Visualization in Biology and Medicine - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 235

usually in parallel, and connecting lines are drawn between paired features.
Visualization tools that use a connection encoding include Ensemble [ 6 ], Syn-
Browse [ 33 ], Mauve [ 8 ], and Apollo [ 24 ]. A connection encoding shows both the size
and location of paired features on the source and destination genomes. The encoding,
however, suffers from visual clutter when more than a few dozen paired features are
shown—many comparative genomics datasets contain thousands of pairs.
A limitation of many of the visualization tools for comparative genomics data is
that they do not encode similarity scores in conjunction with the size and location
of paired features. A common method for looking at these scores is using a scatter-
plot [ 21 , 35 ]. In this method, each axis is the coordinate system for a chromosome,
and paired features are indicated with a dot at the appropriate coordinates. These
dots are colormapped based on their associated similarity scores.
22.2.4 Case Study: MizBee
In this case study, we worked with two biologists to design a visualization tool
to look at comparative genomics data across multiple scales and in the context of
similarity scores. We conducted a series of interviews with the biologists fromwhich
we compiled a list of fourteen low-level, data-centric questions they were asking in
their research. For each of these questions, we characterized the scale at which it
operated. From an extensive review of existing tools for visualizing comparative
genomics data, we created a taxonomy of the design space for showing conserved
features as well as amenable ways to layout the chromosomes from the source and
destination genomes.
Our question characterization and design space taxonomy guided the design
of MizBee [ 28 ], the first interactive tool to visualize comparative genomics data
across multiple different scales in the context of similarity scores. MizBee, shown in
Fig. 22.1 c uses multiple linked views [ 36 ] to support guided navigation of the data by
incorporating overviews with detailed views [ 41 ]. The tool includes three levels of
views: the genome view (left), chromosome view (middle), and region view (right).
The genome view uses a circular layout of the source and destination genomes, with
the destination chromosomes on the inner ring along with a single selected source
chromosome from the outer ring. Both color and connection are used as a redun-
dant encoding of paired features. The connecting edges are bundled [ 13 ] to reduce
visual clutter and to highlight spurious pairs. In the chromosome view , the selected
source chromosome is shown in more detail to see the size of conserved regions
along with a bar chart that indicates each region's similarity score. The lowest level
region view shows individual paired features within a conserved region. This view
gives information about the size, location, and orientation of the features, along with
a bar chart of similarity scores. All three views are linked together through selection
and highlighting.
To validate MizBee, we conducted a series of case studies with our biology col-
laborators. These case studies highlight how MizBee improved communication of