|Home | About | Journals | Submit | Contact Us | Français|
Biomedical research relies increasingly on large collections of data sets and knowledge whose generation, representation and analysis often require large collaborative and interdisciplinary efforts. This dimension of ‘big data’ research calls for the development of computational tools to manage such a vast amount of data, as well as tools that can improve communication and access to information from collaborating researchers and from the wider community. Whenever research projects have a defined temporal scope, an additional issue of data management arises, namely how the knowledge generated within the project can be made available beyond its boundaries and life-time. DC-THERA is a European ‘Network of Excellence’ (NoE) that spawned a very large collaborative and interdisciplinary research community, focusing on the development of novel immunotherapies derived from fundamental research in dendritic cell immunobiology. In this article we introduce the DC-THERA Directory, which is an information system designed to support knowledge management for this research community and beyond. We present how the use of metadata and Semantic Web technologies can effectively help to organize the knowledge generated by modern collaborative research, how these technologies can enable effective data management solutions during and beyond the project lifecycle, and how resources such as the DC-THERA Directory fit into the larger context of e-science.
Biomedical research is increasingly reliant on large collections of data and knowledge that require computational approaches for their management and analysis . Deriving knowledge from large amounts of data requires it to be properly organized so that relationships among data elements are understood and put into the context of current knowledge . This is a particularly challenging task in the biomedical domain where information is complex and often relates data with multiple levels of granularities and that pertain to different disciplines . In recent years we have witnessed the development of tools and techniques, the focus of which has evolved from the basic storage and retrieval of data to more versatile tools that enable the integration of heterogeneous data and their annotation through standard terminologies. More recently, we have also seen the emergence of tools that support the social aspect of collaborative and interdisciplinary research.
As biomedical research started to become an information-intensive discipline, the focus of bioinformatics research initially was the creation of data-specific databases to store and enable searches over a growing quantity of data such as sequences  and gene expression [5, 6], or more complex information such as pathways  and scientific knowledge represented in the literature .
However it soon became clear that a proper metadata framework to annotate data was essential for making sense of the information stored in these databases [9, 10]. For instance, the functional genomics community pioneered the development of shared and computable terminologies (ontologies) to define experimental conditions , which resulted in the construction of the Ontology for Biomedical Investigation (OBI) .
Biomedical research often requires the integration and analysis of different types of information in a biological system, which is a complex task, as this information is often stored in different databases and represented differently. As a consequence, much research has been carried out on how best to manage, interrelate and interrogate biomedical data [12–14]. The task of ‘Data integration’ poses both technical and semantic challenges, which are often interconnected. The technology for relating information artifacts has evolved from the linking of flat files, through specialized software solutions , to web-based information systems that capitalize on the use of ontologies to provide distributed knowledge bases . Underpinning these technologies are tools that allow the composition of data and services, which in turn have evolved from middleware such as CORBA  to web services, orchestrated web services  and advanced user interfaces and interactive environments . We are now witnessing the convergence of solutions that merge ontology-enabled web services with the declarative nature of the web .
Beyond the techniques that have evolved to relate and exchange information across distinct databases, there is a need for the definition of common ‘languages’ to describe integrated information. When data integration was carried out within homogenous research communities, those languages could rely on a shared understanding of their concepts and the semantic challenges of data integration were addressed through the definition of ‘exchange languages’ [21, 22].
As research became increasingly interdisciplinary, however, the necessity for a common understanding of terms across different disciplines prompted the development of ontologies, such as the Gene Ontology (GO) , the success of which has led to the development of coherent ontology libraries, such as the Open Biomedical Ontologies (OBO) collection .
The definition of these biomedical ontologies has evolved both in its ontological foundations, with the commitment to common upper ontologies such as the Basic Formal Ontology (BFO) , and in its representation, which has become increasingly logic based, via the adoption of ontology definition languages such as the Web Ontology Language (OWL) [26–29]. Ontologies now comprise the backbone of biomedical informatics, with dedicated institutions such as the National Center for Biomedical Ontology (NCBO)  and resources such as the BioPortal . Use of ontologies is not limited to the annotation of databases : ontologies have provided a significant contribution to high-throughput data analysis and increasingly are seen as a device to make scientific literature more machine processable. Hence, they enable researchers to make better use of the increasing amount of knowledge available in this format. The gap between databases and scientific literature is narrowing [33–35].
