D1018–D1027 Nucleic Acids Research, 2019, Vol. 47, Database issue Published online 22 November 2018 doi: 10.1093/nar/gky1105 Expansion of the Human Phenotype Ontology (HPO) knowledge base and resources Sebastian Köhler 1,2,3 , Leigh Carmody3,4 , Nicole Vasilevsky 3,5 , Julius O.B. Jacobsen3,6 , Daniel Danis3,4 , Jean-Philippe Gourdine 3,5 , Michael Gargano3,4 , Nomi L. Harris3,7 , Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 Nicolas Matentzoglu3,8 , Julie A. McMurry 3,9 , David Osumi-Sutherland3,8 , Valentina Cipriani 3,10,11,12 , James P. Balhoff 3,13 , Tom Conlin 3,9 , Hannah Blau 3,4 , Gareth Baynam14,15,16,17,18 , Richard Palmer17 , Dylan Gratian14 , Hugh Dawkins18 , Michael Segal19 , Anna C. Jansen20,21 , Ahmed Muaz3,22 , Willie H. Chang23 , Jenna Bergerson24 , Stanley J.F. Laulederkind 25 , Zafer Yüksel 26 , Sergi Beltran 27,28 , Alexandra F. Freeman24 , Panagiotis I. Sergouniotis29 , Daniel Durkin4 , Andrea L. Storm30,31 , Marc Hanauer32 , Michael Brudno23 , Susan M. Bello 33 , Murat Sincan34 , Kayli Rageth34 , Matthew T. Wheeler 35 , Renske Oegema36 , Halima Lourghi32 , Maria G. Della Rocca30,31 , Rachel Thompson 37 , Francisco Castellanos4 , James Priest38 , Charlotte Cunningham-Rundles39 , Ayushi Hegde4 , Ruth C. Lovering 40 , Catherine Hajek34 , Annie Olry32 , Luigi Notarangelo24 , Morgan Similuk24 , Xingmin A. Zhang 3,4 , David Gómez-Andrés41 , Hanns Lochmüller 27,42,43,44 , Hélène Dollfus45 , Sergio Rosenzweig46 , Shruti Marwaha35 , Ana Rath 32 , Kathleen Sullivan47 , Cynthia Smith 33 , Joshua D. Milner24 , Dorothée Leroux45 , Cornelius F. Boerkoel34 , Amy Klion24 , Melody C. Carter24 , Tudor Groza3,22 , Damian Smedley3,6 , Melissa A. Haendel3,5,9 , Chris Mungall3,7 and Peter N. Robinson 3,4,48,* 1 Charité Centrum für Therapieforschung, Charité––Universitätsmedizin Berlin Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Berlin 10117, Germany, 2 Einstein Center Digital Future, Berlin 10117, Germany, 3 Monarch Initiative, monarchinitiative.org, 4 The Jackson Laboratory for Genomic Medicine, Farmington, CT 06032, USA, 5 Oregon Health & Science University, Portland, OR 97217, USA, 6 Genomics England, Queen Mary University of London, Dawson Hall, Charterhouse Square, London EC1M 6BQ, UK, 7 Environmental Genomics and Systems Biology, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA, 8 European Bioinformatics Institute (EMBL-EBI), Wellcome Trust Genome Campus, Cambridge, UK, 9 Linus Pauling institute, Oregon State University, Corvallis, OR, USA, 10 William Harvey Research Institute, Queen Mary University College of London, 11 UCL Genetics Institute, University College of London, 12 UCL Institute of Ophthalmology, University College of London, 13 Renaissance Computing Institute, University of North Carolina at Chapel Hill, 14 Western Australian Register of Developmental Anomalies and Genetic Services of Western Australia, Department of Health, Government of Western Australia, WA, Australia, 15 School of Paediatrics and Telethon Kids Institute, University of Western Australia, Perth, WA, Australia, 16 Institute for Immunology and Infectious Diseases, Murdoch University, Perth, WA, Australia, 17 Spatial Sciences, Department of Science and Engineering, Curtin University, Perth, WA, Australia, 18 The Office of Population Health Genomics, Department of Health, Government of Western Australia, Perth, WA, Australia, 19 SimulConsult, Chestnut Hill, MA, USA, 20 Neurogenetics Research Group, Vrije Universiteit Brussel, Brussels, Belgium, 21 Pediatric Neurology Unit, Department of Pediatrics, UZ Brussel, Brussels, Belgium, 22 Garvan Institute of Medical Research, Darlinghurst, Sydney, NSW 2010, Australia, 23 Centre for Computational Medicine, Hospital for Sick Children and Department of Computer Science, University of Toronto, Toronto, Canada, 24 National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD, * To whom correspondence should be addressed. Tel: +1 860 837 2095; Email: peter.robinson@jax.org  C The Author(s) 2018. Published by Oxford University Press on behalf of Nucleic Acids Research. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Nucleic Acids Research, 2019, Vol. 47, Database issue D1019 USA, 25 Rat Genome Database, Department of Biomedical Engineering, Medical College of Wisconsin & Marquette University, 8701 Watertown Plank Road Milwaukee, WI 53226, USA, 26 Bioscientia GmbH, Ingelheim, Germany, 27 CNAG-CRG, Centre for Genomic Regulation (CRG), The Barcelona Institute of Science and Technology, Baldiri Reixac 4, Barcelona 08028, Spain, 28 Universitat Pompeu Fabra (UPF), Barcelona, Spain, 29 University of Manchester & Manchester Royal Eye Hospital, Manchester, UK, 30 ICF, Rockville, MD, USA, 31 National Center for Advancing Translational Sciences, Office of Rare Diseases Research, National Institutes of Health, Bethesda, MD, USA, 32 INSERM, US14––Orphanet, Plateforme Maladies Rares, 75014 Paris, France, 33 The Jackson Laboratory, Bar Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 Harbor, ME, USA, 34 Sanford Imagenetics, Sanford Health, Sioux Falls, SD, USA, 35 Center for Undiagnosed Diseases, Stanford University School of Medicine, Stanford, CA, USA, 36 Department of Genetics, University Medical Center Utrecht, the Netherlands, 37 Institute of Genetic Medicine, Newcastle University, Newcastle upon Tyne, UK, 38 Department of Pediatrics, Stanford University School of Medicine, Stanford, CA, USA, 39 Mount Sinai School of Medicine, New York, NY, USA, 40 Institute of Cardiovascular Science, University College London, UK, 41 Child Neurology Unit. Hospital Universitari Vall d’Hebron, Vall d’Hebron Research Institute (VHIR), Barcelona, Spain, 42 Department of Neuropediatrics and Muscle Disorders, Medical Center––University of Freiburg, Faculty of Medicine, Freiburg, Germany, 43 Children’s Hospital of Eastern Ontario Research Institute, University of Ottawa, Ottawa, Canada, 44 Division of Neurology, Department of Medicine, The Ottawa Hospital, Ottawa, Canada, 45 Centre for Rare Eye Diseases CARGO, SENSGENE FSMR Network, Strasbourg University Hospital, Strasbourg, France, 46 Immunology Service, Department of Laboratory Medicine, NIH Clinical Center, Bethesda, MD, USA, 47 Department of Pediatrics, Division of Allergy Immunology, The Children’s Hospital of Philadelphia, University of Pennsylvania Perelman School of Medicine, 3615 Civic Center Boulevard, Philadelphia, PA 19104, USA and 48 Institute for Systems Genomics, University of Connecticut, Farmington, CT, USA Received September 17, 2018; Revised October 18, 2018; Editorial Decision October 19, 2018; Accepted October 24, 2018 ABSTRACT INTRODUCTION The Human Phenotype Ontology (HPO)––a standard- A cornerstone of differential diagnostics and translational ized vocabulary of phenotypic abnormalities asso- research is deep phenotyping: the computational analysis ciated with 7000+ diseases––is used by thousands