=Paper=
{{Paper
|id=None
|storemode=property
|title=Analysis of Vaccine-related Networks using Semantic MEDLINE and the Vaccine Ontology
|pdfUrl=https://ceur-ws.org/Vol-1061/Paper6_vdos2013.pdf
|volume=Vol-1061
|dblpUrl=https://dblp.org/rec/conf/icbo/ZhangTHKL13
}}
==Analysis of Vaccine-related Networks using Semantic MEDLINE and the Vaccine Ontology==
https://ceur-ws.org/Vol-1061/Paper6_vdos2013.pdf
Analysis of Vaccine-related Networks using Semantic MEDLINE
and the Vaccine Ontology
Yuji Zhang1,*, Cui Tao1, Yongqun He2, Pradip Kanjamala1, Hongfang Liu1
1
Department of Health Sciences Research, Mayo College of Medicine, Rochester, MN 55905, USA
2
Unit of Laboratory of Animal Medicine, University of Michigan, Ann Arbor, MI 48109, USA
ABSTRACT medicines and vaccines were developed based on previ-
A major challenge in the vaccine research has been to ous marketed products. This suggested that drug reposi-
identify important vaccine-related networks and logically tioning has drawn great attention from the both industry
explain the results. In this paper, we showed that network- and academic institutes (Graul, Sorbera et al. 2010).
based analysis of vaccine-related networks can discover However, many of these drug repositioning have been
the underlying structure information consistent with that serendipitous discoveries (Ashburn and Thor 2004) or on
captured by the Vaccine Ontology and propose new hy- observable clinical phenotypes, which are lack of system-
potheses for vaccine disease or gene associations. First, a atic ways to identify new targets. Recent research has
vaccine-vaccine network was inferred using a bipartite shown that bioinformatics-based approaches can aid to
network projection strategy on the vaccine-disease net- reposition drugs based on the complex relationships
work extracted from the Semantic MEDLINE database. In among drugs, diseases and genes (Liu, Fang et al. 2013).
total, 76 vaccines and 573 relationships were identified to Such approaches can also be applied in the future vaccine
construct the vaccine network. The shortest paths between development.
all pairs of vaccines were calculated within the vaccine In recent years, high-throughput biological data and
network. The correlation between the shortest paths of computational systems biology approaches has provided
vaccine pairs and their semantic similarities in the Vac- an unprecedented opportunity to understand the disease
cine Ontology was then investigated. Second, a vaccine- etiology and its underlying cellular subsystems. Biologi-
gene network was also constructed, in which several im- cal knowledge such as drug-disease networks, and bio-
portant vaccine-related genes were identified. This study medical ontologies have accelerated the development of
demonstrated that a combinatorial analysis using literature network-based approaches to understanding disease etiol-
knowledgebase, semantic technology, and ontology is ogy (Ideker and Sharan 2008; Barabasi, Gulbahce et al.
able to reveal unidentified important knowledge critical to 2011) and drug action (network pharmacology) (Berger
biomedical research and public health and generate testa- and Iyengar 2009; Mathur and Dinakarpandian 2011).
ble hypotheses for future experimental verification. Such approaches could also be applied in the vaccine re-
search, aiming to investigate the vaccine-related associa-
1 INTRODUCTION tions derived from public knowledgebase such as
Vaccines have been one of the most successful public PUBMED literature abstracts. For example, a Vaccine
health interventions to date with most vaccine-preventable Ontology (VO)-based literature mining research was re-
diseases having declined in the United States by at least ported last year to study all gene interactions associated
95-99% (1994). However, vaccine development is getting with fever alone or both fever and vaccine (Hur, Ozgur et
more difficult as more complex organisms become vac- al. 2012). This study focused on the retrieval of gene-gene
cine targets. In recent years, drug repositioning has been associations based on their direct interactions in the con-
growing in last few years and achieved a number of suc- text of fever and vaccine. The centrality-based network
cesses for existing drugs such as Viagra (Goldstein, Lue approach (Ozgur, Vu et al. 2008) evaluated the level of
et al. 1998) and thalidomide (Singhal, Mehta et al. 1999). importance for each gene in extracted gene interaction
By definition, drug repositioning is the “process of find- network. Novel gene interactions were identified to be
ing new users outside the scope of the original medical essential in fever- or vaccine-related networks that could
indications for existing drugs or compounds” (Chong and not be found before. A similar VO and centrality-based
Sullivan 2007). In 2009, more than 30% of the 51 new literature mining approach was also used to analyse vac-
cine-associated IFN- gene interaction network (Ozgur,
*
To whom correspondence should be addressed: Xiang et al. 2011). Ball et al. compiled a network consist-
Zhang.Yuji@mayo.edu ing of 6,428 nodes (74 vaccines and 6,354 adverse events)
and more than 1.4 million interlinkages, derived from
1
Zhang et al.
