Enabling dynamic network analysis through visualization in TVNViewer
© Curtis et al.; licensee BioMed Central Ltd. 2012
Received: 20 March 2012
Accepted: 20 July 2012
Published: 16 August 2012
Many biological processes are context-dependent or temporally specific. As a result, relationships between molecular constituents evolve across time and environments. While cutting-edge machine learning techniques can recover these networks, exploring and interpreting the rewiring behavior is challenging. Information visualization shines in this type of exploratory analysis, motivating the development ofTVNViewer (http://sailing.cs.cmu.edu/tvnviewer), a visualization tool for dynamic network analysis.
In this paper, we demonstrate visualization techniques for dynamic network analysis by using TVNViewer to analyze yeast cell cycle and breast cancer progression datasets.
TVNViewer is a powerful new visualization tool for the analysis of biological networks that change across time or space.
KeywordsVisualization Dynamic network analysis Gene expression analysis
The rapid development of high-throughput technology and increasing amounts of biological data promises greater insight into the complex interactions that govern cellular function. In particular, gene expression measurements can be used to infer network relationships between genes in a cell, potentially uncovering important interactions that perturb the cellular state [1–4]. Understanding these network relationships between genes can lead to greater insight into cellular processes, such as the cell cycle or disease progressions . Traditionally, gene networks have been analyzed as static entities. However, biological processes such as development and disease progression evolve over time and react to changing environments. Representing these dynamic interactions with a single static network limits the biological insights that can be derived from analysis. Recently, biologists have attained a deeper knowledge of the functional and regulatory underpinnings of complex biological processes by studying dynamic gene-gene relationships [6–9]. In addition, recent algorithmic advancements allow these time-varying networks to be reverse engineered from a time series of molecular profiles. As techniques in dynamic network analysis continue to advance, tools that can visualize these complex networks will become increasingly important to understanding the systematic rewiring of the transcriptional regulatory circuitry that controls cell behavior.
Once a series of networks is available for analysis, these networks must be explored to find the subtle (and obvious) changes in network topology. Analysts can examine the changing network topology to find key regulators that drive the network evolution. At this point, the focus becomes more exploratory than query driven. Information visualization, “the use of computer-supported, interactive visual representations of data to amplify cognition,” as a field, touts its strengths at generating exploration-based insights, explanatory and persuasive interaction, and aesthetic representations . Visualization techniques excel at providing an explanation of the overall structure of the data or finding weak or unexpected patterns most easily recognized by humans .
Because visualization naturally enables gene network analysis, many visualization tools have been developed to explore biological networks including Cytoscape , Osprey , VisANT , and Graphle . Although many tools exist, the state-of-the-art tools in biological network analysis do not support the exploration of dynamic networks [22, 23]. While information visualization researchers have developed and evaluated techniques for dynamic network analyses of numerous kinds in other contexts, including social networks [24, 25], internet traffic networks , and even literature networks , these tools are not easily applicable in the biological context . Additionally, in our own experience exploring dynamic gene networks, we have found that multiple networks need to be visualized simultaneously and in real-time. To explore these networks, analysts must conveniently load and view a large number of networks and rapidly switch between networks to compare the topologies. Thus, given our experience and current visualization research, we have found that the visualization tools available for gene network analysis, such as Cytoscape, are insufficient to support the analysis of a large number of rewiring networks.
We present TVNViewer, an online visualization toolspecifically designed to support the discovery of spatial or temporal changes in network topology via exploration . In addition to facilitating exploratory analysis, TVNViewer allows analysts to create the intuitive visualizations required to present their discoveries. In this paper, we demonstrate how the visualizations in TVNViewer facilitate dynamic network analysis through the analysis of two real datasets. The first dataset is a yeast (Saccharomyces cerevisiae) microarray dataset that contains 5610 genes measured at 25 time points across two cell cycles . The networks at each time point have been recovered using Time-Varying Dynamic Bayesian Networks (TV-DBN) . The second dataset is a breast cancer progression and reversal dataset ; breast cells grown in a 3D culture begin as normal cells, become malignant (cancerous), and are then reverted by drugs that inhibit various signaling pathways. The networks have been recovered using Treegl . TVNViewer can be used to expose the similarities and differences of these cells states to reveal the effectiveness of various drugs.