of detailed, individual clinical abnormalities (1,2). The Hu- of researchers, clinicians, informaticians and elec- man Phenotype Ontology (HPO) provides the most com- prehensive resource for computational deep phenotyping tronic health record systems around the world. Its and has become the de facto standard for deep phenotyp- detailed descriptions of clinical abnormalities and ing in the field of rare disease––whether for computable computable disease definitions have made HPO the disease definitions, description of clinical abnormalities or de facto standard for deep phenotyping in the field to aid genomic diagnostics. A foundational and integrative of rare disease. The HPO’s interoperability with other component of the Monarch Initiative (3,4), the HPO has ontologies has enabled it to be used to improve di- been adopted internationally by numerous organizations, agnostic accuracy by incorporating model organism both academic and commercial; these include the 100,000 data. It also plays a key role in the popular Exomiser Genomes Project, the NIH Undiagnosed Disease Program tool, which identifies potential disease-causing vari- and Network (UDP and UDN), the Undiagnosed Diseases ants from whole-exome or whole-genome sequenc- Network International (UDNI), RD-CONNECT, SOLVE- ing data. Since the HPO was first introduced in 2008, RD and many others (5–9). The HPO recently achieved status as an International Rare Disease Research Consor- its users have become both more numerous and tium (IRDiRC) recognized resource and is in use by the more diverse. To meet these emerging needs, the Global Alliance for Genomics and Health (10) and the as- project has added new content, language transla- sociated Matchmaker Exchange (3,11). Here we describe in- tions, mappings and computational tooling, as well tegrated HPO resources which we have revised, expanded, as integrations with external community data. The or invented since the previous articles in this series (12,13). HPO continues to collaborate with clinical adopters Previously, we reported on a range of algorithms that to improve specific areas of the ontology and ex- had been developed by our group and others to support tend standardized disease descriptions. The newly phenotype-driven genomic diagnostics (12). Since then, the redesigned HPO website (www.human-phenotype- HPO has been applied to an increasing range of use cases. ontology.org) simplifies browsing terms and explor- Usage of HPO is now commonplace for the analysis of ing clinical features, diseases, and human genes. clinical whole-exome and genome sequencing (WES/WGS) data (14–25) as well as for data integration in transla- tional research and bioinformatics (16,26–39). A phenotype risk score based on a mapping of electronic health-record (EHR)-derived billing codes to HPO terms allowed high- D1020 Nucleic Acids Research, 2019, Vol. 47, Database issue throughput ascertainment of EHR phenotypes such that of pathologies or alternatively gene lists created ‘on the cases and controls of Mendelian diseases could be distin- fly’ based on the HPO terms provided with the individual guished and the pathogenicity of variants associated with case. These major databases are not only contributing to Mendelian diseases was characterized (40). In another set- gene discovery and diagnosis of the unsolved patients in- ting, EHR narratives were explored to extract HPO terms cluded in the platforms (10) but also providing source data by natural language processing and the resulting terms were for many computational developments. Within the Solve- successfully used to prioritize causal genes for Mendelian RD project (https://solve-rd.eu), RD-Connect worked with diseases in pediatric patients (41). Additionally, an increas- Orphanet and HPO to implement the first version of the Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 ing number of commercial applications are using HPO Phenopackets standard (https://github.com/phenopackets) terms. For instance, the SimulConsult Genome-Phenome and export ∼600 cases in Phenopacket format, including Analyzer uses HPO terms to tag findings. This is currently clinical phenotype (HPO annotation), clinical diagnosis being used to document findings entered by the users with (ORDO), molecular diagnosis (OMIM) and gene name of codes in exported reports, and the codes will also be used genes identified as causal or candidate. The export included to identify findings in the electronic health record as in- both solved cases and unsolved cases that contain sufficient puts to be considered in diagnosis (42). A key feature of the information for phenotypic algorithm evaluation. In addi- HPO is its logical interoperability with basic research on- tion, work is ongoing that will enable assessment of the cor- tologies such as the Mammalian Phenotype Ontology (MP) relation between the level, detail and quantity of phenotypic (43), Uberon (44) and the Cell Ontology (45). This inter- annotation and the solve rate, which will provide clinicians operability is leveraged within the Exomiser tool (described with better advice on the level of detail to provide in their below). The International Mouse Phenotyping Consortium annotations and feed back into improvements to algorithms (IMPC) recently identified 360 new candidate molecular such as those implemented in Exomiser. causes of human Mendelian diseases (46); these included Ontologies should be responsive to the community (43). an inherited heart disease ‘Arrhythmogenic Right Ventricu- In the past 2 years we have made improvements to the on- lar Dysplasia’ that affects the heart muscle, and ‘Charcot- tology based on input from clinicians and researchers, as Marie-Tooth disease’, which is characterized by nerve dam- is evidenced by term requests that have been submitted via age leading to muscle weakness and an awkward way of our GitHub tracker (12). There, we provide a template that walking. This discovery was made possible because (i) the guides users through the process of providing information human diseases had been defined in terms of their com- including the suggested term label, definitions and com- ponent HPO phenotypes; (ii) the mouse phenotypes were ments, synonyms, references and diseases that should be an- mapped to the MP; and (iii) Monarch’s phenotype com- notated to the new term. Periodically we also organize col- parison algorithm (47) is designed to traverse HP and MP laborative workshops with clinical groups that would like to with ease. Similarly, the Rat Genome Database (RGD) an- revise and extend entire areas of the HPO. Five such work- notates genes, QTLs and strains for phenotype using pheno- shops have been conducted since the 2017 HPO update (Ta- type terms from the Mammalian Phenotype (MP) Ontology ble 1). (43); more recently, RGD has converted their annotations of The HPO project additionally has a