results)relations are included in the analysis. Therefore, we developed a custom graph parsing scheme that allows these relations to be incorporated without resulting in erroneous term categorizations. This enables a more robust semantic scoping utilization of Gene Ontology, facilitating summarization of annotations in large data sets. We also demonstrate that GOcats is well suited for comparing results from separate annotation sources due to its ability to allow adjustment of categories to the appropriate annotation term granularity. GOcats thus facilitates more robust interpretation and comparison of experimental and knowledgebase annotation sources and provides new tools for semantic scoping utilization and development of Gene Ontology. Furthermore, when used alongside annotation enrichment tools such as categoryCompare2, GOcats' unique method for inferring category membership results in both improved enrichment statistics and the identification of enriched terms otherwise impossible-to-identify with statistical significance when compared to using conventional ontology inference rules. 8 has_part describe part-whole (mereological) correspondence. Therefore, we consider scoping relations to be comprised of is_a, part_of, and has_part, and mereological relations to be comprised of part_of and has_part.There are three versions of the GO database, each containing aspects of the CV with varying complexity: go-basic is filtered to exclude relations that span across multiple sub-ontologies and to include only relations that point toward the root of the ontology; go or go-core contains additional relations, such as has_part that may span sub-ontologies and which point both toward and away from the root of the ontology; and go-plus contains yet more relations in addition to cross-references to entries in external databases like the Chemical Entities of Biological Interest (ChEBI) ontology (Munoz-Torres & Carbon, 2017). The first and second versions are available in the Open Biomedical Ontology (OBO) flat text file formatting, while the third is available only in the Web Ontology Language (OWL) RDF/XML format.
Categorization-relevant issues in GOOntological graphs are typically designed as directed graphs, meaning that every edge has directionality or directed acyclic graphs (DAGs), meaning that no path exists that leads back to a node already visited if one were to traverse the graph stepwise.This allows the graph to form a complex semantic model of biology containing both general concepts and more-specific (fine-grained) concepts. The "parent-child" relation hierarchy allows biological entities to be annotated at any level of specificity (granularity) with a single term code, as fine-grained terms intrinsically capture the meaning of every one of its parent and ancestor terms through the linking of relation-defining is_a edges in the graph. However, it is deceptively non-trivial to reverse the logic and organize 1 0 Author Contributions E.W.H. worked on the design of GOcats, implemented GOcats, performed all analyses,...