Brucella is a Gram-negative, facultative intracellular bacterium that causes brucellosis in humans and animals [1]. Brucella are taxonomically placed in the alpha-2 subdivision of the class Proteobacteria. Traditionally there are six species of Brucella based on the preferential host specificity: B. melitensis (goats), B. abortus (cattle), B. suis (swine), B. canis (dogs), B. ovis (sheep) and B. neotomae (desert mice). The first four species listed in decreasing order of severity are pathogenic to humans making brucellosis a zoonotic disease. These bacteria are also amenable for use in biological warfare and bio-terrorism. Recently, two new species B. cetaceae (cetacean) and B. pinnipediae (seal) have been described [2]. Complete genome sequences of 10 Brucella strains are currently available in the NCBI RefSeq database. Four genomes from B. abortus, B. melitensis, and B. suis have been extensively analyzed [3, 4, 5, 6]. While animal brucellosis vaccines are commercially available, there is no effective and safe human vaccine against virulent Brucella infections. Extensive studies on Brucella have recently been concentrated on understanding the mechanisms for protective Brucellaimmunity and the development of effective human brucellosis vaccines.

VIOLIN (http://www.violinet.org) is a web-based vaccine database and analysis system. VIOLIN contains general information on microbial pathogenesis, host ranges, and host protective immunity, as well as vaccine-specific information such as vaccine type, preparation method, genetically engineered genes, and host responses in various animal models. VIOLIN contains information about 454 vaccines and vaccine candidates for 73 pathogens. VIOLIN contains many bioinformatics tools for vaccine literature mining, vaccine data analysis and integration, and vaccine target prediction. For example, VIOLIN includes Vaxmesh and Vaxpresso programs that may be used to mine vaccine literature based on MeSH controlled vocabulary and natural language processing (NLP), respectively. Dr. Yongqun He, the founder of the VIOLIN initiated and leads community-based development of the Vaccine Ontology to support vaccine integration and automated reasoning. A web-based vaccine target prediction program Vaxign available in VIOLIN is used to predict vaccine targets based on genome sequence analysis using a reverse vaccinology strategy.

As of May 13, 2010, more than 2,000 Brucella vaccine-related literature papers were searchable in PubMed, and 10 Brucella genomes have been published in the NCBI RefSeq database. To support Brucella vaccine research and development, we systematically curated from the literature existing Brucella vaccine information, which are stored in VIOLIN for query and further analyses. Different VIOLIN tools are also used to analyze Brucellavaccines and predict new vaccine targets.

A large number of vaccine-related databases exist on the web. There are many government-supported vaccine databases. For example, the Centers for Disease Control and Prevention (CDC) maintain a Vaccine Information Statements (VISs) system (http://www.cdc.gov/vaccines/pubs/vis/default.htm). The Center for Biologics Evaluation and Research (CBER) under the Food and Drug Administration (FDA) regulates vaccine products and posts relevant information in their vaccine site: http://www.fda.gov/cber/vaccines.htm. There is also a Vaccine Adverse Event Reporting System (VAERS, http://vaers.hhs.gov/), co-sponsored by FDA and CDC in USA. Many agent-specific databases are also available, for example, the HIV vaccine resource (http://www3.niaid.nih.gov/research/topics/HIV/vaccines/default.htm) created by the National Institute of Allergy and Infectious Diseases (NIAID) at the National Institutes of Health (NIH). Other vaccine resources include, the Vaccine Page: http://www.vaccines.org/), the Vaccine Resource Library (PATH, http://www.path.org/vaccineresources/), and the Immunization Action Coalition (http://www.immunize.org/). These databases primarily focus on available information concerning existing licensed vaccines and vaccine regulation. VIOLIN is unique in that it stores and analyzes research data concerning commercial vaccines and vaccines under clinical trial or in early stages of development [8].

The development of the Vaccine Ontology (VO) is a community effort and involves many experts in the vaccine and biomedical ontology communities [27]. With the large number of vaccine data types and publications available, VO is developed as an efficient strategy for vaccine data standardization, retrieval, and integration. VO makes it possible for computer programs to understand various vaccine types and research data associated with different vaccines. VO will also help to ensure that data is annotated in a way that ensures comparability. Therefore, VO-based software programs can be developed to support high throughput vaccine data processing and analysis. We are currently developing a VO-based literature mining and curation program that would increase the efficacy of vaccine literature mining and manual curation. The VO-based literature mining program will also relieve the burden of continuous database updating. VO will also be used to integrate all vaccine data in VIOLIN, making vaccine information exchange more efficient.

Compared to the traditional vaccine development approach that starts from a wet laboratory, reverse vaccinology begins with dry laboratory bioinformatics analysis, which makes the vaccine development more specific and efficient. Reverse vaccinology was first used by Rino Rappuoli in the development of a vaccine against serogroup B Neisseria meningitidis (MenB), the major cause of sepsis and meningitis in children and young adults [28]. Since then, this strategy has been applied to many other pathogens such as Bacillus anthracis [29], Streptococcus pneumoniae [30], and Mycobacterium tuberculosis [31]. While the criteria for vaccine prediction are known and many individual programs are available, it is still time consuming and requires expertise in these individual programs to predict vaccine targets using genome sequences. Vaxign is the first web-based automated pipeline that identifies potential vaccine targets based on the reserve vaccinology strategy [19]. Vaxign has been applied successfully to predict vaccine targets for uropathogenic E. coli [19]. This study demonstrated that Vaxign can predict novel Brucella vaccine targets. Experimental verification of many of these targets is currently under way. Vaxign also contains a program to predict immune epitopes that bind to MHC class I and II molecules in different animal species. Studies analyzing and ranking potential immune epitopes from predicted Brucellaproteins are in progress. Promising epitopes will be tested in a wet laboratory.

