoe's readers, deprecating the old functions
• Renamed with_file_context to with_path_context for errors, deprecating the old functions
• Renamed is_valid_codon to is_resolvable_codon in CodonExtension
• is_amino_acid now behaves the same as is_known_amino_acid, and is_known_amino_acid is removed
• RangeSearch now contains a generic with the original type to allow search methods to restrict the types of data they are used on
• StdGeneticCode now translates codons with mixed - and . bytes to .
• OrFail is now only implemented on errors with a 'static lifetime
• retain_by_recoding now uses ByteMap, and retain_by_validation now uses ByteValidator
• Implemented Copy for several immutable views

Removed

• Removes open_nonempty_file, which produces incorrect results for piped inputs
• itoa is no longer a dependency of Zoe.

Fixes

• Fixes a bug where pipes are interpreted as empty files with FastQReader::from_filename (and similarly for FastaReader and SAMReader)
• Fixes a bug in p_distance_acgt where longer sequences could be normalized incorrectly.
• Fixes a bug in physiochemical distance to ensure invalid sequences are not comparable

Zoe v0.0.26

Release Notes for v0.0.26 - 2026-03-06

Added

• Adds the sneaky_snake pre-alignment filter for imposing an edit distance threshold
• Added convenience methods for checking the variant of SeqSrc
• Implemented to_reverse and make_reverse for MaybeAligned<Alignment<_>>
• Implemented SequenceReadable for profile sets
• Implemented PartialOrd for Alignment and MaybeAligned

Changed

• Development features now only use internal documentation on GitHub
• Generalized sampling methods to accept any random generator (behind rand feature gate)
• Updated to use rand v0.10.0, which includes breaking changes (behind rand feature gate)

Fixed

• Fixed the constructors for SharedProfiles to match those of LocalProfiles

Zoe v0.0.25

Release Notes for v0.0.25 - 2026-02-04

Added

• Adds methods to pad and extend Nucleotides and AminoAcids

Changed

• Zoe's MSRV is now 1.95 (nightly).

Fixes

• Fixes Zoe to use the latest sync'd portable SIMD in Rust nightly 1.95. For earlier versions, try Rust nightly 1.94.

Zoe v0.0.24

Release Notes for v0.0.24 - 2026-01-26

Added

• Stabilized the dev-3pass feature for performing a memory-efficient three-pass local alignment routine, as well as other helper algorithms
• Added WithErrorContext and ResultWithErrorContext traits using the new ErrorWithContext struct for updated error-handling
• Added prepend_state, prepend_ciglet, and prepend_inc_op to AlignmentStates
• Added get_first_codon, get_last_codon, and get_tail_codon to GetCodons trait, with mutable equivalents in GetCodonsMut
• Added CodonExtension trait for [u8; 3] with convenience codon characterization methods
• Added StatesSequenceMut trait and AlignmentStates::as_mut_slice to provide mutable access to ciglets while iterating/processing
• Added ToCigletIterator trait, implemented on structs which can be converted to an iterator of ciglets
• Added LenInAlignment trait, replacing the previous match_len and offering a more flexible API
• Added methods and implementations to iterate over CIGAR views
• Added to_alignment to convert SamData to Alignment
• Added ability to push tags to SamTags
• Added InspectCigar for checking for indels in CIGAR strings or AlignmentStates
• Added VARIANTS constants and index comparison methods for pHMMs (behind dev-phmm feature gate)
• Added row and column accessors to TransitionParams (behind dev-phmm and alignment-diagnostics feature gates)

Changed

• Modified alignment methods on profiles to take a SeqSrc enum as an argument, allowing the returned alignment to properly reflect which sequence is the query and which is the reference
• Renamed alignment methods/functions to use the terminology sw instead of smith_waterman and align instead of alignment
• Alignment methods returning indices or ranges now return specialized structs
• The arbitrary module now uses specification structs to allow various restrictions to be enforced when generating Zoe data types. See the module level documentation for more details
• Updated error handling logic and style for OrFail trait
• Standalone function zoe::data::err::open_nonempty_file is now public
• SharedProfiles no longer holds the sequence as Box<u8>, instead using &'a [u8]
• Condensed the PhmmParamKind variants in the visit_params API (behind dev-phmm feature gate)

Fixes

• Fixed spelling of filter_to_dna_uanligned to filter_to_dna_unaligned in trait ToDNA
• Fixed bug for parsing -128 in a weight matrix using MatParser

Removed

• Removed the match_len method from CIGAR strings in favor of LenInAlignment
• Removed CheckedCigar in favor of LenInAlignment (checked methods are behind the fuzzing feature gate)

Zoe v0.0.23

Release Notes for v0.0.23 - 2025-11-24

Added

• Added nw_scalar_score and nw_scalar_alignment for performing global sequence alignment
• Added random downsampling for iterators of known and unknown lengths in iter_utils::sampling
• Added the method process_results to more robustly work with iterators of results
• Added get_aligned_query, similar to get_aligned_seqs but returning only the query sequence
• Added additional getters and convenience methods on SamTags and SamTagValue
• Added to_eq_x to Alignment and AlignmentStates, which converts M to either = or X in the alignment
• Added SequenceReadable getter trait to record types and sequence types
• Added from_ciglets_unchecked and from_cigar_unchecked to AlignmentStates
• Added decode_iter to KmerEncoder
• Added nw_score_from_path for rescoring a global alignment (behind alignment-diagnostics feature gate)
• Introduced the AlignmentAndSeqs and AminoAcidsIupacX arbitrary wrappers (behind fuzzing feature gate)
• Added visit_params methods to pHMMs as a diagnostic tool for inspecting the parameters that are visited along a given path (behind dev-phmm and alignment-diagnostics feature gates)
• Added the ability to remove layers from a pHMM (behind dev-phmm and alignment-diagnostics feature gates)