The outline of the paper is as follows: we first discuss the implementation and design of TVNViewer and then show how the visualizations available in TVNViewer enable the analysis of the yeast and breast cancer datasets through several visualization strategies. Finally, we demonstrate the power of dynamic network analysis in TVNViewer via biological analysis of the breast cancer and yeast datasets.
TVNViewer runs as a freely available online visualization tool, accessed from http://sailing.cs.cmu.edu/tvnviewer. We present several resources for analysts to learn how to use TVNViewer: extensive online documentation, video tutorials, and five example preloaded networks. Analysts who create an optional login can store up to ten datasets directly on the TVNViewer website. However, all TVNViewer functionality is available without a login through a temporary session. Data for TVNViewer is stored securely on the website in a MySQL database. Analysts can upload data onto the website as described in the online documentation.
TVNViewer itself is implemented using Adobe ActionScript, and thus runs on all major browsers with the freely-available Adobe Flash plug-in. TVNViewer is an open-source project; the source code for TVNViewer can be downloaded from the main website. To implement TVNViewer, we built off of Flare, an easily-customized, open-source web-visualization project (flare.prefuse.org).
In addition to providing different visual representations of the data, TVNViewer allows the analyst to customize network views to the analyst’s preferred visual representation. Specifically, the analyst can adjust the size of the data nodes, choose to have the data nodes sized based on degree, adjust the font size of the labels, or change the visual thickness of the edges. Based on the size of the analyst’s screen, TVNViewer dynamically resizes the visualization to ensure that all labels and nodes fit within the visualization window. The opacity of each edge in TVNViewer represents its weight in the network and can be adjusted by the analyst. Additionally, the analyst can select what edges and node labels are visible in the visualization. For example, consider the case where a network has many edges with a low weight. In this case, the analyst increases the minimum edge threshold and all edges below this threshold disappear, revealing the strongest interactions. Another scenario is where the analyst is interested in only a handful of genes or gene groups. In this scenario, the analyst can remove all other labels from the visualization, highlighting the specific genes of interest. Providing customizable, interactive visualizations like these allows analysts to enhance their own cognition by putting their knowledge into the analysis. Rather than constantly having to remember numeric or ordinal values for edge weights, for example, the visualization off-loads those considerations to the visual cortex, allowing the analyst to focus on analytic activities rather than the trivia of edge weights which are only valuable for the analyst in so far as they generate insights .
Results and discussion
In this section, we highlight some ofTVNViewer’s visualizations available for dynamic network analysis. In each case, we use the yeast or breast cancer data to show how an analyst would use TVNViewer to discover patterns and information in the recovered set of networks. After the demonstrations, we will discuss the results of using TVNViewer for dynamic network analysis.
One-level network circle view
An important challenge in dynamic network analysis is the recognition of subtle changes in the network topology over time. In the one-level network circle view, the analyst sees all the genes in the dataset aligned in a circle layout. The genes are represented by circles (nodes) and the connections between genes are represented by edges (lines between nodes). The genes areclustered to minimize the number of edges going across the circle, keeping most edges local to tight clusters of genes around the edge of the circle. Genes are colored by this clustering; details are provided in the online documentation describing how this is done. Also, the analyst can use the tree view to view the sorting tree of how the nodes were clustered.
Two-level network with GO annotations
Analysis of temporally dependent gene-gene interactions across the yeastcell cycle
Budding yeast (Saccharomyces cerevisiae) serves as an excellent model for dynamic network learning because the molecular mechanisms of the cell cycle control system are well known . Budding yeast follows the eukaryotic cell cycle, which is divided into 4 distinct phases . The first is G1-phase (gap 1), which is the interval between mitosis and DNA synthesis where the cell is actively growing. This is followed by S-phase (synthesis) during which DNA replication occurs. The cell continues to grow during G2 (gap 2) and then divides in the M or mitosis phase. For the purpose of this study, we group the G2 and M phase and refer to it as G2M.
