D. Fung, S.Li, A. Goel, S. Hong and M. Wilkins, "Visualization of the Interactome: what are we looking at?", Proteomics, to appear.
Network visualization of the interactome has been become routine in systems biology research. Not only does it serve as an illustration on the cellular organization of protein-protein interactions, it also serves as a biological context for gaining insights from high-throughput data. However, the challenges to produce an effective visualization have been great owing to the fact that the scale, biological context, and dynamics of any given interactome are too large and complex to be captured by a single visualization. Visualization design therefore requires a pragmatic tradeoff between capturing biological concept and being comprehensible. In this review, we focus on the biological interpretation of different network visualizations. We will draw on examples predominantly from our experiences but elaborate them in the context of the broader field. A rich variety of networks will be introduced including interactomes and the complexome in 2D, interactomes in 2.5D and 3D, and dynamic networks.
D. Fung, M. Wilkins, D. Hart and S. Hong, "Using the Clustered Circular Layout as an Informative Method for Visualizing Protein-protein Interaction Networks", Proteomics, Volume 10, Issue 14, No. 14, July 2010, pp. 2723-2727, 2010.
The force-directed layout is commonly used in computer-generated visualizations of protein-protein interaction networks. While it is good for providing a visual outline of the protein complexes and their interactions, it has two limitations when used as a visual analysis method. The first is poor reproducibility. Repeated running of the algorithm does not necessarily generate the same layout, therefore, demanding cognitive readaptation on the investigator's part. The second limitation is that it does not explicitly display complementary biological information, e.g. Gene Ontology, other than the protein names or gene symbols. Here, we present an alternative layout called the clustered circular layout. Using the human DNA replication protein-protein interaction network as a case study, we compared the two network layouts for their merits and limitations in supporting visual analysis.
D. Fung, S. Hong, D. Koschutzki, F. Schreiber and K. Xu, "2.5D Visualisation of Overlapping Biological Networks", Journal of Integrative Bioinformatics, 5(1), 2008.
Biological data is often structured in the form of complex interconnected networks such as protein interaction and metabolic networks. In this paper, we investigate a new problem of visualising such overlapping biological networks. Two networks overlap if they share some nodes and edges. We present an approach for constructing visualisations of two overlapping networks, based on a restricted three dimensional representation. More specifically, we use three parallel two dimensional planes placed in three dimensions to represent overlapping networks: one for each network (the top and the bottom planes) and one for the overlapping part (in the middle plane). Our method aims to achieve both drawing aesthetics (or conventions) for each individual network, and highlighting the intersection part by them. Using three biological datasets, we evaluate our visualisation design with the aim to test whether overlapping networks can support the visual analysis of heterogeneous and yet interconnected networks.
K. Xu, R. Williams, S. Hong, Q. Liu and J. Zhang, "Semi-Bipartite Graph Visualization for Gene Ontology Networks", Proceedings of Graph Drawing 2009, LNCS, Springer, pp. 244-255, Springer, 2010.
In this paper we propose three layout algorithms for semi-
bipartite graphs|bipartite graphs with edges in one partition|that emerge
from microarray experiment analysis. We also introduce a method that
effiectively reduces visual complexity by removing less informative nodes.
The drawing quality and running time are evaluated with five real-world
datasets, and the results show significant reduction in crossing number
and total edge length. All the proposed methods are available in visual
ization package GEOMI, and are well received by domain users.
D. Fung, S. Hong, D. Koschutzki, F. Schreiber and K. Xu, "Visual Analysis of Overlapping Biological Networks", Proceedings of International Conference on Information Visualisation (IV 2009), IEEE Computer Society 2009, pp. 337-342, 2009.
This paper investigates a new problem of visualizing a set of overlapping networks. We present two methods for constructing visualization of two and three overlapping networks in three dimensions. Our methods aim to achieve both drawing aesthetics (or conventions) for each individual network and exposing the common nodes between the overlapping networks. We evaluated our approaches using biological networks including protein interaction network, metabolic network, and gene regulatory network, from the bacterium Escherichia coli and crop plants to demonstrate their usefulness to support biological analysis.