Changed

• TryFrom for AlignmentStates now merges adjacent ciglets with the same operation
• An iterator of ciglets now collects into a Result<AlignmentStates, CigarError> instead of AlignmentStates, and performs merging as above
• The view traits have been refactored so that all associated types are defined in the ViewAssocTypes trait
• reborrow_view has been added to allow invariant lifetimes to be shortened in views
• Vec<u8>, &[u8], and &mut [u8] now implement the various view traits
• Pairs of methods accepting encoded and decoded k-mers have been merged
• KmerCounter has had its methods renamed from insert to tally
• Iterators over encoded k-mers from KmerSet now return owned values
• KmerCollectionContains is renamed to FindKmersInSeq
• The kmer module has been restructured to simplify the docs
• Implemented std::fmt::Binary for ThreeBitEncodedKmer, and added it as a trait bound to AnyInt
• sw_score_from_path and sw_scalar_alignment_override are now behind the alignment-diagnostics feature gate
• Modified the behavior of AlignmentArbitrary and AlignmentStatesArbitrary (behind fuzzing feature gate)
• Refactored the organization/visibility of pHMM code; modified SamHmmParser constructors; applied compact bit representation for Viterbi traceback; switched to eager evaluation for next_index and prev_index (behind dev-phmm feature gate)
• The inner fields of SAMReader and FastQReader are now properly private, and SliceRange is properly public

Fixed

• The fuzzing feature now successfully compiles even when dev-phmm is not enabled
• Fixed another NaN error that would occur when certain transitions in a pHMM were zero probability (behind dev-phmm feature gate)

Removed

• Removed checked methods from KmerEncoder (they only provided partial checking and proved not useful)

Zoe v0.0.22

Release Notes for v0.0.22 - 2025-09-25

Added

• Added banded Smith Waterman implementation as a more memory-efficient alternative for long sequence alignment
• Added HeaderReadable getter trait for record types
• Added slice_to_ref_range method for Alignment, in order to slice/clip an alignment to a desired portion of the reference
• Added CheckedCigar trait for summing the lengths for operations consuming the query/reference (behind fuzzing feature fate)
• Implemented Arbitrary for AlignmentStates and Alignment along with additional wrapper types (behind fuzzing feature gate)
• Added Viterbi algorithm for DomainPhmm (behind dev-phmm feature gate)
• Implemented Arbitrary for SemiLocalPhmm and DomainPhmm (behind fuzzing and dev-phmm feature gates)
• Added sw_scalar_alignment_override for more flexible testing of dynamic programming alignment algorithms (behind fuzzing feature gate)

Changed

• The wrapper types for generating arbitrary CIGAR strings have been modified (behind fuzzing feature gate)
• pHMM types and algorithms now rely on the more general PhmmNumber trait instead of Float (behind dev-phmm feature gate)

Fixes

• ProfileSets is re-exported in the prelude
• Fixed the initialization of the first match score for the semilocal Viterbi algorithm (behind dev-phmm feature gate)
• Fixed the calculation of the end score for semilocal Viterbi algorithm to allow passing through the END state (behind dev-phmm feature gate)
• Avoid error with NaN that may appear when using the score_from_path function on local or domain pHMMs (behind dev-phmm feature gate)

Zoe v0.0.21

Release Notes for v0.0.21 - 2025-09-05

Added

• Added sw_simd_score_ends for getting the score and end coordinates of the alignment
• Added byte index map for amino acids, BLOSUM scoring matrices, and PAM scoring matrices
• Added display implementation for WeightMatrix
• Added get_subset to WeightMatrix for subsetting a weight matrix with a different alphabet
• Added parse_matrix for parsing a WeightMatrix from a file at compile time
• Added accessor byte_keys() for ByteIndexMap
• Added unchecked constructors for ScalarProfile and StripedProfile
• Added SamDataView and SamDataViewMut
• Added map attribute syntax to define_whichever and impl_traits, so that a map can be applied when implementing iterator
• Added dev-3pass feature gate for additional alignment functions that return coordinates, as well as a three-pass alignment function with reduced memory usage
• Added score_from_path and viterbi for LocalPhmm (behind dev-phmm feature gate)
• Added DomainPhmm, along with score_from_path for it (behind dev-phmm feature gate)
• Added SemiLocalPhmm, along with score_from_path and viterbi for it (behind dev-phmm feature gate)

Changed

• MaybeAligned now can hold other types besides Alignment, and score-only functions now return MaybeAligned<u32>
• into_biased_matrix has been renamed to to_biased_matrix and takes &self
• WeightMatrix is now imported under data::matrices in addition to data
• WeightMatrix now has an additional lifetime generic for the ByteIndexMap
• WeightMatrix no longer implements Copy
• The @ symbol is no longer included at the beginning of a header in a FastQ record (the reader strips it, and the display implementation re-adds it)
• Added tags field to SamData, as well as the ability to opt out of storing the tags in SamReader
• Changed Debug impl of CigarViewMut and derived more traits
• Various Smith-Waterman profile set methods have been moved to trait methods
  for ProfileSets.
• Updated traceback in viterbi to not use nested vecs (behind dev-phmm feature gate)
• Removed as_u32 from Alignment, since the integer type for scores has been standardized

Fixes

• Fixes a bug in AlignmentStates::add_ciglet() for inc > 1