Added mean field protocol for complexes#223
Conversation
|
Great, thanks Kelly! I think it makes sense to consider some refactoring at this point to avoid code duplication - just like the PLM-based protocols have infer_plm(), the mean-field based protocols could have infer_meanfield() that contains most of the non-specific code shared between monomers and complexes etc. |
|
@thomashopf agree wholeheartedly - will better modularize after current research push :) |
aggreen
left a comment
aggreen
left a comment
There was a problem hiding this comment.
This looks good- there's only minor documentation and whitespace changes that need to be made. I've been using this protocol successfully with no issues.
| [ | ||
| "prefix", "min_sequence_distance", | ||
| "scoring_model", "use_all_ecs_for_scoring", | ||
| "alignment_file","segments","focus_mode", |
There was a problem hiding this comment.
put spaces after commas in list
| "prefix", "min_sequence_distance", | ||
| "scoring_model", "use_all_ecs_for_scoring", | ||
| "alignment_file","segments","focus_mode", | ||
| "focus_sequence","theta","pseudo_count", |
|
|
||
| prefix = kwargs["prefix"] | ||
|
|
||
| # infer ECs and load them with plmc |
There was a problem hiding this comment.
delete these obsolete lines
|
|
||
| #infer ECs with mean field/DCA | ||
|
|
||
| # Below here is added material from mean_field protocol: |
| #infer ECs with mean field/DCA | ||
|
|
||
| # Below here is added material from mean_field protocol: | ||
| model = prefix+".model" |
| ) | ||
|
|
||
|
|
||
|
|
There was a problem hiding this comment.
limit to one whitespace line
| if segments is not None: | ||
| # create index mapping | ||
| seg_mapper = mapping.SegmentIndexMapper( | ||
| # kwargs["focus_mode"], outcfg["region_start"], *segments |
| ecs = mapping.segment_map_ecs(ecs,seg_mapper) | ||
|
|
||
|
|
||
|
|
There was a problem hiding this comment.
limit to one whitespace line
| # file_format="plmc_v2" | ||
| # ) | ||
|
|
||
| ###### End added material |
|
This branch is included in #225 - ok to close this pull request? |
|
Does this already have the suggested refactoring to reduce code redundancy? |
|
No, there's been no refactoring that I know of |
3 participants
Based directly off of the PLM code, but used MeanField model and fit. Note: for region_start when mapping segments, I used the number in kwargs instead of the one taken from the MeanField model - the latter led to incorrect numbering.