Bug fixes (the rare substr error). Added option to build msa to
add unobserved positions. Rewrote algorithm for paired chain annotator
to improve support for single chains supplied by accident (or
intentionally). Added prefiltering for alignment which improves speed
by roughly 2x. Updated CLI to include unobserved positions by default
and to include percent identity. Updated assign_cdr_labels to
always use the same numbering scheme as the Annotator object was
assigned upon creation (avoids possible confusion).
Minor changes to docstrings in 0.3.5.
Revised and updated API for PairedChainAnnotator, SingleChainAnnotator. Added bug fix for unusual alignment errors. Updated numbering tools to accept letter X in input. Converted PairedChainAnnotator, SingleChainAnnotator, VJGeneTool to C++. Updated vj gene assignment to use either e-value or percent identity and either IMGT or OGRDB, and to report multiple matches if multiple equivalent matches are found. Switched C++ wrapping to Nanobind. Rewrote core alignment code to produce a 30% improvement in numbering speed / efficiency. Added Aho numbering scheme. Added code to convert between numbering schemes when assigning VJ so that VJ can be assigned using any scheme not just IMGT. Added command line interface for easy use on fasta files from the command line.
Added cibuildwheel support to build cross-platform automatically.
Updated the VJ gene tool to return BOTH the assigned V and J genes and the percent identity.
Added the sort position codes function to SingleChainAnnotator, which enables the user to sort a list of position codes (this is useful and convenient when merging many lists of numbers). Set SingleChainAnnotator up to return a list indicating whether each position is cdr or framework if the user so requests.
Added MultiChainAnnotator, LiabilitySearchTool and VJGeneTool. This considerably expands the capabilities of the library to cover most basic antibody amino acid sequence analysis tasks. Removed BioPython as a dependency (it is not necessary to include a fasta file generator in SingleChainAnnotator since any users using Biopython can easily add this themselves.)
Minor updates to alignment; fixed a bug relevant to old versions of GCC that prevents compilation with old versions of GCC; removed the scipy dependency; fixed the double-normalization issue.
Added scoring with a user-supplied mask, so that the user can filter out regions they would like to not include in the score.
Fixed bug that caused gapped scoring (scoring where all gaps are ignored) to yield masked scores only.
Added sequence scoring functionality for humanness. Updated all numbering tools to improve behavior in the presence of N-terminal deletions.
Added Kabat, Martin schemes.
Initial release.