Changelog
2.1.3 - Green Kakapo
Fixed
Create output directory in the translate-igblast command if it doesn’t exist.
Allow different prefixes and organism names for igblast germline reference files.
Pinned version of transformers to avoid incompatibility with Antiberty.
2.1.2 - Green Weka hotfix II
Fixed
Improved TCRT5 and AbLang parallelization.
Minor doc fixes.
Added multithread processing for translate-igblast.
2.1.1 - Green Weka hotfix
Fixed
Fixed automatic locus determination.
Fixed example command in documentation.
2.1 - Green Weka
Added
Added fine-tuning tutorial
Fixed
Make locus uppercase to avoid missmatches.
Fixed bug on translation when
sequence_vdjcolumn was empty.Fix compatibility for Python 3.14 (does not support AbLang).
2.0 - Brown Kiwi
Added
Allow calling Python API as well as CLI.
Allow choosing sequence column for translation.
Added PROTT5, TCRT5 and TCRbert embeddings.
Added Ablang embeddings.
Added AnnData export
Added possibility to store residue-level embeddings for all models that support them.
Fixed
Refactored code to allow extending to TCR embeddings.
Used lazy imports to decrease the response time to load help message.
1.1 - Blue Tui
Added
Added explicit command for BALM-paired embedding
Added Tutorial
Added better docs on included embeddings
Fixed
When multiple heavy or light chains are present, pick the one with highest duplicate count, not consensus count.
Consider IGH, TRB and TRD as heavy chains.
1.0 - Green kea
First release on PyPI of amulety