long-term collabora- human phenotypes from MP to HPO (48). tion with Orphanet in the framework HIPBI-RD (harmo- HPO has been adopted as the phenotypic annota- nizing phenomics information for a better interoperability tion ontology of choice for many large-scale rare dis- in the rare disease field), a project that was funded by the E- ease genome-phenome databases and analysis tools includ- Rare 3 ERA-NET program (50) and will be continued in the ing the RD-Connect Genome-Phenome Analysis Platform framework of the SOLVE-RD project, as well as in the Eu- (GPAP) (49), the Broad Center for Mendelian Genomics ropean Joint Co-fund Programme for Rare Diseases (EJP- and its SEQR platform, the rare disease arm of the UK RD). This project has resulted in more than 60 000 HPO 100,000 Genomes Project, the NIH Undiagnosed Diseases annotations for diseases in the Orphanet database and over Program and the Undiagnosed Diseases Network Interna- one thousand new term requests and other improvements tional (UDNI). This is creating a vast body of clinically of existing HPO terms. Phenotype-disease annotations in- validated, linked genome-phenome data that not only as- clude the frequency of occurrence of a phenotype in a dis- sists in the diagnosis of the subjects themselves but can ease (see Table 2), as well as the fact that a phenotype is be exploited for further developments of the ontology and part of established diagnostic criteria or is a pathognomonic associated diagnostic algorithms. For example, the RD- sign. These annotations are available for download and can Connect GPAP mandates submission of HPO-coded phe- be consulted in the Orphanet website. Furthermore, this notypic data through the PhenoTips tool, using custom- collaboration has produced the HPO-ORDO Ontological designed disease-specific data collection forms on top of the Module (HOOM in which the HPO and Orphanet Rare ‘enter-what-you-see’ HPO entry box. The average number Diseases Ontology can be used together). of phenotypic annotations per index case is eight (with an average of six observed and two excluded features) and the LOGICAL ENHANCEMENTS AND INTEROPERABIL- GPAP now contains linked genome-phenome datasets on ITY 5000 individuals. Through data submission from European Reference Networks in the Horizon 2020-funded Solve-RD The HPO provides textual definitions for ease of use, but project this number will increase to >20 000 datasets in it also has a robust logical representation with OWL- the coming 2–3 years. The GPAP allows the user to filter based logical definitions based on species-neutral ontolo- variants using predefined gene panels for specific groups gies such as Uberon, the Gene Ontology, the Cell On- Nucleic Acids Research, 2019, Vol. 47, Database issue D1021 Table 1. Community workshops and collaborations aimed at HPO content expansion and refinement Organization Location Focus Undiagnosed Diseases Network (UDN); Stanford University, CA, USA (March 2017) Cardiology Stanford Center for Inherited Cardiovascular Diseases (SCICD) European Reference Network for Rare Eye Mont Sainte-Odile, France (October 2017) Ophthalmology Disease (ERN-EYE) National Institute of Allergy and Infectious National Institutes of Health, Bethesda, MD, Allergy and immunology Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 Disease (NIAID) USA (May and July 2018) Neuro-MIG European network for brain St Julians, Malta; Lisbon, Portugal (February Malformations of cortical development (MCD) malformations (www.neuro-mig.org) 2018; September 2018) European Society for Immunodeficiencies Vienna Austria (September 2018) Inborn errors of immunity. (ESID) and the European Reference network on rare primary immunodeficiency, autoinflammatory and autoimmune diseases (ERN-RITA) Table 2. The HPO records the frequencies of phenotypic features in three different ways Frequency categories Term ID Definition Obligate HP:0040280 Always present, i.e. in 100% of the cases. Very HP:0040281 Present in 80–99% of the cases. frequent Frequent HP:0040282 Present in 30–79% of the cases. Occasional HP:0040283 Present in 5–29% of the cases. Very rare HP:0040284 Present in 1–4% of the cases. Excluded HP:0040285 Present in 0% of the cases. Percentage of persons in which a phenotypic feature is observed Percentage x% This is used to record frequency of a feature in a disease if the number of probands is not available, e.g. 42%. Number of persons in a cohort in whom a phenotypic feature was observed N of M n/m This is used to record how many persons with a certain disease were observed to have a notation given phenotypic feature represented by an HPO term, e.g. 5/13. This should be used only if the feature was ruled out in the remaining m-n individuals. Frequency information can be used by differential diagnostic algorithms such as BOQA (62). If possible, HPO annotations are made with the precise counts, but percentages or overall frequency categories are used if that is all that is available. The frequency categories are aligned with those of Orphanet. tology and others. For instance, Delayed patellar ossifica- cal definitions. Pattern-based ontology development prac- tion (HP:0006454) is defined with reference to the PATO tices (52,53) are increasingly used to manage the generation term delayed (PATO:0000502), the Gene Ontology term os- of logical definitions. Rather than encoding logical defini- sification (GO:0001503) and the Uberon term for patella tions manually in OWL using an ontology editor, pattern- (UBERON:0002446). The OBO version of the ontology is based development separates the blueprint of the logi- a simplified version of the full OWL version that contains cal definition––essentially the definition with placeholder all of the terms as well as their subclass (is-a) relations, but variables––from the actual definition of the term, which is does not contain the computational logical definitions. usually encoded in the form of a spreadsheet record. Mem- 'has part' some bers of the Monarch Initiative are contributing to commu- (delayed nity tools for pattern-based development using Dead Sim- and ('inheres in' some ple Ontology Design Patterns (DOSDP, (52)) and the On- (ossification tology Development Kit (ODK). and ('occurs in' some patella))) To support the use of model organisms to further hu- and ('has modifier' some abnormal)) man health research, developers of the Mammalian Pheno- These logical definitions can be used for quality con- type (MP) ontology (54) have collaborated with the HPO trol (51), to infer new classifications (is a/subclass relation- team to develop compatible logical definitions, but these ef- ships) that were not explicitly asserted and for cross-species forts were restricted to comparison of individual definitions phenotype analysis (46). However, this can only work if and resulted in manual changes to the respective ontolo- compatible