=Paper=
{{Paper
|id=Vol-2037/paper44
|storemode=property
|title=eMiRo: An Ontology-Based System for Clinical Data Integration and Analysis
|pdfUrl=https://ceur-ws.org/Vol-2037/paper_44.pdf
|volume=Vol-2037
|authors=Pietro Cinaglia,Pierangelo Veltri,Mario Cannataro
|dblpUrl=https://dblp.org/rec/conf/sebd/CinagliaV017
}}
==eMiRo: An Ontology-Based System for Clinical Data Integration and Analysis==
https://ceur-ws.org/Vol-2037/paper_44.pdf
eMiRo: an ontology-based system for clinical
data integration and analysis
Discussion Paper
Pietro Cinaglia, Pierangelo Veltri, and Mario Cannataro
Department of Medical and Surgical Science, University of Catanzaro
{surname}@unicz.it
Abstract. Biological experiments and bioinformatics analysis are pro-
ducing large datasets of omics data (e.g. genomics, proteomics, interac-
tomics, etc.), stored in diversified sources and databases. Biological data
and bioinformatics results can be related to various information, such as
clinical data (e.g. cancer stage) or environment data (e.g. place where a
patient lives), but this requires novel data integration mechanisms and
analysis algorithms. Novel data structures are needed for data integra-
tion, while efficient algorithms are necessary for managing integrated
data and to analyze results in order to extract knowledge and underline
correlation with environmental factors. In this paper an ontology-based
system for clinical data integration and analysis is presented. The sys-
tem, called eMiRo (Electronic MedIcal RecOrd), includes geo-reference
features for epidemiological analysis based on geographical and clinical
information. Using eMiRo, a physician is able to handle and integrate bi-
ological data; moreover, the system supports the retrieval of additional
information such as ontology terms and information about genes and
diseases. When conducting a study, for each gene relevant to the study,
biological function and relations with other genes as well as involvement
in biological processes is reported.
1 Introduction
The growth of biological resources and information led to the storage of large
data, such as: physiological processes, gene patterns, disease-gene associations,
clinical trials; in this context it is necessary the development of bioinformatics
tools for data management in order to index heterogeneous resources in com-
mon structures and to provide a comprehensive view of the state of knowledge.
In recent years, many bioinformatics studies have been focused on gene role in
biological processes for preventing and treating diseases. Tools for heterogeneous
data integration and association are developed to improve knowledge in experi-
ments related to gene samples from Microarray and Next Generation Sequencing
(NGS) technologies. In this scenario it is necessary to develop algorithms able
to merge data from multiple sources, with heterogeneous models and formats,
to obtain a global information which satisfies the requirements formulated by
users (e.g. researchers). The literature presents various solutions to improve and
to automatize gene data analysis, for example [1] presents a R-based tool for
miRNA data analysis and correlation with clinical ontologies related to a study
about modeling and management of biological data. Ontologies are used in com-
puter science for knowledge management; its structure is based on information
related to each other by Terms (e.g. concepts, processes, and methods). In bi-
ology, Gene Ontology (GO) [2] is used by many algorithms and web systems to
retrieve biological information such as: molecular functions, biological processes,
and subcellular localizations in drug discovery. Analogously, Disease Ontology
(DO) [3] correlates human diseases through the cross-mapping of medical vocab-
ularies (e.g. MeSH, ICD, NCI Thesaurus, SNOMED and OMIM); furthermore,
it allows the examination and comparison of genetic variations, phenotypes,
proteins, and drugs. In genomics, the data heterogeneity is highly grown in con-
junction with the new experimental platforms (e.g. microarray experiments) and
consequently data size, type and structure is greatly diversified. Data-integration
techniques are crucial to obtain an interdisciplinary view which allows, for exam-
ple, a biologist to retrieve information of interest from large and heterogeneous
dataset [4]. Some applications are able to manage several resources (e.g. phys-
iology, pharmacology, clinical data) using ontologies [5] [6] or correlating con-
cepts with information through integration process based on semantic-analysis
[7] [8]. In OAHG [9] a correlation among human protein-coding genes (PCGs),
miRNAs, and lncRNAs is established in order to generate a comprehensive func-
tional annotation resources; for this purpose, it uses multi-level ontologies, such
as: Gene Ontology, Disease Ontology, and Human Phenotype Ontology (HPO)
[10]. The semantics analysis of biomedical data is often necessary to integrate
the information from sources that use different schemes as in ontology databases,
for example: SNOMED CT (Clinical Terms), ICD-9 and ICD-10 (International
Classification of Disease), and Human Phenotype Ontology (HPO) [10]. In re-
cent years the interest of Geographic Information System (GIS) technologies
are growing in biomedical studies in order to solve health issues, such as: to
evaluate the prevalence of diseases, to plan health interventions for epidemiolog-
ical studies, to perform a geographical analysis for monitoring of population for
prevention purposes. In [11] the biological information are integrated with GIS
geographic data for the handling and management of microbial genome data.