From a wider perspective, the definition of ontologies and the increasing relevance of web-based technologies are part of a larger evolution of science (and knowledge creation in general), characterized by a computationally enabled social dimension . This evolution has far-reaching consequences that touch the role of the public in scientific research, for instance, through ‘crowd sourcing’  and through ownership of information .
So far, development of web-based resources that represent information through shared computable languages has focused on ‘primary products’ of research, such as datasets and literature. There are reasons for developing similar resources that focus on the research process itself. Research is often organized into projects that involve a network of collaborating participants who need to communicate and share intermediate results, best practices and, in general, their know-how. This necessity for communication and knowledge sharing is not dissimilar to the needs of the biomedical community at large, and sometimes such networks have adopted knowledge management solution that mimic the functionalities of public repositories . More often, they have relied on tools commonly used for project information and communication, such as wikis, mailing lists or content management systems (CMS).
It is useful to devise (web-based) resources that bring together these types of tools and that can both support the project-related activities of research communities and at the same time integrate their information with that of distributed repositories. Too often, metadata and data curation are left as a final step of research, causing precious information, useful for qualifying the output of research, to be lost . Furthermore, there is a clear potential for error detection and the reduction of duplicate efforts.
Such resources can facilitate the sharing of much more relevant and useful information than can traditional methods such as publications: small facts and negative results can be published via web-based systems more widely and efficiently than via scientific literature, and they can be managed by means of computational systems that can provide credit for their generation .
Finally, such resources can integrate the social side of scientific research with the information that it generates, thus improving communication and collaboration among researchers. This is particularly true for the ‘long tail’ of researchers who share some specific interest, but who otherwise might be remote from the core community.
DC-THERA  is a European Network of Excellence (NoE) established under the European Commission’s Sixth Framework Program, which has integrated many researchers and clinicians, working collectively on basic scientific and therapeutic aspects of dendritic cells (DC), a topic central to immunology. The network has brought together at least 32 partners and 38 associated partners, from 18 different European countries. It is a typical example of a translational and distributed research project, which has prompted the need for a computational, community-based approach to manage a wide range of heterogeneous information. The organization of information in DC-THERA poses additional challenges, since research focusing primarily on DC requires a characterization of resources by their cell-type specificity that often transcends the characterization provided by generic tools and information resources. The nature of DC-THERA as a research project also highlights issues about the way the information generated can be maintained after the project has ended, and how such information can be absorbed or re-used by other efforts that can emerge from the DC-THERA and from the wider community.
In this article we describe the DC-THERA Directory ([43, 44], hereafter called also ‘the Directory’). This is a web-based knowledge management system, initially designed to address the collaborative and sharing needs of the DC-THERA community. The Directory focuses on the ‘network knowledge’, which is the set of technical resources, research expertise, personnel and their relationships that make up the core of a NoE and similarly organized communities.
The design of the DC-THERA Directory addressed three main goals.
First, to provide an information gateway for the DC research community that enriches proprietary and public information through annotations and search functions and provides focused information set for consumption by its researchers and other computational systems.
Second, to represent information in the Directory through languages and terminologies that are ‘compatible’ with the biomedical information ecosystem.
And, last, to maximize the ‘integrability’ of the represented information with external resources, so as to maximize its usefulness and visibility, beyond the boundaries of the specific research network that was initially served.
The DC-THERA Directory is a public web site  that provides information on research assets available within the DC-THERA community and, at the same time, integrates external resources to provide a coherent access point for researchers. Like other e-science resources, the Directory relies on annotations through ontologies and standard languages to provide advanced search and organization functions. It also offers functions for editing information and for managing its privacy. At the time of writing, the Directory provides summary information on 237 data sets, 79 protocols, 524 biological materials, 122 laboratory tools (which include both equipment and consumables), 79 organizations and 328 persons. In addition, it integrates internal and external microarray repositories and provides literature and pathway analysis services.
The DC-THERA Directory is organized such that one research asset corresponds to one information item in the Directory. Each information item is annotated via a brief textual description, a type, a set of attributes and its relationships to other resources. Complex resources are represented in more detail, as is the case for protocols, where protocol steps, their order and their requirements and results are represented explicitly. It is worth mentioning that a relevant feature of the Directory is its annotation of resources, protocols, data sets and eventually tools in terms of the specific cell type to which they relate, namely DC. This is of particular importance in immunology, as the interplay of different cell types is key aspect of the immune system. As shown in the next section, the Directory addresses the specificity of DC biology by annotating cell-type specific reactions and reagents and by using more general types to cross-connect the results stemming from the interaction of different cell types.