sets of definitions are used. gies. Pattern-based development offers a more accurate and Manually maintaining compatible logical definitions scalable alternative by developing common patterns that across large ontologies such as the HPO is error-prone and all phenotype ontologies (i.e. all organisms) can refer to may lead to inconsistent description in one ontology and and that can be applied to a whole branch of an ontology especially across different phenotype ontologies. Even spe- at once. For example, the ‘increasedSize’ pattern defines a cialized branches of the ontology, such as the ones address- blueprint for a logical definition as follows: “has part’ some ing morphological abnormalities, can have divergent logi- (‘increased size’ and (‘inheres in’ some %s) and (‘qualifier’ D1022 Nucleic Acids Research, 2019, Vol. 47, Database issue some ‘abnormal’))’. Using DOSDP in conjunction with the Thirty-six new molecular phenotypes have been added ODK, any phenotype ontology developer who needs to de- to the HPO. These new terms were identified from fine a phenotype describing the increased size of something metabolomics data provided by the Metabolomics Core (such as an anatomical entity) can now simply commit to from the Undiagnosed Disease Network, the Human the increasedSize pattern. More than 40 patterns specif- Metabolome Database (HMDB) and articles related to ically for phenotype ontology development are currently inborn errors of metabolism. The new terms were cu- available in the Uber-Phenotype (UPheno) repository. rated in a spreadsheet that captured information about The clinical features represented in HPO are connected metabolite name, corresponding chemicals and their Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 via subclass relations. Other relationships between those identifiers (ChEBI and HMDB), direction of change classes hold as well, but have not previously been encoded (increase/decrease), location of the abnormal metabo- computationally. For example, phenotype ontologies may lite concentration (blood, urine, cerebrospinal fluid), have two separate classes to represent the increase and de- synonyms, gene/locus association, disease identifiers for crease in size of an anatomical entity such as the liver. To associated diseases (OMIM or MONDO IDs) and key represent such relations, we have added opposite relations publication (PubMed IDs). For instance, an increased level to all terms in HPO using a text and logic-based approach of galactonate in red blood cell (HP:0410063) is associated (see phenopposites GitHub repository under ‘Availability’). with patients with galactosemia (MONDO:0018116; gene: The Monarch Initiative has been a key organizer of a GALT). community effort to use pattern-based ontology develop- The new Clinical modifier subontology allows more ex- ment to reconcile logical definitions on a large scale across pressive and precise disease definitions and can also be used well-established and emerging phenotype ontologies includ- to annotate individual patients. This subontology contains ing HPO, MP, and phenotype ontologies for Caenorhabditis terms to describe severity, positionality and external fac- elegans, Xenopus and Drosophila. To that end, we recently tors that tend to trigger or ameliorate the features of a dis- organized a Phenotype Ontology development and recon- ease. The previous Onset subontology has been expanded to ciliation workshop (Phenotype Ontologies Traversing All a Clinical course subontology, which additionally contains The Organisms: POTATO). At this workshop, more than 40 terms to describe mortality, progression of disease and the ontology curators, developers and biomedical experts came temporal pattern of features of disease (Figure 1). The fre- together to learn about our updated tool-chain for pattern- quency of features can be described in one of three methods based development and to discuss discrepancies between (Table 2). the logical definitions across various phenotype ontologies. The HPO annotation file format had remained un- As a result of the meeting, representatives of all the phe- changed since the first publication of the HPO in 2008 (57); notypes ontologies have committed to an ongoing collab- to accommodate the aforementioned new annotation re- oration to align their respective ontologies by developing sources, we have updated the annotation file format. This sets of common design patterns and using these to define format has slots to capture clinical modifiers, sex-specific terms in their ontologies. The outcome of these community features of disease and to track the history of biocuration efforts will be an integrated ecosystem of phenotype ontolo- of terms (Table 3). gies that can be leveraged in HPO-based clinical diagnostics A new tool called HPOWorkbench has been developed and disease mechanism discovery. to enable browsing through HPO terms and annotations. It can generate GitHub issues directly and can be used by collaborators to provide feedback or new suggestions. DISEASE ANNOTATIONS EXOMISER UPDATE The HPO project provides a comprehensive set of com- putable definitions of rare diseases in the form of anno- Exomiser utilizes the HPO to find potential disease-causing tations which describe the clinical features (HPO terms) variants from whole-exome or whole-genome sequencing that characterize each disease. Each annotated feature can data. The last two major updates to the Exomiser software have metadata including its typical age of onset and the fre- have focused on decoupling the data updates from the soft- quency (for instance, the HPO lists the frequency of Pro- ware release cycle and enabling analysis of either GRCh37 trusio acetabuli [HP:0003179] in persons with Marfan syn- or GRCh38 genomic samples. We updated the variant data drome as 113/146 based on a published clinical study (55)). sources to also include allele frequency data from gnomAD, Such annotation metadata can be used to improve the ac- TOPMed and the UK10 datasets and added annotations curacy of the HPO-based matching algorithms (56). for variant pathogenicity from ClinVar. We also added the Recent updates to our corpus of disease annotations in- ability for users to specify fine-grained maximum allele fre- clude a new file format with robust representation of clini- quencies to be used for prioritizing alleles under different cal modifiers, as well as migration to the Monarch Merged inheritance models and assigning these to likely syndromes Disease Ontology (MONDO), which provides a unified set based on the phenotype matches. Moreover, the Exomiser of disease terms and definitions with computationally de- variant data sources have not only been decoupled from the clared equivalencies to resources such as OMIM and Or- software release cycle, but also from the phenotype ontolo- phanet. The annotation data is readily available