In this paper we present a system designed to manage clinical and biological
data, and to correlate such data by means of available ontologies. The system,
called eMiRo, has been developed to manage clinical and genomic data and
integrates them using ontologies and genomic information from external sources.
Moreover, eMiRo includes a geographical feature for epidemiological analysis.
2 The eMiRo System
We design a system starting by the following requirements: clinical and biological
data management, bioinformatics results from ontologies and semantics data,
support for geographical data. The actors identified for the system are listed
below:
– Physician: all features are available, it is able to manage data patients (us-
ing a model configured by the administrator), performs query to filter the
informations, consults the geographical analysis results, and requires data
integration features for a specific disease; custom access modes are definable
for this profile to create sub-profiles.
– Administrator: refers to the user which is able to handle system and its
features, as well as user profiles, component configuration, and data.
– Client: identified with a person or service (e.g. an algorithm) which accessed
the data-integration function using the genieR as web-service.
2.1 Architecture
System may be roughly divided into three main components, briefly:
– eMiRo (or ’eMiRo-component’, to distinguish this from the entire system
called with the same name): it handles the biological data and the interac-
tions with users, as well as data exchanges with other modules (genieR and
geoP);
– genieR: it concerns data analysis and data-integration features, also sup-
ports retrieving and management of unstructured information from external
sources (e.g. ontologies);
– geoP: it is related to geographical features: analysis, and geographic coordi-
nates managing (e.g. it converts raw text into coordinates using the Google
Maps API).
Figure 1 shows the interactions among the components; external sources and
API calling are also shown.
eMiRo-component is the basic module of the system; it is in charge of the
following functions:
– graphical front-end management;
– querying;
– biological data handling;
– database connection and components interaction;
– login, security and user roles management;
– analytics.
User is able to access the system through a Web-Panel which, essentially, con-
sists in a set of Graphical User Interfaces (GUI) that are generated dynamically
in according to a predefined model for the information required by user. Geo-
graphical data are shown to user using a Map-View (from Google Maps API),
and a summary table-view to show only the information of interest for the user
(e.g. a physician).
Google Maps API
Genomic
& Ontology Geocoder
Database
genieR
Web Datastore
Service
eMiRo
WebApp
geoP
Geocoder
Database
Fig. 1. System deployment
genieR is designed as a web-service to allow external clients (e.g. algorithms
or other information systems) to invoke its public functions. This components
concerns the data-integration features; specifically it unifies in a custom structure
the information given by several external sources. A mapping between Disease
Ontology and Gene Ontology is used to integrate annotations in order to obtain
the gene-disease associations.
Therefore, it is used for the management and the integration of data from
remote genomic and ontology sources. eMiRo supports integration from: Disease
Ontology [3], Gene Ontology [2] and DisGenet [12] (”curated gene-disease asso-
ciation” version); others could be added in future updates. The information are
structured as objects that structure data using lists and hashmap; this approach
allows the management of unstructured data; it is important to emphasize that
this component is implemented for deployment on Google App Engine Environ-
ment.
genieR main sub-components are:
– GenierServlet: it consists in two modules, dealing respectively to handle re-
quests from clients (even Web-Panel accesses genieR as an external resource)
and to structure the output in according to JSON format in order to make
faster the data stream and easy the parsing operations.
– Datastore: it periodically retrieves and parse the information from supported
external sources and organizes these in a structure compatible with the
eMiRo data model.
– DataIntegrator: it is the core module of genieR; it takes care to integrate
information in Datastore and to generate an output for the client.
In summary, a physician is able to choose a disease, from a list, to obtain ad-
ditional information, such as: ontological integration, genes involved for the dis-
ease, and meta-data for each gene founded (e.g. description, biological function,
biological processes, and relations with other genes). Main commands supported
by genieR are ’doget’ and ’dolist’ (’do’: Disease-Object):
– former returns information obtained from data-integration action for a spe-
cific Disease Ontology ID (DOID);
– latter gives the list of diseases that system supports for the integration op-
eration.
geoP is able to convert addresses into geographical coordinates using Google
Geocoders API; using this approach a geographical coordinates management
through GIS systems is allowed. Furthermore, the system supports analytics
features to generate statistical information from geographical and biological data
mapping (for statistical analysis are supported a set of parameter, such as: range
of values, and threshold level).