Vaccine Adverse Event Reporting System (VAERS) (Ball In this paper, we proposed a network-based approach to
and Botsis 2011). This network demonstrated a scale-free investigate the underlying relationships among vaccines
property, in which certain vaccines and adverse events act in the context of the vaccine-related network derived from
as “hubs”. Such network analysis approaches complement Semantic MEDLINE. The distances of the vaccines were
current statistical techniques by offering a novel way to further compared with their semantic similarities in the
visualize and evaluate vaccine adverse event data. How- VO. The results demonstrated that the structure infor-
ever, the relationships among different vaccines in the mation of vaccine network is consistent with that captured
context of vaccine-vaccine and vaccine-gene networks by VO. Such network-based analysis can serve as an in-
have not been well studied yet. A systematic level inves- dependent data resource to construct and evaluate bio-
tigation of such relationships will help us understand bet- medical ontologies. In addition, the vaccine-gene network
ter how vaccines are related to each other and whether was also constructed based on Semantic MEDLINE in-
such information can complement the existing knowledge formation, in which important vaccine-related genes were
such as VO. identified and further investigated by VO and related in-
To analyse the possible common protective immunity dependent resources.
or adverse event mechanisms among different vaccines, it The rest of the paper is organized as follows. Section 2
is critical to study all possible vaccine-vaccine and vac- introduces the data resources and the proposed network-
cine-gene associations using network analysis approaches. based framework. Section 3 illustrates the results generat-
The hypotheses behind this are: (1) if two vaccines have ed from each step in the proposed computational frame-
coupling relationship with common disease(s)/gene(s), work. Section 4 provides a thorough discussion of the
they are linked in the vaccine network; (2) the closer two results and concludes the paper.
vaccines are in the vaccine network, the more similar they
are in the context of literature knowledgebase, such as
Semantic MEDLINE (Rindflesch, Kilicoglu et al. 2011).
Fig. 1. Overview of the proposed framework. The proposed method consists of three steps: 1) Extraction of vaccine-related associations from Semantic
MEDLINE using ontology based terms; 2) Network based analyses to identify vaccine-vaccine associations and vaccine-gene associations; 3) Evaluation of
inferred vaccine-vaccine and vaccine-gene relationships using vaccine ontology hierarchical structure and literature validation.
Medicine (NLM) initiated project which provides a public
2 MATERIALS AND METHODS available database that contains comprehensive resources
In this section, we describe the data resources and prepro- with structured annotations for information extracted from
cessing method in this work. We then introduce our pro- more than 19 million MEDLINE abstracts. Since the
posed network-based approach for investigating vaccine- Semantic MEDLINE is a comprehensive resource that
related associations derived from PubMed literature ab- contains heterogeneous data with different features ex-
stracts. The evaluation of the discovered vaccine-vaccine tracted, our previous research has reorganized this data
and vaccine-gene relationships is conducted based on the source and optimized it for informatics analysis (Tao,
VO hierarchy and logical definitions. Fig. 1 illustrates the Zhang et al. 2012). Using the Unified Medical language
steps of the proposed approach. System (UMLS) semantic types and groups (2012), we
extracted unique associations among diseases, genes, and
2.1 Data Resources and preprocessing drugs, and represented them in six Resource Description
2.1.1. Data Resources Framework (RDF) graphs. In this paper, we used our op-
timized Semantic MEDLINE RDF data as the data source
In this study, we use Semantic MEDLINE as the data
to perform network analysis for vaccine-related networks.
