Define rules for what to import - Implement filter, e.g. for only manual (experimental) annotations about gocamgen HOT 9 OPEN

from gocamgen.

Thanks! So here's my list right now:

acceptable_evidence_codes = ['EXP', 'IDA', 'IPI', 'IMP', 'IGI', 'IEP', 'HTP', 'HDA', 'HMP', 'HGI', 'HEP', 'ISS']

Any codes I'm missing here?

from gocamgen.

We actually also have a handful of manual annotations using these additional codes:

IC
IKR
ISM
ISO
NAS
ND
TAS

If we want to import the whole set of manual annotations, we'll need to include these codes as well.

from gocamgen.

from gocamgen.

@thomaspd are the IEA and PAINT annotations not going through the same pipeline as the rest of the annotations? Will they simply skip the GO-CAM collection? If not, why don't we include all of them as part of this experiment?

from gocamgen.

The goal of this phase is specifically to support manual GO curation in Noctua by Alliance curators, so we won't include PAINT and IEAs at this point. We'll figure out whether/how to include them at a later date.

from gocamgen.

@thomaspd Got it! I can easily filter these out:

unwanted_evidence_codes = ['IEA', 'IBA']
unwanted_evi_code_ref_combos = [('IKR', 'PMID:21873635')]

So non-PAINT IKRs will still get translated.

from gocamgen.

From our perspective it might be easier to filter out rather than in. I think any annotation that is assigned by MGI and doesn't have a ECO:0000266 (ISO) evidence code would be a good first pass. This will allow MGI curators to view and edit the annotations that they have made manually.

from gocamgen.

Now able to filter by rules specific to MOD. Just use --mod option with gen_models_by_gene.py and the filter will use rules coded in [MOD]FilterRule classes.

So far there's WB and MGI along with the default filter:

• default - Remove IEA, IBA, as well as IKRs originating from PAINT.
• WB - So far same as default filter.
• MGI - Only provided_by=MGI lines are valid, also filtering out ISOs along with default filter.

The logic will likely need to be adjusted as rules are refined. We can also think about having these rules coded outside of gocamgen and imported in.

from gocamgen.