The lack of established and widely-used standards to describe and annotate biological assays and screening results in the domain of chemical probe and drug discovery is a severe limitation to utilize these valuable datasets to their maximum potential. Lack of standardized metadata and common terms and definitions for all relevant details of high-throughput screening (HTS) assays hinder targeted data retrieval, integration, aggregation, and analyses across different HTS datasets, for example to infer mechanisms of action of small molecule perturbagens. To address this problem, we created the BioAssay Ontology (BAO, http://bioassayontology.org/). Here we describe the evolution of BAO related to its formal structure, engineering approach, and content to enable modeling of complex assays and integration with other ontologies and complex datasets.
Uma D. Vempati1, Hande Küçük2, Saminda Abeyruwan2, Ubbo Visser2, Ahsan Mir1,Vance Lemmon1,3, Stephan C. Schürer*,1,4
1Center for Computational Science
2Department of Computer Science
3Miami Project to Cure Paralysis
4Department of Molecular and Cellular Pharmacology, University of Miami, Miami FL
Since the first release of BAO,1 we have developed several collaborations, which are focused on the integration, application, and extension of BAO; these include the BioAssay Research Database (BARD, http://bard.nih.gov/) developed in the MLP (Molecular Libraries Program), Library of Integrated Network-based Cellular Signatures (LINCS) Information FramEwork (LIFE, http://lifekb.org/), RegenBase (http://regenbase.org) , PubChem, ChEMBL, and groups in the pharmaceutical industry.
Here we give a brief update of the evolution and latest incorporation of BAO that was developed to facilitate maintenance (and collaborative development), and to provide wider scope in modeling complex assays or to drive software applications. We integrated many modules from established biomedical ontologies, and an upper level ontology. Our modularization approach defines several integral distinct BAO components and separates internal from external modules and domain-level from structural components. This approach facilitates the generation / extraction of derived ontologies (or views) that suit a particular use case or software application. We have modeled many complex assays from LINCS and the MLP in BAO. We highlight some projects in which BAO is used.
BAO core provides domain-specific terms and definitions required to describe and annotate assays and screening results. BAO leverages Description Logic and Web Ontology Language to capture and formalize knowledge about assays and screening results and to enable computational systems to utilize this knowledge.2 BAO enables the classification of assays and screening results into categories that relate to the assay format, assay method, detection technology, target, and endpoints.3 We have used BAO to annotate assays from the largest public HTS data repository, PubChem, and demonstrated its utility to categorize and analyze diverse HTS results from numerous experiments. BAO annotations also group assays into screening campaigns using relationships such as ‘is confirmatory assay of’, ‘is counter assay of’, etc. These relations are used to meaningfully aggregate screening results from different assays (using asserted assay relationships) to identify relevant compounds. A Semantic Web software application leveraging BAO, BAOSearch (http://baosearch.ccs.miami.edu/baosearch/) allows searching and exploring the curated and annotated PubChem assays and results.
We continue to develop BAO with the goal of establishing a standard to report chemical biology assays and their results. This will facilitate the interpretation, re-use and global comparison of the datasets and enable non-experts to answer complex cross-target and cross-compound queries. BAO is currently used in several public and private screening projects and evaluated by a number of organization and projects. BAO is publically available from the NCBO BioPortal.
(1) Vempati, U. D.; Przydzial, M. J.; Chung, C.; Abeyruwan, S.; Mir, A.; Sakurai, K.; Visser, U.; Lemmon, V. P.; Schürer, S. C. Formalization, annotation and analysis of diverse drug and probe screening assay datasets using the BioAssay Ontology (BAO). PloS One
2012, 7, e49198.
(2) Visser, U.; Abeyruwan, S.; Vempati, U.; Smith, R. P.; Lemmon, V.; Schürer, S. C. BioAssay Ontology (BAO): a semantic description of bioassays and high-throughput screening results. BMC Bioinformatics
2011, 12, 257.
(3) Schürer, S. C.; Vempati, U.; Smith, R.; Southern, M.; Lemmon, V. BioAssay Ontology Annotations Facilitate Cross-Analysis of Diverse High-Throughput Screening Data Sets. Journal of Biomolecular Screening
2011, 16, 415–426.