resource to build the networks. Semantic MEDLINE
(Rindflesch, Kilicoglu et al. 2011) is a National Library of
2
Analysis of Vaccine-related Networks using Semantic MEDLINE and the Vaccine Ontology
Our RDF-based Semantic MEDLINE resource current- The vaccine-gene network was constructed by vaccine-
ly contains 843k disease-disease, 111k disease-gene, gene associations extracted from the drug-gene associa-
1277k disease-drug, 248k drug-gene, 1900k drug-drug, tions in our RDF-based data resource. The important vac-
and 49k gene-gene associations. Since this resource con- cine-related genes were identified by their significant
tains high-level terms (e.g., gene, protein, disease) that are higher node degree compared to other vaccine/gene in the
not useful for network analysis, we further manually fil- same network. The Cytoscape tool (Smoot, Ono et al. 2011)
tered out these terms using the following strategy. For was selected to visualize the network. Cytoscape is an
disease terms, we only included those terms that are in- open-source platform for integration, visualization and
cluded in ICD9. For gene terms, we only include those analysis of biological networks. Its functionalities can be
terms that have an Entrez gene ID. extended through Cytoscape plugins. Scientists from dif-
ferent research fields have contributed over 160 useful
2.1.2. Data extraction
plugins so far. These comprehensive features allow us to
We identified those associations relevant to vaccines only. perform thorough network level analyses and visualiza-
Specifically, vaccine terms were identified based on tion of our association tables, and integration with other
SNOMED CT (http://www.ihtsdo.org/snomed-ct). All the biological networks in the future.
terms under the SNOMED CT term Vaccine (CUI:
C0042210) were first extracted. A manual review by 3
experts further removed those common terms (e.g., bacte-
ria vaccine) or animal vaccine terms.
2.2 Network analysis of Vaccine Network
2.2.1. Projection of bipartite vaccine-disease network
In graph theory, a bipartite network is composed of two
non-overlapping sets of nodes and links that connect one
node in the first node set with one node in the second
node set. The properties of bipartite networks are often
investigated by considering the one-mode projection of
the bipartite network. The one-mode projection network
can be created by connecting two nodes in the same node
set if they have at least one common neighboring node in
the other node set. For instance, the vaccine-disease asso-
ciation network is one bipartite network: vaccines and
diseases constitute two node sets, and links are generated
between vaccine and disease if they are associated in the
Fig. 2. The heat map of vaccine-vaccine associations. The shortest path
Semantic MEDLINE. Therefore, the vaccine-vaccine
matrix of all vaccine pairs was used to generate the heat map. Each row
network can be investigated by projecting vaccine-disease
(column) represents a vaccine term. The color scale represents the short-
associations to vaccine-vaccine associations, in which two
est path between any vaccine pair.
vaccines are connected if they are associated with at least
one same disease. In this work, all links were generated 2.3 Analysis of vaccine groups using VO
based on the associations extracted from Semantic The community-based Vaccine Ontology (VO) has in-
MEDLINE as described in Section 2.1. A vaccine-vaccine cluded over 4,000 vaccine-specific terms, including all
network was generated consisting of all the links identi- licensed human and veterinary vaccines currently used in
fied in vaccine-disease associations. the USA. Logical axioms have been defined in VO to
2.2.2. Network distance between vaccines represent the relations among vaccine terms (Ozgur,
The distance between any two vaccines in the vaccine Xiang et al. 2011). The Semantic MEDLINE analysis
network was calculated as the length of the shortest path uses SNOMED terms to represent various vaccines. VO
between them (Fekete, Vattay et al. 2009). The hierar- has established automatic mapping between SNOMED
chical clustering analysis was performed on the distance vaccine terms and VO terms. Based on the mapping, we
matrix of all vaccines (Guess and Wilson 2002). A heat first extracted all vaccine terms from the Semantic
map was generated based on the clustering analysis re- MEDLINE and mapped to VO. The ontology term re-
sults. trieval tool OntoFox (Xiang, Courtot et al. 2010) was then
applied to obtain the hierarchies of the total vaccines or
2.2.3. Analysis of vaccine-gene network subgroups of the vaccines identified in this study.
3
Zhang et al.
3 RESULTS using an ontology reasoner was used to infer that the DTP
is also Bordetella pertussis vaccine (i.e., Pertussis vac-
3.1 The overall network view cine) (Fig. 3). It is likely that the association of the com-
In total, 76 vaccines, annotated by the SNOMED CT term bination vaccine DTP with neurological disorder is at
Vaccine (CUI: C0042210), were used to extract related least partially due to the Pertussis vaccine component.
vaccine-disease and vaccine-gene associations from the Our study also identified many other diseases associat-
drug-disease and drug-gene association tables respective- ing with different vaccines. For example, five vaccines
ly. In the vaccine-disease network, there were 1127 nodes (e.g., pertussis vaccine) were found to be associated with
(178 vaccines and 949 diseases) and 1741 vaccine-disease various types of antimicrobial susceptibility, and eight
associations. In the vaccine-gene network, there were 170 vaccines (e.g., influenza vaccine) have been co-studied
nodes (85 vaccines and 85 genes) and 94 vaccine-gene with patients having the asthma condition. Due to the
associations. One vaccine network was generated by the relative poor annotation of the vaccine data in the Seman-
projection of the vaccine-disease bipartite network, con- tic MEDLINE system, the vaccines identified in the se-
sisting of 76 vaccines and 573 associations. This vaccine mantic analysis were poorly classified. The incorporation
network was then used to analyze the vaccine relation- of VO in the study clearly classifies these vaccines, lead-
ships. The derived vaccine-gene network was also inves- ing to better understanding of the result of the Semantic
tigated by the VO knowledge. MEDLINE analysis.