Studying the yeast cell cycle is a fitting scenario for utilizing TVNViewer as both an exploratory tool and a method of validation.We first generate a series of networks across time from yeast gene expression data using TV-DBN . Then we select subnetworks that are active during certain cell cycle phases and observe their temporal activity as it relates to their function. For example, Figure 7 shows a network with genes that were found to be active during the G2M-phase. Here, we observe functional groups that are clearly relevant to M-phase such as chromosome segregation, mitotic spindle elongation, and telomere maintenance. In addition, we observe GO groups like DNA repair, recombinational repair, and response to DNA damage stimulus which are indicative of G2-phase. One of the major checkpoints occurs in G2 phase, whereby cells are arrested in response to damaged or unreplicated DNA . Thus, we can conclude that these functions are aligned with what we expect from genes that are active in G2M.
From this preliminary overview of the functional significance of the genes provided by TVNViewer, we can then focus on particular genes and investigate supporting biological literature that can both confirm and explain why these genes interact. For instance, the gene HMI1 was found to be a DNA helicase and experimental results indicated that it localized in the mitochondria and was required for the maintenance of the functional mitochondrial genome . The unwinding activity of the helicase requires ATP hydrolysis and has a 3′ to 5′ polarity . Another gene in the subnetwork is YNL208W. While not much is known about the function ofYNL208W, the protein was detected in purified mitochondria . Interestingly, experimental evidence places both HMI1 and YNL208W at the same cellular location, supporting the prediction by our network that these genes interact.
Studying developmental processes such as the yeast cell cycle requires the integration of temporal and functional information. By using TVNViewer, we identify the recurring patterns of the gene subnetworks in S-phase.We also find that the functional roles of the genes in the network are consistent with the timing of network activity. This analysis canguide the exploration of biological literature to link the gene-gene interactions and formulate a summarizing regulatory mechanism.
Exploring the progression and reversal of breast cancer
Using TVNViewer, we also investigate the progression and reversion of breast cancer cells using dynamic network analysis. Functional analysis of 3D culture models of breast cancer has led to a deeper understanding of the effect of a cell’s microenvironment on tumorgenesis and metastasis . It was found that micro-environmental factors and signaling pathways have a dramatic influence on the growth dynamics and malignancy of the cells [38, 39]. Furthermore, treatment with inhibitors of various signaling molecules causes reversion of T4 cells into morphologically-normal-looking cells (T4R cells). Our objective is to analyze the functional differences amongst the different cell states.
As we can readily observe from the figures, the T4R cells are different from the S1 and T4 cells, but are also distinct from each other. The MMP-T4R network is very sparse and thus has few interactions. Notably,cell proliferation and other indicators of cancer are absent in MMP-T4R cells. On the other hand, the PI3K-MAPKK-T4R cells still display considerable cell proliferation and interaction with the extracellular matrix. PI3K-MAPKK –T4R cells also exhibit more activity such as tetrapyrole binding, demethylase activity and carbohydrate binding, all of which are absent in the other cell states. Collectively, these data suggest that although T4 cells can be morphologically reverted back to the normal-looking T4R cells, the underlying molecular mechanisms in the reverted cells are different from those in either S1 or T4 cells and from one another.
The cellular mechanisms responsible for progression through the cell cycle or the development of disease are complex and dynamic. Thus, many machine learning approaches have been designed to construct dynamic networks that model these processes. To fully exploit the information captured in these networks, we must have the visualization framework to simultaneously analyze a series of networks. However, current network visualization tools require extensive adaptation to explore a time series of networks. In many cases, this would require the generation of separate network visualizations, which cannot be easily compared and explored in real time.
In this work, we have demonstrated TVNViewer, a new visualization tool built for exploring the dynamic relationships between genes across a time series or in response to environment or disease. TVNViewer provides a clean interface that can be used to enable high-level functional and topological analysis in addition to highlighting more subtle network interactions over time. It facilitates a convenient and intuitive analysis of a yeast and breast cancer dataset that would not be possible using other gene network visualization tools. To conclude, TVNViewer can enable researchers to leverage the networks produced by machine learning and statistics methods through presenting the temporal context and combination of gene-level and functional-level information to allow for extensive biological analysis and interpretation.
