Charite Pheno Recall Analysis

[Colossus]

Not too sure if this is the best place, but I'm going to put my Charite pheno recall analysis here. I.e. why are we extracting far fewer phenos than Charite?

[Colossus]

This is a query to get all Charite genephenos where we don't ever pick up the pheno:

select distinct 
  hpo_id, 
  canonical_name 
from 
  charite c 
  join genes g 
    on (g.ensembl_id = c.ensembl_id) 
  left join pheno_mentions p 
    on (c.hpo_id = p.entity) 
where p.doc_id is null;

[Colossus]

So many Charite phenos are not "allowed" phenotypic abnormalities (abnormality; no cancer):

select count(distinct hpo_id) from charite;
6074

select count(distinct hpo_id) from charite where hpo_id not in (select distinct hpo_id from allowed_phenos);
346

So almost no "disallowed phenos"

[Colossus]

We should synonym "Abnormality of skin physiology" and all "abnormalities of" to "abnormal blah" automatically.

"Unossified sacrum" ... hard to find. Have only 11 sentences in whole database with word "unossified".

[Colossus]

Why don't we find anything with "Thymoma" (HP:0100522)?? This should be an easy one

EDIT: it's a cancer

[Colossus]

Maybe just dump all "abnormality of" and "abnormal" prefixes such as in "abnormal eye physiology"

[Colossus]

neoplasm; cancer; tumor; should all be tumors. insert synonyms manually

[Colossus]

Why don't we pick up "stillbirth" HP:0003826??

EDIT: It's not a phenotypic abnormality ... we're getting a little morbid here

[Colossus]

Chop off "morphology" and "physiology" suffixes such as in "Abnormal trabecular bone morphology" or "abnormal eye physiology" ... unless all that's remaining is a simple english word such as eye

[Colossus]

Should we leave "sarcoma" phenotypes in? and allow them? like HP:0200058 angiosarcoma?

EDIT: forget about them, it's cancer

[Colossus]

split phenos containing a slash and create synonyms; won't work perfectly but better

[Colossus]

split only slashed word

[Colossus]

If dropping "abnormality of (the)"/"abnormal" leaves only one word, don't add single word; add "physiology", "morphology", "dysplasia", "hypoplasia", "aplasia" to single word and add that

[Colossus]

replace all "physiology" by "morphology"

[Colossus]

when dropping abnormality, in general add "physiology", "morphology", "dysplasia", "hypoplasia", "aplasia"