3.2 Analysis of vaccine network 2) This group of vaccines, including Q fever vaccine,
Parvovirus vaccine, and Tick-borne encephalitis vaccine,
Fig. 2 showed a heat map of hierarchical analysis results, is associated with the common disease “Delayed Hyper-
providing a direct visualization of potential vaccine- sensitivity”. Delayed type reactions may occur at days
vaccine associations. Here we selected four relatively big after vaccination and often raise serious safety concerns.
vaccine-vaccine association groups on the diagonal from Delayed hypersensitivity is not antibody-mediated but
Fig. 2 and explain them in detail: rather is a type of cell-mediated response. The study of
1) This group contains 18 very widely-studied vac- common vaccines and related gene and pathway features
cines. Many interesting results are obtained from the related to the delayed reaction will help to reveal the
analysis of this group of vaccine-disease-vaccine associa- cause of DTR and eventually prevent it. While these vac-
tions. For example, the results from this group show that cines are developed against different bacterial or viral
influenza vaccines and Rabies vaccines have been associ- diseases, there may be similarities among these vaccines,
ated with the induction of a severe adverse event Guillain- such as common vaccine ingredients (e.g., adjuvant) and a
Barré syndrome (GBS) (Hemachudha, Griffin et al. 1988; shared target to some common biological pathway in hu-
Hartung, Keller-Stanislawski et al. 2012). GBS is a rare mans. An identification of these common features may
disorder in which a person’s own immune system damag- indicate a common cause of the DTR.
es their nerve cells, causing muscle weakness and some- 3) This group of vaccines is associated with the com-
times paralysis. This group also includes five other vac- mon disease “Mumps”. The vaccines in this group include
cines associating with nervous system disorder, including Mumps Vaccine, “measles, mumps, rubella, varicella
Pertussis Vaccine (Wardlaw 1988), Diphtheria-Tetanus- vaccine”, and “diphtheria-tetanus-pertussis-haemophilus b
Pertussis Vaccine (Corkins, Grose et al. 1991), Hepatitis conjugate vaccine” (DTP-Hib). The first two vaccines
B Vaccines (Comenge and Girard 2006), Chickenpox protect against Mumps. DTP-Hib was compared with a
Vaccine (Bozzola, Tozzi et al. 2012), and Poliovirus Vac- Mumps vaccine in a study (Henderson, Oates et al. 2004).
cine (Friedrich 1998; Korsun, Kojouharova et al. 2009). 4) This vaccine group consists of seven vaccines (e.g.,
As shown by a VO hierarchical structure layout (Fig. 3), Brucella abortus vaccine and bovine rhinotracheitis vac-
these seven vaccines belong to different bacterial or viral cine) with direct associations between them. They are all
vaccines. The Diphtheria-Tetanus-Pertussis vaccine associated with the common term “calve” in the literature
(DTP) is a combination vaccine that contains three indi- abstracts. Since “calve disease” has a synonym “Scheu-
vidual vaccine components, including a Pertussis vaccine. ermann’s Disease”, these vaccines have all been linked to
DTP is asserted in VO as a subclass of “Diphtheria- “Scheuermann’s Disease”. This is due to the ambiguity of
Tetanus vaccine”. Different from SNOMED, VO logical- the Nature Language Processing (NLP) process. This can
ly defines vaccines based on their relation to the pathogen be improved by future improvement of the disambiguity
organisms defined in the NCBI_Taxon ontology. Since capacity of NLP tools.
