Closed brindakv closed 2 years ago
That doesn't seem very reliable - I could imagine generating multiple models in ihm_model_list
that are not ensembles (e.g. one model for each of several states). Or a protocol could consist of multiple steps, some of which (e.g. preprocessing, alignment) act on 0 or 1 models (i.e. not an ensemble) and some of which (e.g. modeling, filtering) generate or act on an ensemble. What about taking a user-provided flag to protocol.Step
(as is done for ordered/multi_scale/multi_state already)?
If user provides the flag, then that gets priority. I was talking about cases where user doesn't provide the input and seeing if it can be inferred from other information. I see your point about multiple models not necessarily corresponding to an ensemble. So perhaps use ihm_ensemble_info
if user doesn't provide it?
_ihm_modeling_protocol_details.ensemble_flag
is not included in the_ProtocolDumper
class and it would be good to have that data item. The flag should be set toYES
if multiple models are submitted (ihm_model_list
) and/or ifihm_ensemble_info
is populated.