Transforming the study of organisms: Phenomic data models and knowledge bases

Abstract

The rapidly decreasing cost of gene sequencing has resulted in a deluge of genomic data from across the tree of life; however, outside a few model organism databases, genomic data are limited in their scientific impact because they are not accompanied by computable phenomic data. The majority of phenomic data are contained in countless small, heterogeneous phenotypic data sets that are very difficult or impossible to integrate at scale because of variable formats, lack of digitization, and linguistic problems. One powerful solution is to represent phenotypic data using data models with precise, computable semantics, but adoption of semantic standards for representing phenotypic data has been slow, especially in biodiversity and ecology. Some phenotypic and trait data are available in a semantic language from knowledge bases, but these are often not interoperable. In this review, we will compare and contrast existing ontology and data models, focusing on nonhuman phenotypes and traits. We discuss barriers to integration of phenotypic data and make recommendations for developing an operationally useful, semantically interoperable phenotypic data ecosystem.

Fig 1. Phenotypic data integration challenges.

(A) The many names for the mountain gorilla, Gorilla beringei, resulted from years of nomenclatural acts, misspellings, and the quirks of human language and popular culture. (B) The term “paramere” has been ambiguously used to describe 5 different parts of the male genitalia of a gasteruptiid wasp (red). (C) The end-of-season height of a wheat plant can be described by an exact measurement or relative to a “wild type.” (D) With the exception of microorganisms, measurements are collected from specimens but are sometimes represented as a single value representing an entire population or taxon. All 4 of these panels represent 1 or more challenges to phenotypic data integration. Image credit: Panel A by David J. Patterson, used with permission.

Fig 2. TBox versus ABox.

The TBox (A) includes classes (kinds of things), properties (the possible relationships between classes and instances of the classes), and assertions about the classes and properties. The ABox (B) represents instances of the classes represented in the TBox and assertions about those instances. For example, an instance of femur in a frog specimen is 1.2 cm long. Image credit: Photo from National Museum of Natural History, Washington DC.

Fig 3. EQ Formalism for categorical phenotypes versus character states.

From [112]. The EQ Formalism uses ontology terms from an anatomy ontology (green) and a trait ontology (blue) to represent a phenotype and maps to the Character/Character State model (gray). EQ, Entity–Quality.

Fig 4. Darwin Core star schema with traits.

Phenotypes can be represented in the Darwin Core star schema that consists of separate tabular files (blue) linked together by unique identifiers for taxa, occurrences, and measurements (green).

Fig 5. Measurement-Based phenotype data models.

(A) Semantic Morph·D·Base. Pink-bordered boxes: instances; yellow-bordered boxes: classes; gray-bordered boxes: literals (labels or values); boxes with dashed borders: named graphs. (B) TaxonWorks. The underlying goal is to let scientists assert phenotype observations as required for their research. Assertions are persisted in Descriptor–Observation format where subclasses of descriptor (e.g., qualitative, quantitative, statistical, gene, free-text, and media) classify/define observations. Descriptor types anticipate downstream serialization into computable formats, semantic or otherwise. Phenotype assertions are at the class (= Taxon concept, an “OTU” in TaxonWorks) or instance (= Collection object) level (“Entity”). Ultimately, both levels will permit anatomical part assertions. While the approach includes improvements to the overall semantics, it still lacks specifics used in other models (e.g., Fig 5A and 5C); however, the typed descriptor approach provides a flexible software design, whereby incremental improvements to semantics are possible. All data are highly annotatable. Dashed boxes are features in progress. (C) Global Plant Phenological Database. Rounded rectangles represent classes, and hexagons represent instances. The original data set (bottom of figure) indicates that there is an instance of the class/phenophase “open flower presence,” which is a quality of an instance of “whole plant” from the PO. Because the value of the instance of measurement datum is >0, the ontology infers that open flowers are present. Due to the subsumption hierarchy of the PO (left side of figure), the ontology can also infer that nonsenesced flowers, flowers, and plant structures are present. IAO, Information Artifact Ontology; PATO, Phenotype and Trait Ontology; PO, Plant Ontology; OBI, Ontology for Biomedical Investigations; OTU, Operational Taxonomic Unit; RDF, Resource Description Framework; RO, Relations Ontology; UO, Unit Ontology.