VIOLIN is also associated with other existing data resources. For example, many VIOLIN programs (e.g., Vaxign and Protegen) obtain Brucella genome sequences and share Brucella gene annotations with the web-based Pathogen-host Interaction Data Integration and Analysis System (PHIDIAS, http://www.phidias.us) [32]. PHIDIAS focuses on the analysis of pathogen-host interactions. Additionally, PHIDIAS contains the Brucella Bioinformatics Portal, a web-based portal with a special emphasis on Brucella genome annotation and literature mining [7]. PHIDIAS and BBP, also developed in our group, integrate more than 20 existing data resources. The close interaction between PHIDIAS/BBP and VIOLIN makes bioinformatics analysis of Brucellavaccines and vaccine targets more efficient.

VIOLIN currently includes vaccine data for 73 pathogens. The VIOLIN methods described for Brucella vaccine analysis in this report are generic and also feasible for vaccine studies for other pathogens. It is noted that Brucella is one of the most annotated pathogens among these 73 pathogens listed in VIOLIN. The vaccine information for many pathogens is not systematically annotated to the extent of Brucellavaccines. More work and collaborations with the research experts in these pathogens are necessary to curate and analyze vaccines and vaccine candidates for these pathogens.

VIOLIN provides manually curated Brucella vaccine data and ontology representation of these vaccines using the Vaccine Ontology (VO). Many tools are developed in VIOLIN to support literature mining and data curation. Examples of data stored in the VIOLIN database include protective Brucella antigens and host responses induced by different Brucella vaccines. Brucella vaccine targets may be predicted using the VIOLIN Vaxign program. Various Brucellavaccine data can be queried using user-friendly web query programs in VIOLIN. The VIOLIN approach is generic and can be used for analyses of vaccines against other pathogens and infection diseases.

Literature mining of Brucella vaccines using VIOLIN: The information of all PubMed papers associated with Brucella vaccine and vaccination were downloaded from the PubMed web service. The literature contents were processed using VIOLIN literature mining pipelines [8]. The processed results are available for users to analyzed using individual VIOLIN literature mining programs.

Bioinformatics curation of Brucella vaccines in VIOLIN: Brucella vaccine curation was performed on the VIOLIN web page using the Limix literature mining and curation system [7]. Limix allows data curators to submit data to the website, data reviewers to review and approve the submitted data, and eventual publication of high quality data. Specifically, a VIOLIN curator curates and compiles relevant information on vaccine information from peer-reviewed journals, books, and credible websites. The curated information is initially saved as a draft document and, when completed, is submitted to a MySQL database. The data submitted is initially invisible to the public and subject to critical review by an expert reviewer. Once approved, data becomes public and available for users to query. The database administrator manages users’ accounts and curation tasks. The VIOLIN database is routinely maintained by the database administrator. Published database content is periodically reviewed to ensure that new, pertinent information is captured. When new information is found, a curator and/or a domain expert will update the database content using the standard procedure described above. In addition, the VIOLIN team also periodically emails the authors of new vaccine research publications and encourages them to submit their data through the VIOLIN online submission system. VIOLIN also includes internally developed scripts to automatically update gene annotations based on updated records from existing databases (e.g., NCBI Gene database).

VO representation of Brucella vaccines: Manually curated Brucella vaccines are entered into VO by following the VO development standards [27]. The VO is edited by Protégé (http://protege.stanford.edu/). The FACT++ OWL reasoner [10] is used to obtain inferred Brucellavaccine hierarchy.

Vaxign prediction of Brucella vaccine targets: All ten Brucella genomes stored in the NCBI RefSeq database were used for prediction of Brucella vaccine targets. The genome of B. abortus strains 2308 was used as a seed genome. The other genomes include five sequenced virulent strains from three main pathogenic Brucella species: B. abortus strain 9-941), B. melitensis strains 16M and ATCC 23457, and B. suis strains 1330 and ATCC 23445. These strain are pathogenic to human. The genome of Brucella vaccine strain S19 was also included in this study for comparison purposes. The other three Brucella genomes are from B. ovis strain ATCC 25840, B. canis ATCC 23365, and B. microti strain CCM 4915. More Brucella genomes have been sequenced and available at http://www.broadinstitute.org/annotation/genome/brucella_group. Since the annotations are not yet finished and their records are not stored in the NCBI RefSeq database, these genomes were not typically used in this study. The Vaxign pipeline was executed by using the Brucellagenomes as input data. The processed results were stored in the Vaxign database. The Vaxign web query interface was used to query and analyzed the predicted results.

Query of Brucella vaccine information in VIOLIN:All manually curated or computational processed data can be queried through various VIOLIN web pages. Selected query functions are described in detail in the body of this manuscript.