Phenotype and Manifest In

[exosome]

I've got SQL queries for getting both Phenotype and Phenotype Manifest In, but what I can't figure out is how the two are linked together (other than by publication). So for example, allele FBal0031520 lists "cell death increase" and "cell shape defective" as having come from the same publication. For "manifest in," it gives the following list:

nurse cell & actin filament
oocyte & actin filament
nurse cell ring canal
nurse cell
nurse cell ring canal rim
egg chamber
oocyte
interfollicle cell
nurse cell fusome
karyosome
nurse cell & nucleus
actin filament

Of those, the last seven are from the same publication as the cell death / cell shape phenotypes. Hypothetically, cell death could be manifest in some of those locations and cell shape in others. Is there a way that they're linked in the database?

A second question:
If I run an SQL query to get those results, it shows that three of them have no FB identifiers, and are redundant with some that do have identifiers. Why are these in here? Why, also, is there a GO identifier?
The SQL query is:

Code: Select all

SELECT g.uniquename, g.name, a.uniquename, a.name, cvt4.name, fp.value
  FROM feature g, feature_relationship fr, feature a, cvterm cvt, cvterm cvt4,
     featureprop fp
  WHERE g.feature_id = object_id AND subject_id = a.feature_id AND
    fr.type_id = cvt.cvterm_id AND cvt.name = 'alleleof' AND
    a.feature_id = fp.feature_id AND fp.type_id = cvt4.cvterm_id AND
    cvt4.name = 'derived_pheno_manifest' AND a.uniquename = 'FBal0031520';

Thanks for the help!
John

[Josh Goodman]

Hi John,

I can't answer all of these questions because they have more to do with how these terms are curated. I've pinged our curators to make sure they see this post. One thing that you may or may not realize here is that the terms in 'derived_pheno_manifest' can be attributed to different publications. If you add a few more joins from your featureprop table to featureprop_pub and then to pub you will see why some of those featureprops are repeated for the same allele.

Code: Select all

SELECT g.uniquename, g.name, a.uniquename, a.name, cvt4.name, fp.value, pub.uniquename, pub.miniref
      FROM feature g, feature_relationship fr, feature a, cvterm cvt, cvterm cvt4,
         featureprop fp, featureprop_pub fpp, pub
      WHERE g.feature_id = object_id AND subject_id = a.feature_id AND
        fr.type_id = cvt.cvterm_id AND cvt.name = 'alleleof' AND
        a.feature_id = fp.feature_id AND fp.type_id = cvt4.cvterm_id AND
        cvt4.name = 'derived_pheno_manifest' AND a.uniquename = 'FBal0031520' AND
        fp.featureprop_id=fpp.featureprop_id and fpp.pub_id=pub.pub_id;


[dosumis]

Hi John,

> Hypothetically, cell death could be manifest in some of those locations and cell shape in others.
> Is there a way that they're linked in the database?

I'm afraid not. However, the associated free text will generally clarify this. You can retrieve this for specific cases using:

Code: Select all

SELECT pd.description
    FROM phendesc pd, pub, feature_genotype fg, feature a
   WHERE    pd.genotype_id=fg.genotype_id AND fg.feature_id=a.feature_id
   AND pd.pub_id=pub.pub_id
   AND a.uniquename = 'FBal0031520' AND pub.uniquename = 'FBrf0187382';

> A second question:
> If I run an SQL query to get those results, it shows that three of them have no FB identifiers,

From your list - the three terms with '&' in them are combined terms. These are currently not stored correctly in the database. This is a known issue which we hope to fix in the next few months.

> and are redundant with some that do have identifiers. Why are these in here?

I'm not sure what you mean by redundancy here. Can you give an example?

> Why, also, is there a GO identifier?

"Phenotype manifest in" lines are used to curate anatomical and subcellular phenotypes. Cell components affected are curated use the GO cell component vocabulary - hence the GO identifier you found.

Hope you find this useful,

David (Curator at FlyBase)