Migrating from ABDB¶

antibody-utils replaces the reusable parts of ABDB, the closed-source Python package behind the original SAbDab. ABDB needed a per-host configuration file and a local copy of the SAbDab database to import at all; antibody-utils needs neither.

What changes¶

No database. The parts of ABDB that served SAbDab’s database (Database_interface, ABDB_updater, config and the command-line tool) are not included. To find structures, use SAbDab itself.
Numbering with ANARCII. Sequences and structures are numbered with ANARCII, from the optional numbering extra. ABnum, IMGT/DomainGapAlign, MUSCLE-based numbering and the Wolfguy scheme are gone.
IMGT by default. ABDB defaulted to the Chothia scheme and definition. antibody-utils numbers in IMGT by default, and functions that assign regions take the region definition as a required keyword argument, definition=.
GEMMI, not Biopython. Structures are read with GEMMI and wrapped, not subclassed, so residues and atoms are GEMMI objects.
Positions. Residue positions are Position(number, insertion) objects, where no insertion is "" rather than " ". Position.from_tuple and Position.to_tuple convert from and to ABDB’s (number, " ") tuples.
Explicit pairing. Heavy and light chains are not paired automatically, and antigens are not identified. Pair chains with Fv(heavy, light).

Equivalent names¶

Names on the right are relative to antibody_utils.

`ABDB`	antibody-utils
Regions
`AB_Utils.region_definitions.get_region`	`regions.get_region(position, chain, *, scheme, definition)`
`AB_Utils.region_definitions.annotate_regions`	`regions.annotate_regions`, or `NumberedSequence.annotate_regions(definition)`
`AB_Utils.region_definitions.Accept`	`regions.RegionSelector`
`AB_Utils.regions_tuples` and the other region tables	`regions.REGION_GROUPS`, and `get_region`
`AB_Utils.tuple_interpret`	`numbering.Position.parse`, `Position.from_tuple`
Numbering
`Annotate.annotate`, `Annotate.anarci`	`numbering.number_sequence`, `number`, `iter_number`
`AB_Utils.sequence_liabilities.annotate_sequences`	`numbering.number`
Sequences
`AB_Utils.identity`, `fab_identity`	`sequence.identity(a, b, *, definition, regions)`
`AB_Utils.similarity`	`sequence.similarity(a, b, *, definition, regions, normalise)`
`ABchain.get_CDRs`, `get_sequence` (as sequences)	`sequence.cdr_sequences`, `region_sequences`
`AB_Utils.sequence_liabilities.get_liabilities`	`liabilities.find_liabilities(heavy, light)`
Structures
`AbPDB.AntibodyParser().get_antibody_structure(id, file)`	`structure.read_structure(path, scheme)`
`AntibodyStructure.get_fabs()`	`structure.Fv(heavy, light)`, paired by you
`AntibodyStructure.get_abchains()`, `get_chains()`	`AntibodyStructure.chains`, `heavy_chains`, `light_chains`
`Fab.get_VH()`, `Fab.get_VL()`	`Fv.heavy`, `Fv.light`
`ABchain.get_fragments()`, `get_CDRs()`, `residue.region`	`AntibodyChain.region_residues(definition)`
`ABchain.get_sequence()`	`AntibodyChain.numbered`
`AbPDB.Select.fv_only`, `CDRH3`, `fv_only_no_CDRH3`	`structure.select(source, regions, *, definition, exclude)`
`AbPDB.Select.backbone`, `fv_only_backbone`	`structure.select(..., atoms=structure.BACKBONE_ATOMS)`
Geometry
`AB_Utils.superimpose`, `get_equivalent_arrays`	`geometry.superpose`, `geometry.rmsd`
`AB_Utils.calculate_region_rmsd`	`geometry.region_rmsd`
`AB_Utils.calculate_orientation_rmsd`	`geometry.orientation_rmsd`
`ABangle.abangle().calculate_angles(fab)`	`geometry.abangle(fv)`

Not included¶

Antigen identification and classification (get_antigens, get_antigen, is_bound), and automatic pairing of chains into Fabs and scFvs.
Header analysis (is_engineered, is_scfv, hapten detection).
North canonical classes (AB_Utils.canonicals.assign_canonical); for sequence-based canonical classes, see SCALOP.
The interface selection, TMalign superposition, ABangle’s angle predictor and plots, and Visualise.