Was discussed in the SBF facility meeting. Since the pipeline already emits the best predicted structure in top_ranked_structures by default, it that separateness of "rank_0" as the "best" predicted structure should be preserved.
However, it is also helpful to have a single .mmcif files to share with colleagues that has all atomic data of the Assembly without having to re-spec everything about the model run in separate files.
This is provided by .model.Model
Since I learnt how to implement flags with --write_binary to implement #583, should also include a --all_structs with default presence boolean False (action='store_true' in argparse)
EDIT: implemented with fixed rank_{0..4} rather than a glob because nextflow shell parsing was causing issues. Will validate in real-pipeline that structure globbing is being passed (and rejected) properly by:
if all_structs:
for struct_file in struct_files:
try:
biopy_struct = _parse_structure(struct_file)
except Exception as err:
print(
f"Skipping unparseable structure file {struct_file}: {err}",
file=sys.stderr,
)
continue
selected_struct_files.append(struct_file)
biopy_structs.append(biopy_struct)
Was discussed in the SBF facility meeting. Since the pipeline already emits the best predicted structure in
top_ranked_structuresby default, it that separateness of "rank_0" as the "best" predicted structure should be preserved.However, it is also helpful to have a single
.mmciffiles to share with colleagues that has all atomic data of theAssemblywithout having to re-spec everything about the model run in separate files.This is provided by .model.Model
Since I learnt how to implement flags with
--write_binaryto implement #583, should also include a--all_structswith default presence boolean False (action='store_true'in argparse)EDIT: implemented with fixed rank_{0..4} rather than a glob because nextflow shell parsing was causing issues. Will validate in real-pipeline that structure globbing is being passed (and rejected) properly by: