Extending and using anatomical vocabularies in the stimulating peripheral activity to relieve conditions project

Abstract

The stimulating peripheral activity to relieve conditions (SPARC) program is a US National Institutes of Health-funded effort to improve our understanding of the neural circuitry of the autonomic nervous system (ANS) in support of bioelectronic medicine. As part of this effort, the SPARC project is generating multi-species, multimodal data, models, simulations, and anatomical maps supported by a comprehensive knowledge base of autonomic circuitry. To facilitate the organization of and integration across multi-faceted SPARC data and models, SPARC is implementing the findable, accessible, interoperable, and reusable (FAIR) data principles to ensure that all SPARC products are findable, accessible, interoperable, and reusable. We are therefore annotating and describing all products with a common FAIR vocabulary. The SPARC Vocabulary is built from a set of community ontologies covering major domains relevant to SPARC, including anatomy, physiology, experimental techniques, and molecules. The SPARC Vocabulary is incorporated into tools researchers use to segment and annotate their data, facilitating the application of these ontologies for annotation of research data. However, since investigators perform deep annotations on experimental data, not all terms and relationships are available in community ontologies. We therefore implemented a term management and vocabulary extension pipeline where SPARC researchers may extend the SPARC Vocabulary using InterLex, an online vocabulary management system. To ensure the quality of contributed terms, we have set up a curated term request and review pipeline specifically for anatomical terms involving expert review. Accepted terms are added to the SPARC Vocabulary and, when appropriate, contributed back to community ontologies to enhance ANS coverage. Here, we provide an overview of the SPARC Vocabulary, the infrastructure and process for implementing the term management and review pipeline. In an analysis of >300 anatomical contributed terms, the majority represented composite terms that necessitated combining terms within and across existing ontologies. Although these terms are not good candidates for community ontologies, they can be linked to structures contained within these ontologies. We conclude that the term request pipeline serves as a useful adjunct to community ontologies for annotating experimental data and increases the FAIRness of SPARC data.

Keywords: InterLex; SPARC; anatomical terms; ontologies; peripheral nervous system (PNS).

FIGURE 1

Schematic diagram of SciCrunch vocabulary infrastructure showing how content is managed across the different components. Different types of users interact with the system via different interfaces.

FIGURE 2

The relationship of the SPARC vocabulary to infrastructure components. The SPARC Vocabulary is represented by the large circle on the left (solid dotted line). The entire SPARC Vocabulary is available through a SciGraph instance, including the full imports of community ontologies comprising the vocabulary. The totality of InterLex is represented by the small circle on the right (blue dotted line) that partially exists within the SciGraph instance. The overlap between these circles represents the subset of the SPARC Vocabulary that is available in both InterLex and SciGraph that is made available via the SPARC Community Portal (purple dotted line) and augmented through the term request pipeline.

FIGURE 3

Overview of SPARC data workflow modified from Osanlouy et al. (2021). Stars indicate steps in the pipeline where the annotation is performed, and terms are most likely to be added to the SPARC Vocabulary.

FIGURE 4

Accessing the SPARC Vocabulary through an API in the MBF Bioscience software suite (A). The Vocabulary Services tool (B, circle) provides users with a subset of SPARC Terms (D) – specified by organ, species, and parcelation (C) – that can be applied to anatomical annotations. An option to request new terms (D, circle) launches the SAWG portal dashboard. SPARC Terms are also available through Scaffold Tools (B, square), where segmentation data and organ scaffolds are displayed side-by-side to preview concordant fiducial SPARC terms for subsequent scaffold registration.

FIGURE 5

Schematic diagram of the ApiNATOMY model of bladder innervation (Surles-Zeigler et al., 2021). The diagram illustrated major neural circuits involving the urinary bladder and urethra.

FIGURE 6

SPARC term request and review pipeline. The process comprises three iterative steps: (A) submission (Term request), (B) Term review, and (C) Term engineering, shown in the three colored boxes. Details are provided in the text.

FIGURE 7

SPARC term request workflow. Flow chart illustrating the term request and review steps of the term request pipeline.

FIGURE 8

Disposition of all terms submitted to the term request and review pipeline as of August 2021. The star in the figure notates the 10 terms labeled as “not in InterLex.” This label refers to terms that were added to the pipeline but were either not added to the SAWG community or depreciated from InterLex, a subset (SAWG community) of the vocabulary in InterLex via free-text search.

FIGURE 9

Term Engineering for the submitted term “Dorsal part of serosa of urinary bladder.” The term engineering is usually performed by adding appropriate machine-readable relationships to relate submitted terms to the appropriate supercategory and other terms. (A) An example of a relationship term in InterLex is IntersectionOfPartOf. (B) This relationship acts as a linker between the term “Dorsal part” with identifier PATO:0001233 and “serosa of urinary bladder” with identifier UBERON:0001260.

FIGURE 10

SAWG term review. Examples of the SAWG review process. e.g., The Thoracic ansa was submitted to the pipeline and reviewed by the SAWG since the SAWG did not recognize the term after additional investigation. Once contacted, the investigator indicated it was an error.