Most of the types, attributes and relations used in the Directory are drawn from ontologies of the OBO family, which makes the knowledge representation of the Directory contents interoperable with other related biomedical knowledge. To obtain a seamless and simplified framework for using these existing ontologies to annotate the Directory resources, a DC-THERA application ontology has been defined, using the standard ontology language, OWL. This ontology mostly extends the OBI, and integrates several other OBO ontologies (Table 1) in a way that suits the Directory annotation purposes. Moreover, several relationships and classes have been defined in order to achieve a balance between ontological precision and usability. For instance, a relationship is provided to link a cell culture to an ontology concept that represents the type of cells comprising that culture. While this is presented as a ‘cell culture X of type Y’, the relationship is actually a short-cut for the more correct statement: ‘the cell culture X is a population of cells such that each cell has the property of being an instance of the type Y’. Not only does this short-cut simplify the editing tasks for the curator end-user, it also leaves room to derive the correct inference in an ontological framework (e.g. by means of rules). Another more trivial example is the use of an ad-hoc relation ‘is-knowledgeable-about’, defined to cover a range of relationships that could not be specified further (e.g. ‘has produced the bio-material in the laboratory’ or ‘is an expert in the protocol’). A brief overview of the top-level classes and relationships defined in the DC-THERA ontology is provided in Figure 1.
The Directory fulfills its role of information gateway for researchers by providing interactive search, result inspection and editing functions. In the Directory these functions often capitalize on the annotation of its information via ontologies and shared relations.
The Directory provides different ways of accessing information. Ontologies are used as taxonomical indexes to organize and access its content, a significant case being access by cell type (or bio-material). In addition, the Directory also provides a simple ‘Google-like’ query interface that assists the user dynamically by providing predictive suggestions while typing. Keywords entered in the search forms are expanded in their synonyms and morphological variants and are then used to match types and text in the Directory, as well as to query external services. Results of a query are presented as a list with a brief description, where a color code identifies text matching exact terms or synonyms.
In some cases, the Directory tracks the user behavior to restrict free text queries so as to provide more ‘intuitive’ results. As an example, if the user is performing a text query while observing the list of data sets retrieved through an access via data type (taxonomy), the query is limited to data sets in the Directory, external resources that contain information on data sets (e.g. ArrayExpress) and other entries in the Directory that are related to the results found.
The Directory relies on ontologies to expand the scope of queries from more generic to more specific terms. For instance, a query for ‘Leukocyte’ will query the Directory also for all known sub-types of ‘Leukocyte’, as defined in the DC-THERA Ontology and as presented in the Directory in the bio-material taxonomy. Queries can then be used to extract the content of the Directory (e.g. reagents, data sets, protocols), and hence navigate its content, with the desired level of generalization in the specification of cell types.
For each resource, the Directory presents a ‘resource-centric’ view that provides a description of the resource and its context in the Directory: a brief overview of which other entries in the Directory relate to the resource in question, and how (Figure 2).
The description of a resource is generally in the form of a short piece of text and a list of features, organized in property/value pairs. Depending on the resource type, additional detail can be presented. This is the case for protocols, where the detailed description of their workflow is provided via a graph.
From the ‘resource-centric’ view, a user can easily identify other relevant resources in the same context and navigate the content of the Directory to which they relate. For example, a user can navigate from the description of a data set to the analysis protocol that was used to generate this data-set (where ‘generate’ is the property linking the two), then to a specific tool ‘used in’ the protocol, and from there to a member of the DC-THERA Network ‘knowledgeable about’ that tool.
Users can annotate data in a simple way. If they have sufficient permissions, they can annotate entities displayed in the ‘resource-centric’ view with properties/values, or with relationships to other objects. The Directory guides the user by proposing relationships or attributes that are desirable or sensible for the type of entity described. Users can then decide whether to make the created information public or whether to restrict it to a set of participants. The Directory supports a data access model where users can belong to different groups with different roles. For each group, and depending on their role, users may be able to read, edit, delete or even supervise curation of the entries assigned to it. A similar access model is described in .
