Reading outputs

The designed chain gets renumbered from scratch, so a conditioned motif no longer sits at its input indices. The standard annotations written into every generated CIF are how you find where it landed, without hand-parsing the file or reconstructing an unindex mapping yourself.

Selecting a region by annotation

Standard annotations are ordinary annotations, so you can pull out any region with the atomworks selection syntax: aa.mask(...) and aa.idxs(...) evaluate a boolean query where annotation names act as variables.

from atomworks.io import parse
from rfproteina.utils.parse_presets import ANNOTATED_CIF

aa = parse("design.cif", config=ANNOTATED_CIF)["assemblies"]["1"][0]

# Annotation names are variables in the query; index the array with the mask to subset it.
motif  = aa[aa.mask("sa_mask('SA_CRDSRC') & ~hetero")]  # the conditioned motif
ligand = aa[aa.mask("hetero")]                                             # the ligand
ca     = aa[aa.mask("(chain_id == 'A') & (atom_name == 'CA')")]            # the backbone trace

Parse with the STANDARD_ANNOTATIONS_ONLY preset from rfproteina.utils.parse_presets for minimal overhead when all you need is this provenance.

Worked examples

The enzyme design tutorial uses this same selection to colour the motif separately from the scaffold, and the Python API tutorial notebook builds up the annotation system atom by atom. For a quick look without wiring up a viewer, AtomWorks’ view helper renders an AtomArray inline in a notebook, including any masked subset — see the AtomWorks documentation.