for compu- gies and disease annotations. This ensures that we can re- tational use via Monarch’s Biolink API (see resources be- lease Exomiser with the very latest disease and model organ- low). We have also produced a new stand-alone tool to aid ism annotations and that they can be updated on demand. curation of the disease annotations. These user-facing updates have happened against a back- Nucleic Acids Research, 2019, Vol. 47, Database issue D1023 A B Clinical modifier Clinical course Aggravated by Mortality/Aging Aggravated by activity Age of death Exacerbated by head trauma ... … Onset Ameliorated by Infantile onset Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 Ameliorated by carbohydrate ingestion Childhood onset Ameliorated by heat … Ameliorated by pregnancy Pace of progression … Nonprogressive Pain characteristic Progressive Dull … Sharp Temporal pattern … Acute Position Nocturnal Laterality … Bilateral Left … C FCAS2 Spatial pattern OMIM:611762, MONDO:0012724 Acral familial cold autoinflammatory syndrome 2 Axial … has phenotype f=3/3 Severity Borderline Mild Episodic Fever … HP:0001954 Triggered by Triggered by carbohydrate ingestion modifier onset Triggered by cold ... Triggered by cold Infantile Onset Refractory HP:0003593 HP:0025206 Figure 1. Overview of the clinical modifier (A, left) and clinical course (B, right) subontologies. These subontology terms can be used in combination with existing HPO terms to qualify and enrich their meaning. (C) A schematic presentation of one HPO annotation for the disease familial cold autoinflam- matory syndrome 2 (FCAS2). In a publication on this disease, three of three reported patients were found to have episodic fever with infantile (or earlier) onset that was triggered by exposure to cold (63). Table 3. New HPO annotation file format Field Item Required Example 1 Database ID Yes MIM:154700, ORPHA:558 or MONDO:0007947 2 DB Name Yes Achondrogenesis, type IB 3 Qualifier No NOT or empty 4 HPO ID Yes HP:0002487 5 DB Reference Yes OMIM:154700 or PMID:15517394 6 Evidence Yes IEA 7 Onset No HP:0003577 8 Frequency No HP:0003577 or 12/45 or 22% 9 Sex No MALE or FEMALE 10 Modifier No HP:0025257 11 Aspect Yes ‘P’ or ‘C’ or ‘I’ or ‘M’ 12 BiocurationBy Yes HPO:skoehler[YYYY-MM-DD] The file contains 12 tab-separated fields, some of which can be left empty. The ‘Modifier’ and ‘BiocurationBy’ fields can contain multiple items separated by semicolons. For instance, to indicate that a disease is characterized by a skin rash (HP:0000988) that is Recurrent (HP:0031796) and Triggered by cold (HP:0025206) one would annotate HP:0031796;HP:0025206 in the Modifier column. Many annotations go through multiple stages of biocuration. In this case, the individual biocuration events are also added as a semicolon-separated list. D1024 Nucleic Acids Research, 2019, Vol. 47, Database issue ground of continued engineering and performance improve- playing the plain-language vocabulary along with the medi- ments. As a result of the continued development and usage, cal terminology, patients and families become familiar with the Exomiser also recently received the approval of the In- the language they are commonly exposed to in the litera- ternational Rare Diseases Research Consortium (IRDiRC) ture and clinical settings. The public utilizes the HPO medi- as a recognized resource. We have also been able to build cal terms and layperson synonyms to better understand the on HPO being chosen as the terminology for clinical phe- broad spectrum of clinical findings associated with a spe- notype data collection by the UK National Health Service cific disease and to search and navigate the GARD web- (NHS) by introducing Exomiser as a key variant prioriti- site and other resources to retrieve information about mul- Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 zation service for the 100 000 Genomes Project and future tiple diseases associated with a given phenotype. Inclusion NHS-commissioned service for rare disease genetic testing. of the HPO data on the GARD website makes the disease Benchmarking on the solved cases to date shows Exomiser webpages more robust, educates the rare disease commu- can identify over 80% of the diagnoses in the top five candi- nity and empowers them to become partners in their medi- dates (unpublished communication from the 100K Genome cal care. project). The labels, synonyms and textual definitions of the HPO are also being translated into several languages includ- ing French, Spanish, Italian, German, Dutch, Portuguese, SYNONYMS AND TRANSLATIONS Turkish, Japanese, Russian and Chinese; this is critical to One of the key advantages of ontologies is that semantic ensure equitable health care and precision public health (See meaning is attached to concepts, rather than to their names. project homepage below). Tools such as PhenoTips (60) al- This enables each entity to have one or more synonyms, as ready make use of the existing Spanish and French trans- well as translations into other languages. Multiple groups lations, together with a user interface in those languages have taken advantage of this ability to create synonyms for to enable HPO-based phenotyping for clinicians who are HPO concepts for diverse settings, including enabling self- not fluent in English. In the Spanish Undiagnosed Disease phenotyping by patients without medical expertise and en- Network clinicians phenotype patients in Spanish, and then abling capture of data in diverse languages, with subsequent share with the Matchmaker Exchange (13). One further ex- international sharing and analysis. ample is the Life Languages project in Western Australia Patients themselves are an eager and untapped source (WA), which is using the HPO to translate medical and of information about symptoms and phenotypes, however, biological terms into partner Aboriginal Australian Lan- medical terminology is often perplexing to them, making guages. This is being integrated with HPO term extraction it difficult to use resources like the HPO. Further, some from 3D facial images as part of the Pilbara Faces program phenotypes go unnoticed by the clinician (such as those in remote WA. only seen at home). To enable patients to use the HPO di- rectly and to improve collaboration and communication be- NEW HPO WEBSITE tween patients and their physicians, we have recently added ‘layperson’ synonyms to the entirety of the HPO (58). Ap- The HPO website application has been redesigned and re- proximately 36% of the HPO terms have at least one layper- built from the ground up to be both more responsive and son synonym, 89% of the MONDO diseases annotated to more intuitive (Figure 2). Made possible by the new single- HPO have at least one HPO annotation with a layperson page app approach and lightweight microservices, the new synonym and 60% of all disease annotations refer to HPO application loads