In the design of the Directory we have addressed data management issues that relate to the lifecycle of a research project.
One goal has been to guarantee the longevity of the information in the Directory beyond the duration and the scope of the project. This has posed two problems: an economic problem, since resources to maintain the system cannot be guaranteed beyond the duration of the project, and a ‘usability’ problem, to ensure that the information in the Directory can be found easily, transported and manipulated with other systems to maximize ‘re-use’ of the information generated within the network.
To address these issues, we have leveraged on the formal annotation of the entries in the Directory by making them available through standard technologies of the Semantic Web framework , such as RDFa  or SPARQL , details on which will be briefly presented later.
We present a few examples here that show how adoption of these technologies can improve the data management lifecycle.
Most of the information in the Directory has been made publicly available on the web, after an initial phase in which access was restricted to DC-THERA participants. The opening of the Directory to the public was motivated by its potential value to the wider scientific community, including the potential for spawning new collaborations and ideas, due to the links that the Directory contains between the scientific information and the people involved in its production and usage. Because of that, the reach of this public content is highly desirable.
In particular, reachability via web search engines can have a high impact over the lifespan of the project knowledge and its spread. Because of its curated content and its interrelation with internal and external resources, the Directory has a potential for enhanced visibility in search engines. We have built on this potential by enriching some of the web content through RDFa, a mark-up language that makes the types, relationships and attributes in the Directory understandable by other software, and in particular by search engines such as Yahoo or Google [50, 51].
We have monitored traffic data since the Directory went public in January 2010 (Figure 3). Traffic initially decreased towards the end of the project but thereafter started to increase by a steady 10% on a month-on-month basis, with the vast majority of traffic originating from web searches. This pattern suggests a shift in the usage of DC-RESEARCH.EU from a project-specific resource to the wider external public, which shows that the information generated within the network is still ‘alive’ after the end of the project.
An inspection of the most used search keywords reflects the content of the Directory: people, resources and the combination of the two without revealing any particular ‘artifact’ (see also Figure 5 in the Discussion section).
The use of Semantic Web technologies also allows the Directory to integrate external tools and functionalities at minimal cost. We show this point by means of RelFinder , a tool that was originally developed in the context of the DBPedia project  for analysis and visualization of entities represented in a Semantic Web-enabled knowledge base.
RelFinder asks the user for two or three entities and, after a disambiguation step, searches for relevant connections in the knowledge base that connect such terms and displays the result as a graph. This functionality fits well within the Directory, allowing, for instance, the discovery of connections amongst participants, or between a given researcher and a specific data set. An example of a result found via RelFinder is presented in Figure 4.
Integration of RelFinder in the Directory is a significant functional enrichment, which required only the configuration of the SPARQL end-point of the Directory: a single web address.
Conversely, adoption of Semantic Web technologies allows the Directory content to be readily integrated with external resources. As an extreme case of this integrability, we have migrated the entire contents of the Directory to the Talis platform : a public infrastructure offering reliable and efficient storage and access of both unstructured and structured data. (Talis, which is behind the publication of UK government data online , offers free access to its platform to qualifying academic projects under the terms of the Talis Connected Commons program ).
Tools to access and query the content of the Directory (e.g. RelFinder) can be directed seamlessly to the Directory SPARQL end-point, or to the end-point that is provided by the Talis platform .
In this way we have achieved two important results. First, we have guaranteed maintenance of the Directory content even beyond the availability of funds to operate the current web infrastructure.
Second, we have accessed a range of functionalities offered by the Talis platform, including access to new dissemination channels such as those being explored on the data market platform, Kasabi , which currently offers the content of the Directory to interested early adopters.
The DC-THERA Directory makes extensive use of the Semantic Web framework, which is a set of standards and technologies designed to make the web a distributed, query-able, knowledge base. Annotation in the Directory closely matches the data model defined by the Resource Description Framework (RDF) , a key component of this framework. The Directory provides information on the web through different Semantic Web technologies, such as RDFa and SPARQL, mentioned above. More precisely, each resource in the Directory is associated with a URI, which is also an URL (e.g. ‘http://dc-research.eu/rdf/protocol/10’) pointing to an RDF representation of resource information (serialized in XML/RDF). A related URI/URL (e.g. ‘http://dc-research.eu/protocol/10’) resolves to an HTML representation of the information, which is enriched via an RDFa mark-up. Interaction with the information content in the Directory is based on the REST paradigm .
