Update gorule-0000006 to exclude annotations with source = GOC

[pgaudet]

Hello,

As stated, gorule-0000006 is creating false positives for interontology links derived from BP.

We would like to change the rule so that any aspect is allowed with IEP/HEP annotations when assigned by =GOC.

Thanks, Pascale

[kltm]

Doc PR #1520

[kltm]

To clarify, when you mean "false positive", you referring to the on line here?
http://snapshot.geneontology.org/reports/tair-report.html#gorule-0000006

[pgaudet]

For example GO:0005623 (obsolete cell ;)
yes, this should not be reported anymore.

[kltm]

@pgaudet Are there any other examples of this?

[pgaudet]

In snapshot there is still a TAIR annotation that is a false negative for this error:

ERROR - Violates GO Rule: GORULE:0000006: IEP and HEP usage is restricted to terms from the Biological Process ontology -- TAIR locus:2092221 STP10 GO:0005623 TAIR:Publication:501768255|PMID:26893494 IEP C AT3G19940 AT3G19940|STP10|SUGAR TRANSPORT PROTEIN 10|MPN9.18 protein taxon:3702 20160301 GOC TAIR:locus:2092221

but GO:0005623 is obsolete so this is filter via the gorule-0000001 parser (I think).

Is that what you were asking ?

[kltm]

@pgaudet Apologies, I should have brought this up today.
The question we were asking (as part of clarifying this issue): are there any examples of this besides the single one from TAIR.

[pgaudet]

I dont think so

[pgaudet]

The problem here is interontology kinks keep the same evidence code (AFAIK). The best solution may be to change that behavior.

[mugitty]

Modify rule definition to:
The IEP and its high throughput equivalent, HEP, evidence codes are used where process involvement is inferred from
the timing or location of expression of a gene, particularly when
comparing a gene that is not yet characterized with the timing or
location of expression of genes known to be involved in a particular
process. This type of annotation is only suitable with terms from the
Biological Process ontology, except when the annotation comes from an inter ontology link, in which case all GO aspects are allowed.

For GAF
IF evidence code (col 7) = IEP or HEP, and Assigned_by (Col 15) GOC, then - > OK

ELSE if Aspect (Col 9) = BP -> OK
ELSE error = wrong GO aspect for evidence

ELSE if assigned by is NOT GOC , check Aspect (Col 9) = IF col 9 = BP -> OK
ELSE error = wrong GO aspect for evidence

Implement similar rule for GPAD (column 10 and get aspect from goid)

[mugitty]

@pgaudet , we can simplify
IF evidence code (col 7) = IEP or HEP, and Assigned_by (Col 15) GOC, then - > OK
ELSE IF evidence code (col 7) = IEP or HEP, and Aspect (col 9) NOT BP -> error wrong GO aspect for evidence
ELSE -> OK

[pgaudet]

Pascale to add some tests to check that there is an exception for annotations assigned by GOC.

[pgaudet]

Added
! FAILS GORULE:0000006 - passes because uses assigned_by is GOC, even if there is an IEP with a molecular function term.
GeneDB PF3D7_0507500 SUB1 enables GO:0004672 PMID:12764150 IEP F GORULE:0000006-4 subtilisin-like protease 1 gene NCBITaxon:36329 20090624 GOC

Now we need to check that we dont get an error for this line

[pgaudet]

Working on the GAF test script

[pgaudet]

Check at the next release

[pgaudet]

Working except for merged terms, see #2081