faster and supports intuitive search capa- terms with lay translations. This coverage suggests that the bilities, such as auto-complete and term highlight features, layperson HPO would be useful in a diagnostic setting de- to allow the user to efficiently browse through the ontol- spite incomplete coverage. Efforts are currently underway ogy data and corresponding hierarchy. The HPO website to evaluate the diagnostic utility of the layHPO, both syn- uses the ProtVista tool to display genes and genetic vari- thetically as well as in cohorts of previously diagnosed rare ants associated with Mendelian diseases (61). The redesign disease patients. also sets the stage for better integration with monarchinitia- The Sanford Health Imagenetics program has deployed tive.org to facilitate exploration of similar genes and pheno- an online screening tool for patients to self-report traits, types across species. signs, and symptoms in a questionnaire format that is mapped to HPO and leverages the layperson synonyms. HPO FOR MEDICAL EDUCATION This is integrated with the Sanford Imagenetics population- based genotyping initiative. The Genetic and Rare Dis- Clinical features in HPO are also connected to disease eases Information Center (GARD), a program of the Na- nosologies (medical classification schemes) such as ORDO, tional Center for Advancing Translational Sciences Office OMIM, and MONDO. These relationships are typically of Rare Diseases Research (NCATS-ORDR), provides re- curated from literature; however, they can also be crowd- liable, public-friendly information for over 7000 genetic sourced. Phenotate (http://phenotate.org), which was de- and/or rare diseases (59). GARD recently incorporated ta- veloped in the framework of the HIPBI-RD project, is a bles on the disease webpages that display information from web-based tool that allows undergraduate or medical stu- the HPO including the medical terms for associated symp- dents, as well as medical residents, to annotate OMIM and toms and phenotypic abnormalities, the related layperson ORDO diseases with HPO phenotypes by completing class- synonyms, the frequency of the phenotypic features and room exercises. Students are encouraged to refer to the lit- the link to the HPO webpage for the specific term. By dis- erature to select the correct symptoms and enter the refer- Nucleic Acids Research, 2019, Vol. 47, Database issue D1025 Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 Figure 2. Screenshot of the new HPO Website application. Users can search for HPO terms, annotated diseases, or disease-associated genes using an autocomplete widget. The hierarchical structure of the ontology is shown in an abbreviated fashion for clarity’s sake. Only the direct parent and child terms of the currently displayed term are shown in the hierarchy. The total number of decedent terms is shown for each term in the hierarchy to help users decide which parts of the ontology to explore. ences used into their annotations. In a second-year under- notypes in the HPO have been greatly improved for several graduate molecular genetics class (MGY200) at the Uni- areas of medicine thanks to community interactions. versity of Toronto, 78 students used Phenotate to annotate three genetic diseases: Marfan syndrome (MFS), Friedre- ich’s ataxia (FRDA) and congenital myasthenic syndrome. DATA AVAILABILITY Overall, students collectively provided more comprehen- The Human Phenotype Ontology and its resources are sive annotations than clinicians who also submitted anno- available at http://www.human-phenotype-ontology.org tations. Phenotate is an open platform, available for use by Web version of the Exomiser: https://phenomics-dev.kccg. anyone teaching genetics. By crowdsourcing annotations, garvan.org.au/web-exomiser/ Phenotate hopes to improve the HPO and related nosolo- Orphanet: http://www.orpha.net gies, while also offering students an educational tool that HOOM: HPO-ORDO Ontological Module: http://www. supplements their coursework. orphadata.org/ Biolink: https://api.monarchinitiative.org/api/ MONDO: http://obofoundry.org/ontology/mondo.html Monarch Initiative: https://monarchinitiative.org CONCLUSION Phenotate: http://phenotate.org In the 2 years since the previous Nucleic Acids Research HPOWorkbench: https://github.com/ database article (12), the HPO has continued to grow in TheJacksonLaboratory/HPOworkbench both reach and scope. The HPO has put a strong emphasis Ontology Development Kit: https://github.com/ on working with interested members of the community to INCATools/ontology-development-kit revise and extend individual areas of the HPO, and we wel- UPheno: https://github.com/obophenotype/upheno come interactions with more groups in any area of medicine. Phenopposites: https://github.com/Phenomics/ The HPO project has begun to develop resources for layper- phenopposites sons to interact with the HPO and software designed for HPO Translation Project: https://crowdin.com/project/hpo- patients. Annotations and improved representation of phe- translation D1026 Nucleic Acids Research, 2019, Vol. 47, Database issue FUNDING Campbell,J. et al. (2014) The Human Phenotype Ontology project: linking molecular biology and disease through phenotype data. National Institutes of Health (NIH), Monarch Initia- Nucleic Acids Res., 42, D966–D974. tive [OD #5R24OD011883]; Forums for Integrative Phe- 14. Taylor,R.L., Parry,N.R.A., Barton,S.J., Campbell,C., Delaney,C.M., nomics [U13 CA221044-01]; NCATS Data Translator Ellingford,J.M., Hall,G., Hardcastle,C., Morarji,J., Nichol,E.J. et al. (2017) Panel-Based clinical genetic testing in 85 children with [1OT3TR002019]; NCATS National Center for Digital inherited retinal disease. Ophthalmology, 124, 985–991. Health Informatics Innovation [U24 TR002306]; NIH Data 15. Fang,H., Wu,Y., Yang,H., Yoon,M., Jiménez-Barrón,L.T., Commons [1 OT3 OD02464-01 UNCCH]; Cost Action Mittelman,D., Robison,R., Wang,K. and Lyon,G.J. (2017) Whole Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 CA 16118 Neuro-MIG; British Heart Foundation Pro- genome sequencing of one complex pedigree illustrates challenges gramme Grant [RG/13/5/30112]; Division of Intramu- with genomic medicine. BMC Med. Genomics, 10, 10. 16. Posey,J.E., Rosenfeld,J.A., James,R.A., Bainbridge,M., Niu,Z., ral Research; NIAID; NIH; E-RARE project Hipbi-RD Wang,X., Dhar,S., Wiszniewski,W., Akdemir,Z.H., Gambin,T. et al. [01GM1608]; European Union’s Horizon 2020 Research (2016) Molecular diagnostic experience of whole-exome sequencing and Innovation Programme [779257]. Funding for open in adult patients. Genet. Med., 18, 678–685. access charge: NIH; Donald A. Roux Family Fund (to 17. Retterer,K., Juusola,J., Cho,M.T., Vitazka,P., Millan,F., Gibellini,F., Vertino-Bell,A., Smaoui,N., Neidich,J., Monaghan,K.G. et al. (2016) P.N.R.). Clinical application of whole-exome sequencing across clinical Conflict of interest statement. None declared. indications. Genet. Med., 18, 696–704. 