2.2 Implementation details
eMiRo system is designed for dynamic creation of reports referred to biologi-
cal information management; this solution allows users to create new examina-
tion and handling data models for existing ones. Data-Type supported by the
Database Engine are available during models creation.
A ’model’ is composed by a set of information related to a graphic layout
and its low-level parameters (e.g. field-name, field-data-type, bounds for graphic
elements, and the order of the fields in the resulting GUI), furthermore the
persistence of models is granted by a Relational Database Management System
(RDBMS). In according to this dynamic solution the Figure 2 shows the data
structure for a generic examination (named ’Report-1’) which consists of three
fields (’date’ is a default field whose representation is not required); Figure 2-A
and 2-D are referred to a specific examination for a given patient; while, 2-B
and 2-C contain information related to the model for GUI generation: fields,
data-type, and generic meta-data (e.g. field-order, and field-width in pixel). To
implement the system are chosen multiple programming languages related to
the purpose for each component; this approach optimizes the deployments and
the user experience. genieR and geoP are implemented using Servlet technology
to deploy Web-Applications [13] and to expose their features as a service (in
according with server policy configuration), for genieR the Google Cloud App
Engine SDK is also integrated. eMiRo-component has been developed using the
PHP and HTML5 languages, the latter extended using Bootstrap and jQuery
frameworks in order to support graphical interface with dynamic elements and
a responsive design based on CSS and JavaScript. Oracle MySQL has been
chosen as RDBMS for biological and geographic data persistence; in addition, it
contains information about the structures and the models used by system during
GUI generation.
Report-1
Date 2016-07-27
Field-1 test
Field-2 13.13
Field-3 13.0
Save
or
A
id date reportsid patientid userid logtime
1 2016-07-27 1 1 1 2016-07-27 13:13
B
id name description attachmentModule
1 Report-1 report for test 0 [NO]
C
id label type description size order reportsid
1 field-1 3 [text] description-1 0 [default] 1 1
2 field-2 1 [double] description-2 35 [px] 2 1
3 field-3 1 [double] description-3 0 3 1
D
id value fieldsid reportslistid
1 test 1 1
2 13.13 2 1
3 13.0 3 1
Fig. 2. Data model (low level representation)
2.3 Data-Integration: approach
During the test, the eMiRo data-integration component (named genieR) has
been used to retrieve the Open Biological Ontologies datasets from the server of
Gene-Ontology, Disease-Ontology, and Disgenet. OBO Foundry initiative sup-
ports the evolution of ontologies for biomedical data integration promoting on-
tologies designed to be interoperable and logically well-formed in order to in-
corporate accurate representations of biological information [14]. A dataset in
OBO format contains several tags (e.g. id, name, description, synonyms, cross-
references) organized in Terms linked to each other; thus, a dataset may be rep-
resented using, for example, a graph data structure. genieR pipeline is based on
two phases: preprocessing and analysis. During Preprocessing the data are inte-
grated and organized in a single structure: (i) the ontological datasets are down-
loaded on cloud-memory (the data can be refreshed when the sources update
their datasets); (ii) subsequently, each dataset is parsed to extract information
of interest, and it is collapsed within a single structure with others. Using this
approach, genieR is able to create a new large ontology (managed by the Data-
store component) that contains the heterogeneous information extracted from
each ontological dataset, and the novel relations found among these. Preprocess-
ing reduces the data in the cloud-memory to improve the overall performance
during the analysis. For the Analysis phase, genieR implements a component
named ’DataIntegrator’: when a client requests information about a disease the
DataIntegrator performs an analysis of its integrated data (that represents its
knowledge), and returns a results in JSON Format. Summing, genieR is able to
provide for a specific disease the relevant genes and for each the biological func-
tion, and its involvement in biological processes, and the relations with other
genes.
3 Security
In eMiRo, the user requests and the access policy are handled by web-application
and DBMS Engine [15]; first grants the user access based on his role, second
checks the read and write operations by the components, as well as the direct
requests to database. The communication between the components (internal and
external) is performed through HTTPS protocol which allows server authenti-
cation, privacy protection and maintaining data integrity, as well as checks data
exchanged between the parts and allows SSL/TLS encryption. Furthermore, ac-
count credentials are encrypted using the Secure Hash Algorithm (SHA) before
being stored within the database.