Availability and requirements
·Project name: TVNViewer
·Project home page:http://sailing.cs.cmu.edu/tvnviewer
·Operating system(s): Platform independent
·Programming language: ActionScript
·Other requirements: Adobe Flash Player
·License: Noncommercial research use
·Any restrictions to use by non-academics: license needed
This work is supported by the Defense Advanced Research Projects Agency [Z931302]; National Science Foundation [DBI-0640543, IIS-0713379]; National Institutes of Health [1R01GM087694, 1R01GM093156, 0015699]; National Science Foundation Graduate Research Fellowship (Grant No. 075) to AP; Natural Sciences and Engineering Research Council of Canada Fellowship to JX; and an Alfred P. Sloan Research Fellowship to EPX.
- Li Z, Chan C: Inferring pathways and networks with a Bayesian framework. FASEB J 2004, 18(6):746–748.PubMedGoogle Scholar
- Basso K, Margolin A, Stolovitzky G, Klein U, Dalla-Favera R, Califano A: Reverse engineering of regulatory networks in human B cells. Nat Genet 2005, 37(4):382–390. 10.1038/ng1532View ArticlePubMedGoogle Scholar
- Segal E, Shapira M, Regev A, Pe’er D, Botstein D, Koller D, Friedman N: Module networks: identifying regulatory modules and their condition-specific regulators from gene expression data. Nat Genet 2003, 34: 166–176. 10.1038/ng1165View ArticlePubMedGoogle Scholar
- Schafer J, Strimmer K: An empirical Bayes approach to inferring large-scale gene association networks. Bioinformatics 2005, 21: 754–764. 10.1093/bioinformatics/bti062View ArticlePubMedGoogle Scholar
- Schadt EE: Molecular networks as sensors and drivers of common human diseases. Nature 2009, 461: 218–223. 10.1038/nature08454View ArticlePubMedGoogle Scholar
- Calvano SE, Wenzhong X, Richards DR, Felciano RM, Baker HV, Cho RJ, Chen RO, Brownstein BH, Cobb JP, et al.: A network-based analysis of systemic inflammation in humans. Nature 2005, 437: 1032–1037. 10.1038/nature03985View ArticlePubMedGoogle Scholar
- Dupuy D, Bertin N, Hidalgo CA, Venkatesan K, Tu D, Lee D, Rosenberg J, Svrzikapa N, Blanc A, Carnec A, Carvunis AR, Pulak R, Shingles J, Reece-Hoyes J, Hunt-Newbury R, Viverios R, Mohler WA, Tasan M, Roth FP, Le Peuch C, Hope IA, Johnsen R, Moerman DG, Barabasi AL, Baillie D, Vidal M: Genome-scale analysis of in vivo spatiotemporal promoter activity in Caenorhabditis elegans. Nat Biotechnol 2007, 25(6):663–668. 10.1038/nbt1305View ArticlePubMedGoogle Scholar
- Jiao Y, Tausta SL, Gandotra N, Sun N, Liu T, Clay NK, Ceserani T, Chen M, Ma L, Holford M, Zhang HY, Zhao H, Deng XW, Nelson T: A transcriptome atlas of rice cell types uncovers cellular, functional and developmental hierarchies. Nat Genet 2009, 41: 258–263. 10.1038/ng.282View ArticlePubMedGoogle Scholar
- Keller MP, Choi YJ, Wang P, Davis DB, Rabaglia ME, Oler AT, Stapleton DS, Argmann C, Schueler KL, Edwards S, Steinberg HA, Neto EC, Kleinhanz R, Turner S, Hellerstein MK, Schadt EE, Yandell BS, Kendziorski C, Attie AD: A gene expression network model of type 2 diabetes links cell cycle regulation in islets with diabetes susceptibility. Genome Res 2008, 18: 706–716. 10.1101/gr.074914.107PubMed CentralView ArticlePubMedGoogle Scholar