18. Zhu,Q., Liu,H., Chute,C.G. and Ferber,M. (2015) EHR based genetic testing knowledge base (iGTKB) development. BMC Med. REFERENCES Inform. Decis. Mak., 15, S3. 1. Delude,C.M. (2015) Deep phenotyping: the details of disease. Nature, 19. Fujiwara,T., Yamamoto,Y., Kim,J.-D., Buske,O. and Takagi,T. (2018) 527, S14–S15. PubCaseFinder: A case-report-based, phenotype-driven 2. Robinson,P.N. (2012) Deep phenotyping for precision medicine. differential-diagnosis system for rare diseases. Am. J. Hum. Genet., Hum. Mutat., 33, 777–780. 103, 389–399. 3. Mungall,C.J., Washington,N.L., Nguyen-Xuan,J., Condit,C., 20. Baker,K., Gordon,S.L., Melland,H., Bumbak,F., Scott,D.J., Smedley,D., Köhler,S., Groza,T., Shefchek,K., Hochheiser,H., Jiang,T.J., Owen,D., Turner,B.J., Boyd,S.G., Rossi,M. et al. (2018) Robinson,P.N. et al. (2015) Use of model organism and disease SYT1-associated neurodevelopmental disorder: a case series. Brain, databases to support matchmaking for human disease gene discovery. 141, 2576–2591. Hum. Mutat., 36, 979–984. 21. Thiffault,I., Farrow,E., Zellmer,L., Berrios,C., Miller,N., Gibson,M., 4. Mungall,C.J., McMurry,J.A., Köhler,S., Balhoff,J.P., Borromeo,C., Caylor,R., Jenkins,J., Faller,D., Soden,S. et al. (2018) Clinical Brush,M., Carbon,S., Conlin,T., Dunn,N., Engelstad,M. et al. (2017) genome sequencing in an unbiased pediatric cohort. Genet. Med., The Monarch Initiative: an integrative data and analytic platform doi:10.1038/s41436-018-0075-8. connecting phenotypes to genotypes across species. Nucleic Acids 22. Stokman,M.F., van der Zwaag,B., van de Kar,N.C.A.J., van Res., 45, D712–D722. Haelst,M.M., van Eerde,A.M., van der Heijden,J.W., Kroes,H.Y., 5. Ramoni,R.B., Mulvihill,J.J., Adams,D.R., Allard,P., Ashley,E.A., Ippel,E., Schulp,A.J.A., van Gassen,K.L. et al. (2018) Clinical and Bernstein,J.A., Gahl,W.A., Hamid,R., Loscalzo,J., McCray,A.T. et al. genetic analyses of a Dutch cohort of 40 patients with a (2017) The undiagnosed diseases network: Accelerating discovery nephronophthisis-related ciliopathy. Pediatr. Nephrol., 33, 1701–1712. about health and disease. Am. J. Hum. Genet., 100, 185–192. 23. Short,P.J., McRae,J.F., Gallone,G., Sifrim,A., Won,H., 6. Taruscio,D., Groft,S.C., Cederroth,H., Melegh,B., Lasko,P., Geschwind,D.H., Wright,C.F., Firth,H.V., FitzPatrick,D.R., Kosaki,K., Baynam,G., McCray,A. and Gahl,W.A. (2015) Barrett,J.C. et al. (2018) De novo mutations in regulatory elements in Undiagnosed Diseases Network International (UDNI): white paper neurodevelopmental disorders. Nature, 555, 611–616. for global actions to meet patient needs. Mol. Genet. Metab., 116, 24. Tumienė,B., Maver,A., Writzl,K., Hodžić,A., Čuturilo,G., 223–225. Kuzmanić-Šamija,R., Čulić,V. and Peterlin,B. (2018) Diagnostic 7. Gahl,W.A., Mulvihill,J.J., Toro,C., Markello,T.C., Wise,A.L., exome sequencing of syndromic epilepsy patients in clinical practice. Ramoni,R.B., Adams,D.R., Tifft,C.J. and UDN. (2016) The NIH Clin. Genet., 93, 1057–1062. Undiagnosed Diseases Program and Network: applications to 25. Trujillano,D., Bertoli-Avella,A.M., Kumar Kandaswamy,K., modern medicine. Mol. Genet. Metab., 117, 393–400. Weiss,M.E., Köster,J., Marais,A., Paknia,O., Schröder,R., 8. Gall,T., Valkanas,E., Bello,C., Markello,T., Adams,C., Bone,W.P., Garcia-Aznar,J.M., Werber,M. et al. (2017) Clinical exome Brandt,A.J., Brazill,J.M., Carmichael,L., Davids,M. et al. (2017) sequencing: results from 2819 samples reflecting 1000 families. Eur. J. Defining disease, diagnosis, and translational medicine within a Hum. Genet., 25, 176–182. homeostatic perturbation paradigm: The national institutes of health 26. Meyer,K., Kirchner,M., Uyar,B., Cheng,J.-Y., Russo,G., undiagnosed diseases program experience. Front. Med., 4, 62. Hernandez-Miranda,L.R., Szymborska,A., Zauber,H., 9. Thompson,R., Johnston,L., Taruscio,D., Monaco,L., Béroud,C., Rudolph,I.M., Willnow,T.E. et al. (2018) Mutations in disordered Gut,I.G., Hansson,M.G., ’t Hoen,P.B., Patrinos,G.P., Dawkins,H. regions can cause disease by creating dileucine motifs. Cell. 175, et al. (2014) RD-Connect: an integrated platform connecting 239–253. databases, registries, biobanks and clinical bioinformatics for rare 27. Chen,C., Chen,D., Xue,H., Liu,X., Zhang,T., Tang,S., Li,W. and disease research. J. Gen. Intern. Med., 29, S780–S787. Xu,X. (2018) IDGenetics: a comprehensive database for genes and 10. Boycott,K.M., Rath,A., Chong,J.X., Hartley,T., Alkuraya,F.S., mutations of intellectual disability related disorders. Neurosci. Lett., Baynam,G., Brookes,A.J., Brudno,M., Carracedo,A., den 685, 96–101. Dunnen,J.T. et al. (2017) International cooperation to enable the 28. Haghighi,A., Krier,J.B., Toth-Petroczy,A., Cassa,C.A., Frank,N.Y., diagnosis of all rare genetic diseases. Am. J. Hum. Genet., 100, Carmichael,N., Fieg,E., Bjonnes,A., Mohanty,A., Briere,L.C. et al. 695–705. (2018) An integrated clinical program and crowdsourcing strategy for 11. Philippakis,A.A., Azzariti,D.R., Beltran,S., Brookes,A.J., genomic sequencing and Mendelian disease gene discovery. NPJ Brownstein,C.A., Brudno,M., Brunner,H.G., Buske,O.J., Carey,K., Genome Med., 3, 21. Doll,C. et al. (2015) The Matchmaker Exchange: a platform for rare 29. Doğan,T. (2018) HPO2GO: prediction of human phenotype ontology disease gene discovery. Hum. Mutat., 36, 915–921. term associations for proteins using cross ontology annotation 12. Köhler,S., Vasilevsky,N.A., Engelstad,M., Foster,E., McMurry,J., co-occurrences. PeerJ, 6, e5298. Aymé,S., Baynam,G., Bello,S.M., Boerkoel,C.F., Boycott,K.M. et al. 30. Rao,A., Vg,S., Joseph,T., Kotte,S., Sivadasan,N. and Srinivasan,R. (2017) The human phenotype ontology in 2017. Nucleic Acids Res., (2018) Phenotype-driven gene prioritization for rare diseases using 45, D865–D876. graph convolution on heterogeneous networks. BMC Med. Genomics, 13. Köhler,S., Doelken,S.C., Mungall,C.J., Bauer,S., Firth,H.V., 11, 57. Bailleul-Forestier,I., Black,G.C., Brown,D.L., Brudno,M., Nucleic Acids Research, 2019, Vol. 47, Database issue D1027 31. MacLennan,A.H., Kruer,M.C., Baynam,G., Moreno-De-Luca,A., 47. Robinson,P.N., Köhler,S., Oellrich,A. and Sanger Mouse Genetics Wilson,Y.A., Zhu,C., Wintle,R.F., Gecz,J. and members of the ProjectSanger Mouse Genetics Project, Wang,K., Mungall,C.J., International Cerebral Palsy Genomics Consortium. (2018) Cerebral Lewis,S.E., Washington,N., Bauer,S., Seelow,D.S. et al. (2014) palsy and genomics: an international consortium. Dev. Med. Child Improved exome prioritization of disease genes through cross-species Neurol., 60, 209–210. phenotype comparison. Genome Res., 24, 340–348. 