Autocomplete behavior for evidence codes

[vanaukenk]

Hi,

Curators are probably most familiar with the three-letter evidence code abbreviations. Typing those in the box brings up the commonly used GO ECO codes for some, but not all, evidence codes.
For example, IMP, IGI, and IPI (with a space after the abbreviation) all return the appropriate code in the list, but IDA and IEP, don't.
Do we want the autocomplete to give priority to matching synonyms in ECO, or should we just encourage curators to search with the text string of the name or the ECO ID?

Thx.
--K.

[cmungall]

I suspect this is a solr tuning issue. It's odd that 'transport assay
evidence' comes up when you type IDA. I suspect it's something to do
with the ':' in the synonym

id: ECO:0000134
name: transport assay evidence
synonym: "IDA: transport assay" RELATED []

vs

id: ECO:0000314
name: direct assay evidence used in manual assertion
synonym: "IDA" RELATED [GOECO:IDA]
synonym: "inferred from direct assay" EXACT [GOECO:IDA]

We have an open ticket about this in the amigo tracker, as this is
shared code
geneontology/amigo#131

The solution is to switch to solr5... but we don't intend to do that
until next year.

What to do in the interim? Unfortunately it's not very obvious with ECO
how the names map to the codes. For example if you type full name of
IDA, "inferred from direct assay" is comes up with:

 direct assay evidence used in manual assertion/ECO:0000314

A curator might think, "no I don't want that, I want IDA". But in fact
that is IDA.

I think for the meeting we just have to give people a table that maps
the code to the full name in ECO.

ECO:0000245 RCA computational combinatorial evidence used in
manual assertion
ECO:0000247 ISA sequence alignment evidence used in manual
assertion
ECO:0000303 NAS non-traceable author statement used in manual
assertion
ECO:0000304 TAS traceable author statement used in manual
assertion
ECO:0000314 IDA direct assay evidence used in manual assertion
ECO:0000315 IMP mutant phenotype evidence used in manual
assertion
ECO:0000316 IGI genetic interaction evidence used in manual
assertion
ECO:0000317 IGC genomic context evidence used in manual
assertion
ECO:0000318 IBA biological aspect of ancestor evidence used in
manual assertion
ECO:0000319 IBD biological aspect of descendant evidence used in
manual assertion
ECO:0000320 IKR phylogenetic determination of loss of key
residues evidence used in manual assertion
ECO:0000320 IMR phylogenetic determination of loss of key
residues evidence used in manual assertion
ECO:0000321 IRD rapid divergence from ancestral sequence
evidence used in manual assertion
ECO:0000353 IPI physical interaction evidence used in manual
assertion
ECO:0000501 IEA evidence used in automatic assertion

A key rule is that if you're manually entering it, the full name in ECO
will end in "used in manual assertion"

If people want to use more specific evidence types, this table shows how
GOREFs deepen evidence types:
http://purl.obolibrary.org/obo/eco/gaf-eco-mapping.txt

On 3 Dec 2015, at 5:14, vanaukenk wrote:

Hi,

Curators are probably most familiar with the three-letter evidence
code abbreviations. Typing those in the box brings up the commonly
used GO ECO codes for some, but not all, evidence codes.
For example, IMP, IGI, and IPI (with a space after the abbreviation)
all return the appropriate code in the list, but IDA and IEP, don't.
Do we want the autocomplete to give priority to matching synonyms in
ECO, or should we just encourage curators to search with the text
string of the name or the ECO ID?

Thx.
--K.

---

Reply to this email directly or view it on GitHub:
#234

[kltm]

This is indeed a bit of a mapping issue and a bit more of the issue Chris mentioned. I'm not a fan of duplicate issues tracking, so since this is tagged for searching, I'm going to close out.

[cmungall]

Now we have nowhere to indicate that this is an unresolved issue. Even
if the resolution is to have a web page showing people the mapping and
they consult that, it should be noted.

On 3 Dec 2015, at 11:07, kltm wrote:

Closed #234.

---

Reply to this email directly or view it on GitHub:
#234 (comment)

[vanaukenk]

I've added the mappings above to the end of the LEGO Google doc where David also put a Glossary of terms. Having the info in more than one place would not be a bad idea, though.

[kltm]

Multiple places brings up the risk of drift. What is the current best location of the mapping documentation?

[ukemi]

It's not really mapping. The originals live in the ECO ontology. Maybe a link to Ontobee in the LEGO documentation?

[kltm]

Okay, my understanding is that ECO is the real deal, with codes being legacy, and not completely bijective with ECO. I was under the impression that there was some documentation to clarify that cases where there is no clear map?

[ukemi]

Ah. I see what you mean. Why not just get people used to using ECO and away from codes? When I work in Noctua, I put on my ECO hat. Although at first I was a bit confused by the 'used in manual assertion' codes, once Marcus explained them in DC, it became pretty clear.

[cmungall]

+1 to that, but we don't want ontological fussiness to get in the way of
adoption. It may take some time to wean curators off codes. It may be
better to do this as part of a larger rethink about how capturing assay
results fits into GO curation

On 3 Dec 2015, at 11:34, ukemi wrote:

Ah. I see what you mean. Why not just get people used to using ECO and
away from codes? When I work in Noctua, I put on my ECO hat. Although
at first I was a bit confused by the 'used in manual assertion' codes,
once Marcus explained them in DC, it became pretty clear.

---

Reply to this email directly or view it on GitHub:
#234 (comment)