- Ahmed A, Xing EP: Recovering time-varying networks of dependencies in social and biological studies. PNAS 2009, 106(29):11878. 10.1073/pnas.0901910106PubMed CentralView ArticlePubMedGoogle Scholar
- Song L, Kolar M, Xing EP: KELLER: estimating time-varying interactions between genes. Bioinformatics 2009, 12(i128):25.Google Scholar
- Song L, Kolar M, Xing EP: Time-Varying Dynamic Bayesian Networks. Proceedings of the 23rd Neural Information Processing Systems NIPS, ; 2009.Google Scholar
- Robinson JW, Hartemink AJ: Non-stationary dynamic Bayesian networks: Learning Non-Stationary Dynamic Bayesian Networks. J Mach Learn Res 2010, 11: 3647–3680.Google Scholar
- Lozano AC, Abe N, Liu Y, Rossert S: Grouped graphical Granger modeling for gene expression regulatory network discovery. Bioinformatics 2009, 25(12):i110-i118. 10.1093/bioinformatics/btp199PubMed CentralView ArticlePubMedGoogle Scholar
- Parikh AP, Wu W, Curtis RE, Xing EP: TREEGL: reverse engineering tree-evolving gene networks underlying developing biological lineages. Bioinformatics 2011, 27(13):i196-i204. 10.1093/bioinformatics/btr239PubMed CentralView ArticlePubMedGoogle Scholar
- Card SK, Mackinlay JD, Shneiderman B: Information Visualization: Using Vision to Think. Morgan-Kaufmann, San Francisco, California; 1998.Google Scholar
- Card S, Mackinlay J, Shneiderman B, Kaufmann M: Readings in Information Visualization. Morgan Kaufmann, San Francisco, CA; 1999.Google Scholar
- Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker T: Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 2003, 13(11):2498–2504. 10.1101/gr.1239303PubMed CentralView ArticlePubMedGoogle Scholar
- Breitkreutz BJ, Stark C, Tyers M: Osprey: a network visualization system. Genome Biol 2003, 4(3):R22. 10.1186/gb-2003-4-3-r22PubMed CentralView ArticlePubMedGoogle Scholar
- Hu Z, Hung JH, Wang Y, Change YC, Hugan CL, Huyck M, Delisi C: VisANT 3.5: multi-scale network visualization, analysis, and inference based on the gene ontology. Nucleic Acids Res 2009, 37: W115-W121. 10.1093/nar/gkp406PubMed CentralView ArticlePubMedGoogle Scholar
- Huttenhower C, Mehmood SO, Troyanskaya OG: Graphle: Interactive exploration of large, dense graphs. BMC Bioinforma 2009, 10: 417. 10.1186/1471-2105-10-417View ArticleGoogle Scholar
- Pavlopoulos GA, Wegener AL, Schneider R: A survey of visualization tools for biological network analysis. BioData Mining 2008, 1: 12. 10.1186/1756-0381-1-12PubMed CentralView ArticlePubMedGoogle Scholar
- Suderman M, Hallett M: Tools for visually exploring biological networks. Bioinformatics 2007, 23(20):2651–2659. 10.1093/bioinformatics/btm401View ArticlePubMedGoogle Scholar
- Bonsignore EM, Dunne C, Rotman D, Smith M, Capone T, Hansen DL, Shneiderman B: First Steps to Netviz Nirvana: Evaluating Social Network Analysis with NodeXL. Computational Science and Engineering 2009, 4: 332–339.Google Scholar
- Heer J, Boyd D, Vizster, Vizster: Visualizing Online Social Networks.IEEE Symposium on Information Visualization. InfoVis:5, Minneapolis, MN; 2005:32–39.Google Scholar
- Cox D, Patterson R: NSFNET growth until. , ; 1995. http://www.caida.org/projects/internetatlas/gallery/nsfnet/index.xmlGoogle Scholar
- van Ham F, Wattenberg M, Viegas FB: Mapping Text with Phrase Nets. IEEE Trans Vis Comput Graph 2009, 15(6):1169–1176.View ArticlePubMedGoogle Scholar