Public information presented by the Directory can also be queried by means of SPARQL, a query language for RDF-based knowledge bases. A SPARQL end-point (i.e. a server that can answer SPARQL queries) is available at the address http://dc-research.eu/sparql.
Development of the Directory within the lifecycle of the project has required rapid prototyping and agile methodologies, as discussed in . The Directory has been developed in cycles of releases (Table 2). At the end of each cycle, feedback on functionalities and prioritization of the next functionalities to be implemented has been collected from its end-users: the network participants.
The Directory is based on an ad-hoc software engine that combines object oriented modeling of the main types in the Directory with a schema-less RDF-like modeling of information, following a similar approach to that presented in . This engine was developed to enable the usage of established web development techniques and frameworks and thus to maximize the maintainability of the code base and the effectiveness of the deployment cycle. In particular, the design of the Directory follows a Model-View-Controller (MVC) approach , implemented through the Symfony  framework  and the Relational mapper (ORM) engine Doctrine . Implementation of the Semantic Web functionalities has been based largely on the ARC Library .
The Directory integrates a range of features from other computational resources, both public ones and resources where access is restricted to DC-THERA participants.
Other resources are imported into the Directory through specialized scripts, such as BioLexicon , which is used to expand terms in their synonym and morphological variants, and the body of ontologies, expressed in OWL, which constitute the DC-THERA Ontology.
The DC-THERA Directory addresses data management issues typical of a large collaborative biomedical research effort, and in particular the need for the information produced to be part of a larger shared information space. From a wider perspective, the Directory is part of a range of modern developments that affect the way science is conducted and communicated.
The commitment of the Directory to the web and Semantic Web standards reflects the increasing role of the web as a knowledge mediation platform  and the emergence of standard publication practices such as Linked Data . The attention that the Directory pays to both the social aspect of annotation and its machine readability reflects the trend towards the development of communities of interest  and towards the formalization of the research process, which is explored in . The cell-type specific annotation of resources within the Directory provides a significant improvement in the way researcher can access, share and relate information.
Several other resources are pioneering the development of computational collaborative research tools to support research in the Life Sciences. We provide a brief review here of some representative examples, rather than an exhaustive list.
Some social e-science sites have been developed with a specific need or data type as a unifying item on which a community was later built. This is the case of myExperiment , a social site designed for the exchange of bioinformatic workflows. It supports annotation via RDF and ontologies and publishes information via SPARQL in a similar way to the Directory. Similar features are offered by BioCatalogue  for the annotation of bioinformatic web services. Within the Systems Biology project SysMO, SymoDB  has been developed to support sharing of models and simulations among participants.
Other sites are intended as gateways for specific research communities. Similar to the Directory, they aggregate and organize heterogeneous information, but they vary in the specifics of implemented solutions, and include the following.
The Neuroscience Information Framework  is a comprehensive web resource for the Neuroscience field that makes several types of ontology-annotated resources available, providing data federation for many different biological databases and advanced search features.
The Alzheimer Research Forum  collects information about Alzheimer disease in a similar way. It allows users to link resources to scientific hypotheses and to discussions about them.
The Trial Item Manager  is an application similar to the Directory, allowing for collaborative editing of clinical trial information by means of detailed case report forms.
Particularly similar to the Directory and to its design as an ‘actionable’ inventory of research assets is Eagle-I , a recently formed consortium of several US organizations, aiming to, in their words: ‘build a prototype of a national research resource discovery network—one that will help biomedical scientists search for and find previously invisible, but highly valuable, resources’. Mentioned examples of these resources are: animal models, reagents, cell and tissue banks, core facilities and training opportunities.
A different class of information systems to support the collaborative development of information resources are wikis, which are the backbone of many research project information systems as well as of large-scale annotation efforts [82–84]. However, while wikis are an effective tool in many cases, they have limitations in scaling up with structured non-regular data . Tools such as the Directory are designed to address information with these characteristics.
