A lightweight genome browser for data integration, exploration and interactive figures

Genome browsers are important tools for visualising, interpreting, and
understanding the wealth of information now available about genome
structure, its functions, and the variations that occur between
individuals. In research environments, they are frequently used for
visually checking the quality of new data, looking for patterns and
correlations, and inspecting the results of analysis tools. Browsers
are also important for education, and -- especially with the arrival of
direct-to-consumer genetic testing -- may also be of interest to the
general public.

Historically, the most commonly-used genome browsers have been relatively
static web applications, displaying a portion of the genome as a single
image which must be reloaded to move or zoom the view, leaving major
barriers to interactivity. The alternative is heavy-client desktop
software which is more interactive but requires installation. Dalliance
takes advantage of newer features of web browsers to build a fully
interactive genome browsing tool as a web application.

Because Dalliance does all the data integration and drawing work in code
that runs within the end users' web browsers, it is relatively
straightforward to add an instance to a new web page without any
server-side work. The Dalliance instance can potentially be just one of
many elements on a complex web page. This leads to the possibility of
creating many kinds of applications which combine spatial views of
genomic data with other information. We are already seeing a number of
cases where academic and industrial researchers have used Dalliance as a
browser component of their website, including the interpretation of DTC
genetic tests (see letters of support). You can even think of Dalliance
as allowing the creation of interactive figures, which can accompany
database records, analysis results, or publications.

We propose a program of further development work, rewriting the main
rendering code to improve performance, and splitting the user interface
into components. This means embedders have the flexibility to offer
either a simplified interface (e.g. more appropriate to an interactive
figure), a fully-featured interface with a complete set of navigation
controls tools to control the integration of additional datasets, or
anything in between.

We will add support for additional genomic data formats -- notably,
better support for genetic variation data. We will also update the
metadata model, and support additional metadata sources such as the
"track hub" structures used by the UCSC browser. This will enable easy
integration of existing data from the ENCODE project, and offer a simple
route for sharing biological datasets with enough metadata to allow
display in a genome browser with descriptions and track selection user
interfaces.

Finally, we plan to organise a federated data workshop to interact with
developers and encourage the development of federated solutions for sharing
biological data. This will be modelled on the popular DAS workshop that
has been organised at Hinxton in the past, but with a broader aim of
supporting and encouraging data integration in a variety of formats,
notably the modern style of indexed binary data and trackhub-style
metadata which allows discoverable data to be published with extremely
low requirements for server hardware or administration.

In the course of this project, we plan to develop the existing
Dalliance genome-browser code base and documentation. Our objectives
are to:

1. Maximise the value of genomic data by encouraging visualisation,
exploration, and understanding. To this aim, we will provide tools
which allow these data to be seen in context.

2. Facilitate genome exploration by making it easy to embed genome
browser components in web pages, scientific papers, database front
ends, and web interfaces for analysis tools.

3. Make it easier to customise the browser to fit the styling of
its enclosing application, and to interact seamlessly with the
enclosing application's user interface, increasing the breadth
of ways in which users can interact with these data, and promoting
experimentation in the user interface space.

4. Make it easier to discover and integrate new biological data
sets using established technologies for metadata publication and
sharing, encouraging the combination of data in new ways.

5. Support all the important formats that are being used to share
such data. We have a particular focus on compact and efficient
formats which make access to large datasets practical, and make
access to genomic data more feasible to those with limited network
bandwidth.

6. Maximise the performance of the browser, especially when accessing
large datasets, to facilitate interactivity and data exploration.

Genome browsers are vital tools for both academic and commercial
exploitation of genomic data. The ability to visualise experimental
biological datasets against existing genome annotation is a fundamental
requirement for interpretation and analysis. The proposed development
work will generate software that will lower the barriers to the creation
of custom visualisation tools and analysis. The open availability of the
software code will ensure its use is maximised.

Since genome sequence based assays have become pervasive in biological
research, this work has the potential to assist a large fraction of
individuals who carry out biological research. This includes those
working in the pharmaceutical industry where such analysis is important
when using model organisms of human in drug development and in the
analysis of genomic differences between human individuals as part of
efforts to interpret gene disease relationships. It will directly
benefit researchers developing tools that need to include a
visualisation of data against genome coordinates by greatly simplifying
the task of including that component. It will also enable a much wider
researchers who are not specialist bioinformatics developers to
construct web pages presenting their data in the form of interactive
figures containing a fully functioning genome browser.

As genomics become more widely adopted in medicine there will be new
visualisation requirements for healthcare systems. Already
direct-to-consumer genetic test companies provide browsers as part of
their service. Both NHS and commercial groups developing analysis and
decision support systems for doctors and patients that incorporate
genomic data will similarly benefit through a simplified development
path.

Visualisation of biological information on a genome coordinate system
has become such a fundamental principle of biology that it will become
an important concept to teach in schools and as well as undergraduate
classes. Simplified representations may also make good interactive
displays for museum presentations about biology or health. Dalliance
will bring the construction of genome browser interactive displays
within reach of these groups by simplifying it to the level of creation
of a stand alone web page.

All these groups will benefit from the availability of Dalliance through
the popular Github platform, making it easy for third-party developers
to follow development, stay up to date, and -- if they wish -- submit
their own changes for integration back into the main branch of
development. Dalliance is released under a BSD-style license which means
that it can be freely embedded into other systems, regardless of the
licensing model of the enclosing application. Because Dalliance is an
open source project, individual components can be used in other
contexts. This will benefit all these groups by encouraging
collaboration and code-sharing.

To maximise the impact of this project, it is vital to engage developers
from all sectors that might benefit from including genomics support in
their applications. As well as engaging users through journal,
conference, and web presence with extensive documentation, we plan to
organise a federated data workshop to interact with developers and
encourage the development of federated solutions for biological data. We
see this workshop as a valuable opportunity to meet existing embedders,
discuss new projects, and strengthen the community that is growing up
around these technologies.