multiple inheritances are not used in VO, an inference
4
Analysis of Vaccine-related Networks using Semantic MEDLINE and the Vaccine Ontology
text of vaccine network extracted from PubMed literature
abstracts. The investigations of vaccine-vaccine, vaccine-
disease, and vaccine-gene networks demonstrate that such
literature-based associations can be better analyzed using
VO and such a combinatorial analysis is able to reveal the
association patterns and identify new knowledge. The
identified vaccine-vaccine associations based on vaccine-
disease distance analysis are consistent with their VO
categories and often lead to the generation of new hy-
potheses. Our studies discovered some novel vaccine-
vaccine relationships by discovering a group of vaccines
associated with some common diseases as demonstrated
Fig. 3. The VO hierarchical structure of the seven vaccines associating in the heat map analysis in the Results section. Due to the
with neurological disorder. A reasoning process assigned the Diphtheria- incompleteness of such information existing in the litera-
Tetanus-Pertussis vaccine under Bordetella pertussis vaccine. The Pro- ture abstracts, such vaccine-vaccine associations need
tégé-OWL editor 4.2 was used for the figure generation. further validation in independent databases or through
3.3 Vaccine-gene network future experimental studies. For example, while our anal-
ysis reveals associations between a group of vaccines and
In the vaccine-gene network, many genes were found to neurological adverse events, it is noted that the evidences
interact with different vaccines. For example, our study of these associations, although reported by some PubMed
identified that CD40LG (CD40 ligand) is closely associ- abstracts, are not necessarily commonly acknowledged
ated with five vaccines: Diphtheria toxoid vaccine, Chol- (Sarntivijai, Xiang et al. 2012). More analysis may be
era vaccine, Tetanus vaccine, Chickenpox vaccine, and required for clarification.
inactivated poliovirus vaccine (Fig. 4). CD40LG plays an Future extensions of this work include: (1) integra-
important role in antigen presentation and stimulation of tion of more comprehensive vaccine-disease association
cytotoxic T lymphocytes (Kornbluth 2000). CD40LG can databases (e.g., VAERS system) to construct more com-
also be used in rational vaccine adjuvant design plete vaccine-related networks; (2) generation of vaccine-
(Kornbluth and Stone 2006). Our finding confirms the related gene network by extending the neighbour genes of
important role of CD40LG and provides specific details vaccine-associated genes; (3) network-based investigation
on how this immune factor interacts with various bacterial of the relationships among vaccines and other drugs using
and viral vaccines. vaccine-drug associations; (4) investigation on possible
ways to improve the network by assigning weights or
confident rates to different types of associations or associ-
ations from different sources.
ACKNOWLEDGEMENTS
This project was supported by the National Institute of
Health grants 5R01LM009959-02 to HL, and
R01AI081062 to YH.
REFERENCES
(2012). "The UMLS Semantic Groups." from
http://semanticnetwork.nlm.nih.gov/SemGroups/.
Ashburn, T. T. and K. B. Thor (2004). "Drug repositioning:
identifying and developing new uses for existing
Fig. 4. A vaccine-gene subnetwork. The associations between vaccines
drugs." Nature reviews. Drug discovery 3(8): 673-683.
and related genes were visualized by the Cytoscape tool (Smoot, Ono et
Ball, R. and T. Botsis (2011). "Can network analysis improve
al. 2011). Purple rectangular node represents vaccine, and yellow circle
pattern recognition among adverse events following
node represents gene.
immunization reported to VAERS?" Clinical
pharmacology and therapeutics 90(2): 271-278.
4 DISCUSSIONS AND FUTURE WORK
Barabasi, A. L., N. Gulbahce, et al. (2011). "Network medicine:
In this paper, we proposed a novel network-based ap- a network-based approach to human disease." Nature
proach to investigate the vaccine relationships in the con- reviews. Genetics 12(1): 56-68.
5
Zhang et al.
Berger, S. I. and R. Iyengar (2009). "Network analyses in Kornbluth, R. S. and G. W. Stone (2006). "Immunostimulatory
systems pharmacology." Bioinformatics 25(19): 2466- combinations: designing the next generation of
2472. vaccine adjuvants." Journal of leukocyte biology
Bozzola, E., A. E. Tozzi, et al. (2012). "Neurological 80(5): 1084-1102.
complications of varicella in childhood: case series Korsun, N., M. Kojouharova, et al. (2009). "Three cases of
and a systematic review of the literature." Vaccine paralytic poliomyelitis associated with type 3 vaccine
30(39): 5785-5790. poliovirus strains in Bulgaria." Journal of medical
Chong, C. R. and D. J. Sullivan, Jr. (2007). "New uses for old virology 81(9): 1661-1667.
drugs." Nature 448(7154): 645-646. Liu, Z., H. Fang, et al. (2013). "In silico drug repositioning:
Comenge, Y. and M. Girard (2006). "Multiple sclerosis and what we need to know." Drug discovery today 18(3-
hepatitis B vaccination: adding the credibility of 4): 110-115.
molecular biology to an unusual level of clinical and Mathur, S. and D. Dinakarpandian (2011). "Drug repositioning
epidemiological evidence." Medical hypotheses 66(1): using disease associated biological processes and
84-86. network analysis of drug targets." AMIA ... Annual
Corkins, M., C. Grose, et al. (1991). "Fatal pertussis in an Iowa Symposium proceedings / AMIA Symposium. AMIA
infant." Iowa medicine : journal of the Iowa Medical Symposium 2011: 305-311.