We have attempted a comparison of the usage of these resources with that of the DC-THERA Directory (Figure 5). While the numbers reported can only be considered as indicative, they show that the Directory is substantially less frequented than information resources that have an established web presence (as hinted by the number of incoming links) and which appeal to a relatively generic public. This is not surprising, as the Directory is both a new and specialized resource. However, the specificity of queries is among the highest of the resources presented (all top queries that lead to the website are relevant for its content). Together with results reported in Figure 3, this indicates a healthy status for the Directory, which has evolved from a project-specific information resource into a web resource that is attracting (and retaining) new users clearly focused on its content.
There remain limitations in the adoption of collaborative web environments in the research practice that vary depending on the characteristics of the project and its social environment.
In the case of the Directory, there was no problem of creating a community, since a research network was already in place at the time of its design. In our experience, the bottleneck in the uptake of this environment was the engagement of users to provide information for the Directory, which was solved partly via automated information importers and curation. By means of user experience surveys and feedback collected at demonstrations, we found that the Directory had a good reception among participants. Beside qualitative observations, we organized polls from a representative panel of selected participants, and the Directory was rated high (>7/10) on aspects such as the overview it provides, its intuitiveness and its search functionalities. Users' feedback was also important to reveal limitations of the Directory, and it lead to the introduction of new features and improvements, as highlighted in Table 2. The use of the Directory through tools such as RelFinder has been of particular interest as it provides an innovative and intuitive way to mine connections among participants and knowledge. However its usage still requires the mediation of an expert, as the low level representation of information presented by RDF and its mix of domain and ‘meta’ statements can be confusing to a biomedical research public. Even more promising, though, is the increasing number of visits that the Directory is collecting from Web Searches (Figure 3).
Overall, the main issue in the development of social resources for science is rewarding content provision. ‘Web visibility’ could be a reward that we are exploring in the Directory. In the past, funding agencies and scientific journals had a key role in consolidating the role of databases and standards in the research community . Similar incentives could benefit the development of coherent data management strategies in the research practice. Another interesting incentive could be linking knowledge management systems to project administration, for instance, by automating project reporting.
A distinct problem relates to the complexity of curation, for which there is not an easy solution. In general, there is a trade-off between coverage and precision of annotation. In the Directory we have resorted to curated ontological information, which would have been difficult to crowd-source, at this stage of the evolution of technology.
The DC-THERA Directory has explored the use of a Semantic Web-based data management platform for the curation of the research assets, or the ‘network know-how’ of a research network.
Within the DC-THERA community, the Directory has proved important in stimulating data integration and collaborative research by sharing information.
Controlled vocabularies facilitated data integration in immunology as well as comparison of large ‘omics’ data sets by annotation of cell type specific processes and variables. Emphasis on ontological annotations and standards makes it a resource valuable beyond the limits of the project and, in particular, we have shown how the Directory can support important aspects of the data management lifecycle, providing a resource-efficient way to integrate the information content with external resources, such as tools and knowledge bases.
This work was supported by the DC-THERA Network of Excellence (European Commission NoE contract number: LSHB-CT-2004-512074).
The authors would like to thank all the DC-THERA participants, in particular those who participated in the definition of the application requirements and who took part in the evaluation panel, Maria Cristina Gauzzi, Damariz Rivero, Éva Rajnavölgyi and Rita Nunes. They wish to thank Olivier Lefevre for the work on the DC-BASE integration and Luca Beltrame for his work in the integration of Pathway Analysis functionalities.
Andrea Splendiani is a Founder of IntelliLeaf ltd and a Senior Bioinformatic Scientist at Rothamsted Research. His research interests are in Semantic Web technologies, biomedical ontologies and Systems Biology.
Michaela Gündel holds an MSc degree in Life Science Informatics. Her main research interest lies in semantic knowledge management, ontology development and their application to the life sciences.
Jonathan M. Austyn is Professor of Immunobiology at the University of Oxford and Project Coordinator for the DC-THERA Network. His research interests are the immunobiology and therapeutic applications of dendritic cells.
Duccio Cavalieri is Professor at the Faculty of Pharmacy, University of Florence, applies computational biology to the investigation of the signaling networks of the dendritic cells through pathway analysis of functional genomics data sets.
Ciro Scognamiglio is a Freelance Web Professional based in Paris. His main expertise is in Software Development and Unix System Administration.
Marco Brandizi is a Founder of IntelliLeaf Ltd and a software engineer at the European Bioinformatics Institute, where he works on the management of microarray data and other multi-omics data.