D. Fung, S. Hong, D. Hart, K. Xu, Visualizing the Gene Ontology-Annotated Clusters of Co-expressed Genes: A Two-Design Study, Proceedings of IV (Information Visualisation) 2008, IEEE, pp. 9-14, 2008.
In molecular biology, Gene Ontology (GO) has often
been used for annotation and as a data mining
dimension. A frequently performed step in microarray
analytics is the clustering of co-expressed genes by their
GO bioprocesses. Biological deductions are then made
from the visual representation of the cluster pattern.
Thus far, the question of how different representations of
GO-annotated clusters affect biological interpretation
and usability has not been investigated. In this paper, we
evaluated two representations of GO-annotated clusters
of co-expressed genes. Using a published cDNA
microarray dataset, we tested the effect of each
representation on biological interpretation. We also
reported the results of the user evaluation conducted
with bench biologists from different areas of expertise.
Our study suggests that the bipartite graph may be more
suitable for microarray analytics.
Joshua Wing Kei Ho, Tristan Manwaring, Seok-Hee Hong, Uwe Rohm, David Cho Yau Fung, Kai Xu, Tim Kraska, David Hart, PathBank: Web-Based Querying and Visualization of an Integrated Biological Pathway Database, Proceedings of CGIV (Computer Graphics, Imaging and Visualization) 2006, pp. 84-89, IEEE, 2006.
PathBank is a web-based query and visualization system
for biological pathways using an integrated pathway
database. To address the needs for biologists to visualize
and analyze biological pathways, PathBank is designed to
be user-friendly, flexible and extensible. It is, to the best
of our knowledge, the first web-based system that allows
biological pathways to be visualized in three dimensions.
PathBank demonstrates the ability to automatically generate
and layout biological pathways in response to webbased
database query about proteins, genes, a gene ontology
and small molecules. Using a novel OWL-to-relational
database schema generation approach, it can automatically
integrate biological data from different sources that
support the BioPAX exchange format (e.g. KEGG, Bio-
Cyc). The system's web interface allows both simple keyword
and complex query-based searches in the database.
The pathway visualization capability is embedded in a
Java applet. PathBank makes extensive use of client sides'
technology to reduce computational load of the server. It
also makes extensive use of open-source technology.
Joshua Wing Kei Ho, Seok-Hee Hong: Drawing Clustered Graphs in Three Dimensions. Graph Drawing 2005: 492-502.
Clustered graph is a very useful model for drawing large
and complex networks. This paper presents a new method for drawing
clustered graphs in three dimensions. The method uses a divide and
conquer approach. More specifically, it draws each cluster in a 2D plane
to minimise occlusion and ease navigation. Then a 3D drawing of the
whole graph is constructed by combining these 2D drawings.
Our main contribution is to develop three linear time weighted tree
drawing algorithms in three dimensions for clustered graph layout. Further,
we have implemented a series of six different layouts for clustered
graphs by combining three 3D tree layouts and two 2D graph
layouts. The experimental results with metabolic pathways show that
our method can produce a nice drawing of a clustered graph which
clearly shows visual separation of the clusters, as well as highlighting
the relationships between the clusters. Sample drawings are available
Weidong Huang, Colin Murray, Xiaobin Shen, Le Song, Ying Xin Wu, Lanbo Zheng: Visualisation and Analysis of Network Motifs. IV 2005: 697-702
Many of the complex networks that occur both in nature and in technology are built up from frequently recurring patterns of basic structural elements. These structural patterns known as motifs play a significant role in the function of the network. Visualisation is a useful tool for understanding the structure in a network. The quality of a visualisation can be significantly improved if it effectively displays these motifs. In this paper we present visualisations designed to highlight motifs detected through analysis. We argue that these visualisations designed to show functionally important subgraphs give a greater insight into the function of the network.