- Curtis RE, Yuen A, Song L, Goyal A, Xing EP: TVNViewer: An interactive visualization tool for exploring networks that change over time or space. Bioinformatics 2011, 27(13):1880–1881. 10.1093/bioinformatics/btr273PubMed CentralView ArticlePubMedGoogle Scholar
- Pramila T, Wu W, Miles S, Noble WS, Breeden LL: The forkhead transcription factor Hcm1 regulates chromosome segregation genes and fills the S-phase gap in the transcriptional circuitry of the cell cycle. Gene & Development 2006, 20(16):2266–2278. 10.1101/gad.1450606View ArticleGoogle Scholar
- Petersen O, Ronnov-Jessen L, Howlett A, Bissell M: Interaction with basement membrane serves to rapidly distinguish growth and differentiation pattern of normal and malignant human breast epithelial cells. PNAS 1992, 89(19):9064. 10.1073/pnas.89.19.9064PubMed CentralView ArticlePubMedGoogle Scholar
- Liu Z, Stasko J: Mental Models, Visual Reasoning and Interaction in Information Visualization: A Top-down Perspective. IEEE Trans Vis Comput Graph 2010, 16(6):999–1008.View ArticlePubMedGoogle Scholar
- Shneiderman B: Dynamic Queries for Visual Information Seeking. IEEE Softw 1994, 11(6):70–77.View ArticleGoogle Scholar
- Chen KC, Calzone L, Csikasz-Nagy A, Cross FR, Novak B, Tyson JJ: Integrative analysis of cell cycle control in budding yeast. Mol Biol Cell 2004, 15(8):3841–3862. 10.1091/mbc.E03-11-0794PubMed CentralView ArticlePubMedGoogle Scholar
- Cooper GM: The Cell. 2nd edition. Sinauer Associates, Sunderland (MA); 2000.Google Scholar
- Sickmann A, Reinders J, Wagner Y, Joppich C, Zahedi R, Meyer HE, Schonfisch B, Perschil I, Chacinska A, Guiard B, Rehling P, Pfanner N, Meisinger C: The proteome of Saccharomyces cerevisiae mitochondria. PNAS 2003, 100(23):13207–13212. 10.1073/pnas.2135385100PubMed CentralView ArticlePubMedGoogle Scholar
- Monroe D, Leitzel A, Klein H, Matson S: Biochemical and genetic characterization of Hmilp, a yeast DNA helicase involved in the maintenance of mitochondrial DNA. Yeast 2005, 22(16):1269–1286. 10.1002/yea.1313View ArticlePubMedGoogle Scholar
- Sedman T, Kuusk S, Kivi S, Sedman J: Mitochondrial Genome in Saccharomyces cerevisiae. Mol Cell Bio 2000, 20(5):1816–1824. 10.1128/MCB.20.5.1816-1824.2000View ArticleGoogle Scholar
- Weaver V, Petersen O, Wang F, Larabell C, Briand P, Damsky C, Bissell M: Reversion of the malignant phenotype of human breast cells in three dimensional culture and in vivo by integrin blocking antibodies. Journal of Cell Biology 1997, 137(1):231. 10.1083/jcb.137.1.231PubMed CentralView ArticlePubMedGoogle Scholar
- Itoh M, Nelson C, Myers C, Bissell M: Rap1 integrates tissue polarity, lumen formation, and tumorigenic potential in human breast epithelial cells. Cancer Res 2007, 67(10):4759. 10.1158/0008-5472.CAN-06-4246PubMed CentralView ArticlePubMedGoogle Scholar
- Bissell MJ, Labarge MA: Context, tissue plasticity, and cancer; are tumour stem cells also regulated by the microenvironment? Cancer Cell 2005, 7: 17–23.PubMed CentralPubMedGoogle Scholar
- Weaver AM, Silva CM: Signal tranducer and activator of transcription 5b: a new target of breast tumor knase/protein tyrosine kinase 6. Breast Cancer Res 2007, 9: R79. 10.1186/bcr1794PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.