University of Oxford Big Data Institute: Development & dissemination of efficient analysis methods for large, complex, heterogeneous clinical datasets

Background: Understanding the determinants of common life-threatening and disabling disease is challenging. Such conditions are typically caused by many different factors (genetic, physiological, behavioural, infectious and environmental), which may be occur at different times of life (including developmental factors as well as those that may occur much later; e.g. cigarette smoking, alcohol, co-morbidity, and treatment). Furthermore the effects may manifest over a range of timescales (immediate to many decades) and time courses (acute, chronic stable, progressive, relapsing/remitting).

Vision: The University of Oxford Big Data Institute (BDI) sets out to be an international centre of excellence for the analysis of very large and complex biomedical data sets. The Institute will combine world-class academic leadership with a core infrastructure of cutting-edge technology and high calibre scientists. Built upon Oxford's internationally leading expertise in epidemiology, genomics, imaging, computer science, and infectious disease surveillance, the BDI will be a national resource for the development of new analytical methods to facilitate the generation, storage, analysis and sharing of data in very large clinical studies. This work will provide transformative advances in the speed and range of research into the causes and consequences, prevention and treatment of disease, and will be particularly relevant to understanding the role of genetic and environmental influences on common life-threatening and disabling diseases such as cancer, cardiac disease, stroke, and dementia.

Knowledge Transfer: In addition to primary research, the BDI will play a major role in capacity building through training, dissemination of research methods, and stakeholder engagement:
- A new MRC Big Data Training Academy will deliver an extensive, flexible and outward-looking portfolio of training and career development opportunities accessible to researchers of all levels within and outside Oxford, including a new doctoral training programme, support for post-doctoral training fellows, visitor and exchange programmes, and a broad range of short courses suitable for graduate training and continuing professional development.
- Effective collaboration with local, national and international experts (including partnerships with academia, healthcare, and pharmaceutical and IT industries) will enhance the development, evaluation and adoption of new research methods and tools.
- The Institute will work with regulators and other key stakeholders to develop standards for data storage, sharing and analysis; and to promote appropriate and proportionate regulatory and governance approaches that are fit-for-purpose in areas such as privacy, consent, information security, data access and sharing, intellectual property, and the intersection between research and routine practice.
- Public engagement activities will promote understanding, address concerns, and develop trust in "Big Data" approaches to biomedical research.

Advances in Big Data offer new ways to conduct health research, with advantages in speed, cost and scope of scientific enquiry. In particular, major advances will come from the development of new methods to integrate and interrogate multiple, complementary datasets simultaneously. The University of Oxford Big Data Institute (BDI) will develop flexible and efficient systems, tools and methods for generating and analysing research-optimized data sets that incorporate relevant clinical information (including bespoke research data and routine clinical information), extensive phenotypic measurements (including imaging, physical and function assessments) and genomic and other laboratory data.

Evaluation of clinical phenotype: The BDI will bring together bioinformaticians, clinical specialists and research users to develop validated algorithms that combine multiple data sources to confirm, clarify and classify clinically and biologically relevant phenotypes in very large studies.

Interpretation of genotype: Current statistical methodologies for assessment of genetic influences on disease (many of which were developed in Oxford) fail in regions of high structural or sequence diversity (e.g. around human HLA and KIR loci, and in many bacterial and eukaryotic pathogens). The BDI will develop novel methods that integrate information from multiple reference sequences, thus extending the ability for researchers to investigate genetic influences on disease pathways.

Combining multi-dimensional phenotypic and genomic data: Building on our experience in clinical informatics and genetic analyses, we will address gaps in the current standards and ontologies. We will develop appropriate statistical methods for the analysis of high-dimensional and highly correlated data in order to allow biologically meaningful associations to be identified robustly. We will also establish an infrastructure that supports appropriate sharing and analysis of biomedical Big Data.

The University of Oxford Big Data Institute (BDI) seeks to promote Innovation in Medical Science to produce Transformation in Human Health.

The Institute aims to have broad-ranging influence:

Improved methods and technologies for research: The BDI will bring together existing expertise in large-scale epidemiology, imaging, genomic medicine, bioinformatics, and computer science to create an internationally leading centre of excellence for the analysis of large, complex, heterogeneous data sets for research into the causes and consequences, prevention and treatment of disease. The work of the Institute has the potential to radically extend the boundaries of these individual disciplines (e.g. large-scale epidemiology will require new tools and methodologies to analyse even larger, more complex and rich heterogeneous data), but, the greatest impact is likely to be achieved by close collaboration and working between disparate disciplines (e.g. computer science and epidemiology to develop new analytical tools for routine health data; clinical, imaging and genetic science to better understand the development of dementia). The creation of the Institute will act as a focal point for fostering collaborative engagement with both academic and industrial partners, nationally and internationally.

New advances in science and health: The expertise, methods and systems developed will provide big advances in the ability to understand disease mechanisms and discover new treatment approaches, and will yield significant benefits for clinical care in areas such as infectious disease surveillance & management, and the translation of genomics into routine medical practice. We intend to make our data, methods & tools widely available for scientific researchers.

Capacity building: A new MRC Big Data Training Academy will provide an extensive portfolio of training and development opportunities, suitable for scientists at all stages of their career, and readily accessible to external researchers. These will include a new doctoral training programme in Big Data for Biomedicine, post-doctoral fellowships, a visitor and exchange programme, a large number of short courses suitable for both graduate training and continuing professional development. Much of the training material will be available through a remote learning environment, and the annual Oxford-Stanford Conference on Big Data will provide an opportunity for cross-fertilisation of ideas among the many stakeholders in this research. Thus the multidisciplinary environment of the new Institute will provide substantial opportunities for the development of a new generation of research scientists, equipped to tackle the major medical bioinformatic challenges of Big Data research for the improvement of human health.

Commercial and economic impact: The work seeks to generate new methods, tools and systems for analysing Big Data in biomedicine. These tools will be made widely available through commercial entities spun out of the University, engagement with large IT companies such as Oracle, Microsoft and Google; and strong relationships with pharma who, like academia, are dependent on this type of large sale analysis. We anticipate that the Institute's work will generate substantial investment into UK bioinformatics and health research, and opportunities for spinouts.

Societal impact: The Institute will create a community for engagement, dialogue and debate with government, regulators, policy makers and the public. Developing public trust, understanding and confidence through ongoing active engagement will be a key activity of the new Institute.