Add suggested list of values for tree inference methods

[jimallman]

The property tree['^ot:curatedType'] is currently treated as a free-form text field. While we want to allow this, it's best if people choose from among a preset list of inference methods. The UI for this can include two widgets: a SELECT (pull-down) for suggested values, and a free-form text field that appears if the user choose 'Other' from the suggestion list.

[pmidford]

I've been reviewing both phylont (
http://purl.bioontology.org/ontology/PHYLONT) and cdao
(http://purl.obolibrary.org/obo/cdao.owl). Phylont has better coverage
(models, methods, and tools), but looks like it needs more review
(typos, inconsistent naming patterns) and (at least the version posted
to NCBO) was last updated in October 2012.
I'm not sure how seriously this ontology (version) is intended - the
ontology IRII is the default one offered by Protege for new ontologies.

CDAO has a few model description terms and some potentially useful
methods represented as subtypes of EdgeLengthType. CDAO was last
updated in February 2013 according to the ontology properties in the owl
file.

I started a google doc comparing the vocabularies I can share if
anyone's interested, but I'm not sure that level of review is warranted
at this point.

Thoughts?

On 3/7/14, 2:44 PM, Jim Allman wrote:

The property tree['^ot:curatedType'] is currently treated as a
free-form text field. While we want to allow this, it's best if people
choose from among a preset list of inference methods. The UI for this
can include two widgets: a |SELECT| (pull-down) for suggested values,
and a free-form text field that appears if the user choose 'Other'
from the suggestion list.

—
Reply to this email directly or view it on GitHub
#226.

[jimallman]

Thanks, @pmidford, suggestions for the method list are welcome. I can start with a simple list of inference methods on Wikipedia (http://en.wikipedia.org/wiki/Bayesian_inference_in_phylogeny, see the footer):

Choose an inference method...
===
Maximum parsimony
Maximum likelihood
Neighbor-joining
UPGMA
Bayesian inference
Least squares
Three-taxon analysis
Other (specify)

[jimallman]

@pmidford: I see that the methods listed on BioPortal are in a more extensive hierarchy: http://bioportal.bioontology.org/ontologies/PHYLONT/?p=classes&conceptid=http%3A%2F%2Fwww.co-ode.org%2Fontologies%2Font.owl%23Assessment-of-Uncertainty

I can certainly follow suit in a SELECT widget, if we want this added precision.

[pmidford]

@jimallman: Agree we could do this, though I suspect the tip terms are most common. Many inference methods will specify a model as well as the method (which is a different branch of the ontology). In those cases, both should be specified - one field or two? My intuition is that higher level terms might be useful for the method, but if a model is specified, it will be a tip term.

[pmidford]

Agree we could do this. Probably worth looking at the existing
ot:curatorType fields and see if the parent types in the hierarchy (or
similar) get any use.

On 3/7/14, 5:01 PM, Jim Allman wrote:

@pmidford https://github.com/pmidford: I see that the methods listed
on BioPortal are in a more extensive hierarchy:
http://bioportal.bioontology.org/ontologies/PHYLONT/?p=classes&conceptid=http%3A%2F%2Fwww.co-ode.org%2Fontologies%2Font.owl%23Assessment-of-Uncertainty

I can certainly follow suit in a SELECT widget, if we want this added
precision.

—
Reply to this email directly or view it on GitHub
#226 (comment).

[jimallman]

Many inference methods will specify a model as well as the method (which is a different branch of the ontology). In those cases, both should be specified - one field or two?

Interesting! And now I see that there's a proposed ot:inferenceMethod for trees in the Nexson wiki page:
https://github.com/OpenTreeOfLife/api.opentreeoflife.org/wiki/NexSON#wiki-table-iii---proposed-not-yet-implemented-predicates

I'm not sure if this is just a better name for ot:curatedType, or if the combination fits your model + method description.

[pmidford]

The intent of the ot:inferenceMethod was to capture the method, though
it probably ought to include the model as well - which would leave the
other text information (e.g., figure references) in *ot:curatedType *for
now.

On 3/7/14, 10:02 PM, Jim Allman wrote:

Many inference methods will specify a model as well as the method
(which is a different branch of the ontology). In those cases,
both should be specified - one field or two?

Interesting! And now I see that there's a proposed
ot:inferenceMethod for trees in the Nexson wiki page:
https://github.com/OpenTreeOfLife/api.opentreeoflife.org/wiki/NexSON#wiki-table-iii---proposed-not-yet-implemented-predicates

I'm not sure if this is just a better name for ot:curatedType, or if
the combination fits your model + method description.

—
Reply to this email directly or view it on GitHub
#226 (comment).