4 Conclusion
This paper presents the architecture and the security insights of a cloud-based
information system, named eMiRo, for biological data mapping; georeferencing
of patients is also supported. A light-beta version is used to handling biological
data by infectious diseases group of Magna Graecia University.
References
1. F. Cristiano and et al., “An r-based tool for mirna data analysis and correlation
with clinical ontologies,” Proceeding BCB ’14 Proceedings of the 5th ACM Confer-
ence on Bioinformatics, Computational Biology, and Health Informatics, 2014.
2. M. Ashburner, C. A. Ball, J. A. Blake, D. Botstein, H. Butler, J. M. Cherry, A. P.
Davis, K. Dolinski, S. S. Dwight, J. T. Eppig, M. A. Harris, D. P. Hill, L. Issel-
Tarver, A. Kasarskis, S. Lewis, J. C. Matese, J. E. Richardson, M. Ringwald, G. M.
Rubin, and G. Sherlock, “Gene ontology: tool for the unification of biology. The
Gene Ontology Consortium,” Nat. Genet., vol. 25, pp. 25–29, May 2000.
3. DO Official WebSite, Disease Ontology, 2016.
4. D. Gomez-Cabrero, I. Abugessaisa, D. Maier, A. Teschendorff, M. Merkenschlager,
A. Gisel, E. Ballestar, E. Bongcam-Rudloff, A. Conesa, and J. Tegner, “Data inte-
gration in the era of omics: current and future challenges,” BMC Syst Biol, vol. 8
Suppl 2, p. I1, 2014.
5. S. M. Wimalaratne, P. Grenon, R. Hoehndorf, G. V. Gkoutos, and B. de Bono, “An
infrastructure for ontology-based information systems in biomedicine: RICORDO
case study,” Bioinformatics, vol. 28, pp. 448–450, Feb 2012.
6. B. De Bono, R. Hoehndorf, S. Wimalaratne, G. Gkoutos, and P. Grenon, “The
RICORDO approach to semantic interoperability for biomedical data and models:
strategy, standards and solutions,” BMC Res Notes, vol. 4, p. 313, 2011.
7. K. M. Livingston, M. Bada, W. A. Baumgartner, and L. E. Hunter, “KaBOB:
ontology-based semantic integration of biomedical databases,” BMC Bioinformat-
ics, vol. 16, p. 126, 2015.
8. C. Pang, A. Sollie, A. Sijtsma, D. Hendriksen, B. Charbon, M. de Haan, T. de Boer,
F. Kelpin, J. Jetten, J. K. van der Velde, N. Smidt, R. Sijmons, H. Hillege, and
M. A. Swertz, “SORTA: a system for ontology-based re-coding and technical an-
notation of biomedical phenotype data,” Database (Oxford), vol. 2015, 2015.
9. L. Cheng, J. Sun, W. Xu, L. Dong, Y. Hu, and M. Zhou, “Oahg: an integrated
resource for annotating human genes with multi-level ontologies,” Sci Rep, vol. 6,
p. 34820, Oct 2016.
10. S. Kohler and at al., “The human phenotype ontology project: linking molecular
biology and disease through phenotype data,” Nucl. Acids Res., 2014.
11. S. Jelokhani-Niaraki, M. Tahmoorespur, Z. Minuchehr, and M. R. Nassiri, “An
Ontology-Based GIS for Genomic Data Management of Rumen Microbes,” Ge-
nomics Inform, vol. 13, pp. 7–14, Mar 2015.
12. J. Pinero, N. Queralt-Rosinach, A. Bravo, J. Deu-Pons, A. Bauer-Mehren,
M. Baron, F. Sanz, and L. I. Furlong, “DisGeNET: a discovery platform for the
dynamical exploration of human diseases and their genes,” Database (Oxford),
vol. 2015, 2015.
13. Oracle, Java Servlet Technology Overview, 2016.
14. B. Smith, M. Ashburner, C. Rosse, J. Bard, W. Bug, W. Ceusters, L. J. Goldberg,
K. Eilbeck, A. Ireland, C. J. Mungall, N. Leontis, P. Rocca-Serra, A. Ruttenberg,
S. A. Sansone, R. H. Scheuermann, N. Shah, P. L. Whetzel, and S. Lewis, “The
OBO Foundry: coordinated evolution of ontologies to support biomedical data
integration,” Nat. Biotechnol., vol. 25, pp. 1251–1255, Nov 2007.
15. G. Miklau and D. Suciu, “A formal analysis of information disclosure in data
exchange,” J. Comput. Syst. Sci, 2007.