32. Saklatvala,J.R., Dand,N. and Simpson,M.A. (2018) Text-mined 48. Shimoyama,M., De Pons,J., Hayman,G.T., Laulederkind,S.J.F., phenotype annotation and vector-based similarity to improve Liu,W., Nigam,R., Petri,V., Smith,J.R., Tutaj,M., Wang,S.J. et al. identification of similar phenotypes and causative genes in monogenic (2015) The Rat Genome Database 2015: genomic, phenotypic and disease patients. Hum. Mutat., 39, 643–652. environmental variations and disease. Nucleic Acids Res., 43, Downloaded from https://academic.oup.com/nar/article/47/D1/D1018/5198478 by Universitat Autonoma de Barcelona user on 14 June 2021 33. Adler,A., Kirchmeier,P., Reinhard,J., Brauner,B., Dunger,I., D743–D750. Fobo,G., Frishman,G., Montrone,C., Mewes,H.W., Arnold,M. et al. 49. Lochmüller,H., Badowska,D.M., Thompson,R., Knoers,N.V., (2018) PhenoDis: a comprehensive database for phenotypic Aartsma-Rus,A., Gut,I., Wood,L., Harmuth,T., Durudas,A., characterization of rare cardiac diseases. Orphanet. J. Rare Dis., 13, Graessner,H. et al. (2018) RD-Connect, NeurOmics and 22. EURenOmics: collaborative European initiative for rare diseases. Eur. 34. Cornish,A.J., David,A. and Sternberg,M.J.E. (2018) PhenoRank: J. Hum. Genet., 26, 778–785. reducing study bias in gene prioritization through simulation. 50. Maiella,S., Olry,A., Hanauer,M., Lanneau,V., Lourghi,H., Bioinformatics, 34, 2087–2095. Donadille,B., Rodwell,C., Köhler,S., Seelow,D., Jupp,S. et al. (2018) 35. Singh,T., Kurki,M.I., Curtis,D., Purcell,S.M., Crooks,L., McRae,J., Harmonising phenomics information for a better interoperability in Suvisaari,J., Chheda,H., Blackwood,D., Breen,G. et al. (2016) Rare the rare disease field. Eur. J. Med. Genet., loss-of-function variants in SETD1A are associated with doi:10.1016/j.ejmg.2018.01.013. schizophrenia and developmental disorders. Nat. Neurosci., 19, 51. Köhler,S., Bauer,S., Mungall,C.J., Carletti,G., Smith,C.L., 571–577. Schofield,P., Gkoutos,G.V. and Robinson,P.N. (2011) Improving 36. Posey,J.E., Harel,T., Liu,P., Rosenfeld,J.A., James,R.A., Coban ontologies by automatic reasoning and evaluation of logical Akdemir,Z.H., Walkiewicz,M., Bi,W., Xiao,R., Ding,Y. et al. (2017) definitions. BMC Bioinformatics, 12, 418. Resolution of disease phenotypes resulting from multilocus genomic 52. Osumi-Sutherland,D., Courtot,M., Balhoff,J.P. and Mungall,C. variation. N. Engl. J. Med., 376, 21–31. (2017) Dead simple OWL design patterns. J. Biomed. Semantics, 8, 18. 37. Beck,T., Hastings,R.K., Gollapudi,S., Free,R.C. and Brookes,A.J. 53. Xiang,Z., Zheng,J., Lin,Y. and He,Y. (2015) Ontorat: automatic (2014) GWAS Central: a comprehensive resource for the comparison generation of new ontology terms, annotations, and axioms based on and interrogation of genome-wide association studies. Eur. J. Hum. ontology design patterns. J. Biomed. Semantics, 6, 4. Genet., 22, 949–952. 54. Smith,C.L. and Eppig,J.T. (2012) The Mammalian Phenotype 38. Li,M.J., Liu,Z., Wang,P., Wong,M.P., Nelson,M.R., Kocher,J.-P.A., Ontology as a unifying standard for experimental and Yeager,M., Sham,P.C., Chanock,S.J., Xia,Z. et al. (2016) GWASdb high-throughput phenotyping data. Mamm. Genome, 23, 653–668. v2: an update database for human genetic variants identified by 55. Chun,K.J., Yang,J.H., Jang,S.Y., Lee,S.H., Gwag,H.B., Chung,T.-Y., genome-wide association studies. Nucleic Acids Res., 44, D869–D876. Huh,J., Ki,C.S., Sung,K., Choi,S.H. et al. (2015) Analysis of 39. Sveinbjornsson,G., Albrechtsen,A., Zink,F., Gudjonsson,S.A., protrusio acetabuli using a CT-based diagnostic method in korean Oddson,A., Másson,G., Holm,H., Kong,A., Thorsteinsdottir,U., patients with marfan syndrome: Prevalence and association with Sulem,P. et al. (2016) Weighting sequence variants based on their other manifestations. J. Korean Med. Sci., 30, 1260–1265. annotation increases power of whole-genome association studies. 56. Köhler,S. (2018) Improved ontology-based similarity calculations Nat. Genet., 48, 314–317. using a study-wise annotation model. Database, 2018, 40. Bastarache,L., Hughey,J.J., Hebbring,S., Marlo,J., Zhao,W., Ho,W.T., doi:10.1093/database/bay026. Van Driest,S.L., McGregor,T.L., Mosley,J.D., Wells,Q.S. et al. (2018) 57. Robinson,P.N., Köhler,S., Bauer,S., Seelow,D., Horn,D. and Phenotype risk scores identify patients with unrecognized Mendelian Mundlos,S. (2008) The Human Phenotype Ontology: a tool for disease patterns. Science, 359, 1233–1239. annotating and analyzing human hereditary disease. Am. J. Hum. 41. Son,J.H., Xie,G., Yuan,C., Ena,L., Li,Z., Goldstein,A., Huang,L., Genet., 83, 610–615. Wang,L., Shen,F., Liu,H. et al. (2018) Deep phenotyping on 58. Vasilevsky,N.A., Foster,E.D., Engelstad,M.E., Carmody,L., electronic health records facilitates genetic diagnosis by clinical Might,M., Chambers,C., Dawkins,H.J.S., Lewis,J., Della exomes. Am. J. Hum. Genet., 103, 58–73. Rocca,M.G., Snyder,M. et al. (2018) Plain-language medical 42. Segal,M.M., Rahm,A.K., Hulse,N.C., Wood,G., Williams,J.L., vocabulary for precision diagnosis. Nat. Genet., 50, 474–476. Feldman,L., Moore,G.J., Gehrum,D., Yefko,M., Mayernick,S. et al. 59. Lewis,J., Snyder,M. and Hyatt-Knorr,H. (2017) Marking 15 years of (2017) Experience with integrating diagnostic decision support the genetic and rare diseases information center. Transl. Sci. Rare software with electronic health records: Benefits versus risks of Dis., 2, 77–88. information sharing. EGEMS, 5, 23. 60. Girdea,M., Dumitriu,S., Fiume,M., Bowdin,S., Boycott,K.M., 43. Smith,C.L., Goldsmith,C.-A.W. and Eppig,J.T. (2005) The Chénier,S., Chitayat,D., Faghfoury,H., Meyn,M.S., Ray,P.N. et al. Mammalian Phenotype Ontology as a tool for annotating, analyzing (2013) PhenoTips: Patient phenotyping software for clinical and and comparing phenotypic information. Genome Biol., 6, R7. research use. Hum. Mutat., 34, 1057–1065. 44. Haendel,M.A., Balhoff,J.P., Bastian,F.B., Blackburn,D.C., 61. Watkins,X., Garcia,L.J., Pundir,S., Martin,M.J. and UniProt Blake,J.A., Bradford,Y., Comte,A., Dahdul,W.M., Dececchi,T.A., Consortium. (2017) ProtVista: visualization of protein sequence Druzinsky,R.E. et al. (2014) Unification of multi-species vertebrate annotations. Bioinformatics, 33, 2040–2041. anatomy ontologies for comparative biology in Uberon. J. Biomed. 62. Bauer,S., Köhler,S., Schulz,M.H. and Robinson,P.N. (2012) Bayesian Semantics, 5, 21. ontology querying for accurate and noise-tolerant semantic searches. 45. Bard,J., Rhee,S.Y. and Ashburner,M. (2005) An ontology for cell Bioinformatics, 28, 2502–2508. types. Genome Biol., 6, R21. 63. Jéru,I., Duquesnoy,P., Fernandes-Alnemri,T., Cochet,E., Yu,J.W., 46. Meehan,T.F., Conte,N., West,D.B., Jacobsen,J.O., Mason,J., Lackmy-Port-Lis,M., Grimprel,E., Landman-Parker,J., Hentgen,V., Warren,J., Chen,C.K., Tudose,I., Relac,M., Matthews,P. et al. (2017) Marlin,S. et al. (2008) Mutations in NALP12 cause hereditary Disease model discovery from 3,328 gene knockouts by The periodic fever syndromes. Proc. Natl. Acad. Sci. U.S.A., 105, International Mouse Phenotyping Consortium. Nat. Genet., 49, 1614–1619. 1231–1238.