Society 81(9): 383-384. Ozgur, A., T. Vu, et al. (2008). "Identifying gene-disease
Fekete, A., G. Vattay, et al. (2009). "Shortest path discovery of associations using centrality on a literature mined
complex networks." Physical review. E, Statistical, gene-interaction network." Bioinformatics 24(13):
nonlinear, and soft matter physics 79(6 Pt 2): 065101. i277-285.
Friedrich, F. (1998). "Neurologic complications associated with Ozgur, A., Z. Xiang, et al. (2011). "Mining of vaccine-
oral poliovirus vaccine and genomic variability of the associated IFN-gamma gene interaction networks
vaccine strains after multiplication in humans." Acta using the Vaccine Ontology." Journal of biomedical
virologica 42(3): 187-194. semantics 2 Suppl 2: S8.
Goldstein, I., T. F. Lue, et al. (1998). "Oral sildenafil in the Prevention, C. f. D. C. a. (1994). "Reported vaccine-preventable
treatment of erectile dysfunction. Sildenafil Study diseases--United States, 1993, and the childhood
Group." N Engl J Med 338(20): 1397-1404. immunization initiative." MMWR. Morbidity and
Graul, A. I., L. Sorbera, et al. (2010). "The Year's New Drugs & mortality weekly report 43(4): 57-60.
Biologics - 2009." Drug news & perspectives 23(1): 7- Rindflesch, T. C., H. Kilicoglu, et al. (2011). "Semantic
36. MEDLINE: An advanced information management
Guess, M. J. and S. B. Wilson (2002). "Introduction to application for biomedicine." Information Services &
hierarchical clustering." Journal of clinical Use 31(1/2): 15-21.
neurophysiology : official publication of the American Sarntivijai, S., Z. Xiang, et al. (2012). "Ontology-based
Electroencephalographic Society 19(2): 144-151. combinatorial comparative analysis of adverse events
Hartung, H. P., B. Keller-Stanislawski, et al. (2012). "[Guillain- associated with killed and live influenza vaccines."
Barre syndrome after exposure to influenza]." Der PLoS One 7(11): e49941.
Nervenarzt 83(6): 714-730. Singhal, S., J. Mehta, et al. (1999). "Antitumor activity of
Hemachudha, T., D. E. Griffin, et al. (1988). "Immunologic thalidomide in refractory multiple myeloma." N Engl J
studies of rabies vaccination-induced Guillain-Barre Med 341(21): 1565-1571.
syndrome." Neurology 38(3): 375-378. Smoot, M. E., K. Ono, et al. (2011). "Cytoscape 2.8: new
Henderson, R., K. Oates, et al. (2004). "General practitioners' features for data integration and network
concerns about childhood immunisation and visualization." Bioinformatics 27(3): 431-432.
suggestions for improving professional support and Tao, C., Y. Zhang, et al. (2012). Optimizing semantic
vaccine uptake." Communicable disease and public MEDLINE for translational science studies using
health / PHLS 7(4): 260-266. semantic web technologies. Proceedings of the 2nd
Hur, J., A. Ozgur, et al. (2012). "Identification of fever and international workshop on Managing interoperability
vaccine-associated gene interaction networks using and compleXity in health systems. Maui, Hawaii,
ontology-based literature mining." Journal of USA, ACM: 53-58.
biomedical semantics 3(1): 18. Wardlaw, A. C. (1988). "Animal models for pertussis vaccine
Ideker, T. and R. Sharan (2008). "Protein networks in disease." neurotoxicity." The Tokai journal of experimental and
Genome research 18(4): 644-652. clinical medicine 13 Suppl: 171-175.
Kornbluth, R. S. (2000). "The emerging role of CD40 ligand in Xiang, Z., M. Courtot, et al. (2010). "OntoFox: web-based
HIV infection." Journal of leukocyte biology 68(3): support for ontology reuse." BMC research notes 3:
373-382. 175.
6