Phenoscape and colleagues meet with PATO on ontology and phenotype representation issues, Sept. 25-27, 2010

At the end of September, members of Phenoscape (Mabee, Balhoff), the Hymenoptera Anatomy Ontology (HAO) project (Yoder, Deans, Seltmann) and TAIR (Huala) met with developers of the Phenotype and Trait Ontology (PATO) (Gkoutos, Mungall, Westerfield, Lewis) at the University of Oregon.   Our discussions were focused on finding solutions to problems that have arisen as a result of PATO ontology structure, and problems for representing phenotypes in the EQ model, which have arisen in the course of annotating comparative phenotype data from the fish and hymenoptera literature.  We prepared for this meeting by developing a list of common issues and importantly, specific examples, on a Google doc shared among participants.  We all co-edited this document during the meeting with notes, decisions and examples, and we ‘published’ this Google doc for you all to see.  A number of important changes to the PATO hierarchy were proposed and subsequently made.  We also clarified best practices for modelling some common but tricky phenotypic features. One additional outcome was the participants strong recommendation that a ‘shape jamboree’ be held to improve the usability of this branch of the PATO ontology.

Some proposed changes to PATO:

  • Consolidate relational and monadic branches: One of the more major decisions was to remove the distinction between the “relational” and “monadic” branches of PATO. The relational terms can be descendants of the monadic terms which are pre-composed with a reference to a dependent entity. This change would be a major improvement helping to relate terms dealing with similar concepts and for inference using these concepts.
  • Continuums: Add a relationship ‘ranges_from’ which can be used to specify a continuum of values between two indicated qualities.  Example: a color grading “from yellow to brown”.
  • Position: Remove term ‘position’ after moving its children to more appropriate places.
  • Enable directional references: Add classes describing directions a structure can be pointed.  These classes should include pre-composed logical definitions drawing on the spatial ontology. An example would be ‘directed posteriorly’.
  • Spatial term review: Review all existing PATO terms referencing spatial aspects, and verify that they are based on logical definitions using the spatial ontology.
  • Size vs. shape: Several children of ‘size’ were noted to actually be types of ‘shape’. Many of these were noted and several were immediately revised by George Gkoutos.  This discussion revealed that many free-text characters defined by biologists refer to size-sounding terms when they are actually describing changes in shape.
  • Changes in various term definitions:
    • PATO:1485 should be obsoleted and two new separate terms, ‘condensed’ and ‘compressed’ (as a synonym of ‘flattened’) should be added. ‘Condensed’ is considered to be a type of ‘structure’, while ‘flattened’ is a ‘shape’ instead of a ‘curvature’. All children of ‘flattened’ should be reviewed to remove references to ‘curvature’.
    • Improved definition for ‘morphology’ referencing “shape or size or structure”.
    • Clarified terms ‘surface feature shape’ and ‘texture’. The parent term ‘surface shape’ was obsoleted while ‘surface feature shape’ is retained as a shape with a repeated feature on a surface. A ‘has_repeated_part’ relation should be added to be used in pre-composed subclasses of ‘surface feature shape’, linking them to specific shapes.
    • Add a term defining ‘spatial density’. The existing ‘density’ term refers to the physics concept.
  • Spatial pattern: We proposed reworking the ‘spatial pattern’ term hierarchy to logically reference other terms which the given pattern is in respect to (structures, colors, etc.). The ‘color pattern’ term should be moved under ‘spatial pattern’.
  • Synonyms: We discussed some community-specific term labels for various PATO classes.  There is an existing OBO standard for how to implement these as synonyms. Community-specific applications would need to support display of the appropriate synonym.
  • Comparative relations: Generalize existing comparative relations in PATO.  For example, Instead of ‘increased_in_magnitude_relative_to’, it will be ‘increased_in_value_relative_to’.
  • Qualitative branch: Clean up and better document ‘qualitative’ hierarchy (which is used for various logical “shortcuts”).

EQ representation issues:

  • Size bins: Relative size characters (small, medium, large) can be represented by creating (within a given annotation application) anonymous subclasses of size which are related to each other in the appropriate way using relations such as ‘increased_in_magnitude_relative_to’.  This will provide the appropriate relative size inferencing for the given character states, but, as in the original paper, not be readily comparable to size classes created for characters in other studies.
  • Negation: When describing a phenotype that is simply “not something else”, e.g. ‘not round’, the complement_of operator should be used in an OWL class expression.
  • Comparative/relative qualities should not be conflated with relational (or, better, dependent) qualities. Comparative relations such as ‘increased_in_value_relative_to’ can be used to relate one EQ to another.
  • Phenex and other annotation tools should provide enhanced interfaces for these special representation issues: creating comparative EQs by simply entering a relative entity or taxon, a simple means to say things like ‘not round’, and a way to create local groups of relatively ordered qualities for a given character (e.g. small, medium, large).

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: