Information about cellular processes and pathways is becoming increasingly available in detailed, computable standard formats such as BioPAX and SBGN. Effective visualization of this information is a key recurring requirement for biological data analysis, especially for -omic data. Biological data analysis is rapidly migrating to web based platforms; thus there is a substantial need for sophisticated web based pathway viewers that support these platforms and other use cases.
SBGNViz fills an important gap by making the large and fast-growing corpus of rich pathway information accessible to web based platforms. SBGNViz can be used in a variety of contexts and in multiple scenarios ranging from visualization of the results of a single study in a web page to building data analysis platforms.
Citation: Sari M, Bahceci I, Dogrusoz U, Sumer SO, Aksoy BA, Babur Ö, et al. (2015) SBGNViz: A Tool for Visualization and Complexity Management of SBGN Process Description Maps. PLoS ONE 10(6): e0128985. https://doi.org/10.1371/journal.pone.0128985
Academic Editor: Nicholas James Provart, University of Toronto, CANADA
Received: March 2, 2015; Accepted: May 4, 2015; Published: June 1, 2015
Copyright: © 2015 Sari et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper.
Funding: This work was supported by The Scientific and Technological Research Council of Turkey (#113E161, http://www.tubitak.gov.tr/en), UD Foundation for the National Institutes of Health (US) (#U41HG006623 and #GM103504, http://www.fnih.org), and Defense Advanced Research Projects Agency United States Army Research Office (Big Mechanism Program) (http://www.darpa.mil, grant number: W911NF14C0119, grant recipient: ED (Dr. Emek Demir)). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Our knowledge about cellular processes increases at a breakneck rate, both in the axis of static process information captured in pathway models and interaction networks, and in the axis of dynamic -omic profiles that capture alterations and perturbations occurring in different contexts such as a disease or a condition. This rapid growth necessitates a parallel growth in our analysis and interpretation capabilities. A recurring component for a wide range of analysis scenarios is visualization of these networks. Due to the vast size and complexity of these networks, which often lead to the so-called hairball effect, ability to visualize and dynamically layout arbitrary sub-networks at multiple levels of detail is necessary for effective visualization. To keep up with the rapid, distributed growth of data, visualization tools should also use standard community formats such as BioPAX  and SBGN . Seamless access to the existing pathway data corpus and -omic profile repositories and being able to overlay this data are highly desirable. Although there are existing standalone software applications that satisfy some of these requirements such as ChiBE  and Cytoscape , currently there are no web applications offering a fulfilling solution. Since most analysis tools are migrating towards web and cloud based platforms, there is an urgent need for a web based visualization tool that can meet the aforementioned requirements.
SBGNViz is unique in that it is the only web based visualization tool for SBGN-PD maps with full support for compound structures, including a specialized compound layout capability. In addition, novel methods for reducing complexity of SBGN-PD maps without destroying the validity and integrity of processes were designed and implemented as part of SBGNViz.
In this section, we introduce the algorithms behind the novel complexity management operations provided by SBGNViz JS edition.
Many techniques for navigating and visualizing very large networks to reduce their complexity have been proposed in the literature [7, 8]. These range from visual methods like panning, zooming and fisheye views, to expanding and collapsing the nodes of the compound network, to hiding unwanted parts of the topology. However, application of such methods for process description maps, without the use of the domain information, hardly ever results in valid and complete maps (Fig 1).
Application of generic complexity management techniques without the use of domain specific knowledge typically results in invalid or incomplete process description maps. (A) An SBGN map for ATM mediated phosphorylation of repair proteins in the context of MRN complex, (B) The same map after NBN instances are selected and the rest (in this case, only one of the four processes in the map) are hidden using generic filtering, and (C) The same map when proper filtering that takes domain-specific information into account is applied.
To overcome this problem in the context of SBGN-PD maps, we keep the following invariants intact for a map, in which a specified node group is to be shown (or hidden):
- If a node is specified in a node group to be shown, it must be included.
- If a non-process node (EPN) is selected to be shown, all the processes it is involved with should be included since SBGN-PD shows the transformation of these biological entities via processes.
- If a process is selected to be shown, then all of its substrates, products, and effectors should be included as well. Hiding a substrate, product or effector might simply refer to a different process or lead to an inconsistentcy.
- If a node is selected to be shown, the parent node (a complex or a compartment) should be also included since the complexes are identified with their members and the cellular location that an entity is in might be crucial information.
- A complex molecule should always be displayed with all of its members since the member composition defines the identity of the complex.
Focusing on a node group
We now present an algorithm that takes a group of nodes and expands this group to figure out the minimal sub-map consisting of these nodes, fulfilling the above requirements. Thus, the user can focus on specified nodes, hiding any other part of the map not directly related with the specified entities.
1 nodeGroup ← nodeGroup ∪ nodeGroup.descendants()
2 nodeGroup ← nodeGroup ∪ nodeGroup.parents()
3 nodeGroup ← nodeGroup ∪ nodeGroup(‘complex’).descendants()
4 processes ← nodeGroup(‘process’)
5 nonProcesses ← nodeGroup(‘!process’)
6 neighborProcesses ← nonProcesses.neighborhood(‘process’)
7 nodeGroup ← nodeGroup ∪ processes.neighborhood() ∪ neighborProcesses
8 nodeGroup ← nodeGroup ∪ nodeGroup.parents()
9 nodeGroup ← nodeGroup ∪ nodeGroup(‘complex’).descendants()
10 return nodeGroup
As an example, suppose we are interested in a specific process and a specific molecular complex with the protein myosin as a member in a given SBGN-PD map (those nodes selected in orange in Fig 2A). If the above expand algorithm is applied, the resulting sub-map will be a valid SBGN-PD map of reduced size (Fig 3D). The steps of algorithm ExpandNodes applied to this map are detailed in Figs 2 and 3.
(A) A complex (containing myosin) and a process that consumes it are selected to be shown. (B) First step of the expansion algorithm adds all children of these nodes (myosin as a member of the complex in this case). (C) Then all parents of the current node group are included in the selection (a complex containing the myosin complex and muscle cytosol). (D) Node group is expanded to include any non-selected members of any complexes.
(A) Current selection includes a process (circled in blue) and a process (circled in red) is a neighbor of the current selection. The neighboring nodes of such processes are appended to the selection. (B) Parents (e.g., actin-myosin complex) of the current selected are included. (C) Any non-selected children of selected complexes are added to the selection. (D) Resulting map includes only those processes initially selected or that the initially selected complex is involved in. The rest are hidden.
Ignoring a node group
Sometimes though, one might be interested in doing the inverse: hiding or deemphasizing certain selected entities in a map. It is natural to think that applying the expand algorithm described earlier on the remaining nodes, those not in the specified node group, should handle this situtation. However, this might undesirably expand the complement of the map to include specified nodes as well. For example, consider the map in Fig 4A. Suppose those nodes that are to be hidden are selected in orange, including native state of RSK. If we expand the remaining nodes, the native state of RSK will be included back for display; in fact, no nodes will be hidden.
(A) Selected nodes (in orange) are to be hidden from the drawing. (B) First, selected nodes are expanded. (C) Then, the remaining nodes (in orange) are expanded. (D) Only those nodes in the expanded group calculated in C are those nodes to be displayed.
The proper way to apply expansion and adjust the specified node group for the hide case is to first expand the specified nodes, followed by expanding the remaining nodes to decide on which nodes to show. The algorithm below does exactly that.
algorithm HideSelectedNodes(nodeGroup, allElems)
1 elemsToShow ← ExpandRemainingNodes(nodeGroup, allElems)
2 elemsToHide ← allElems.not(elemsToShow)
3 hide elemsToHide
algorithm ExpandRemainingNodes(nodeGroup, allElems)
1 nodeGroup ← ExpandNodes(nodeGroup)
2 remainingNodes ← allNodes.not(nodeGroup)
3 remainingNodes ← ExpandNodes(remainingNodes)
4 return remainingNodes
SBGNViz AS edition
Project home page: https://code.google.com/p/sbgnviz-as/
Sample deployment: http://www.cs.bilkent.edu.tr/~ivis/SBGNViz.as/
Operating system(s): Web based, platform independent
Other requirements: Most common browsers with recent Adobe Flash
License: GNU Lesser LGPL
SBGNViz JS edition
Project home page: https://code.google.com/p/sbgnviz-js/
Sample deployment: http://www.cs.bilkent.edu.tr/~ivis/SBGNViz.js/
Operating system(s): Web based, platform independent
Other requirements: Most common browsers with recent versions
License: GNU Lesser LGPL
Loading and saving diagrams
Both SBGNViz editions allow input and output in SBGN-ML file format. In addition, the current drawing of the model can be saved as a static image in PNG and PDF (AS edition only) formats.
Both editions of SBGNViz comply with the standard notation of SBGN-PD maps, including support for compound structures such as molecular complexes and cellular locations. Nesting structures are respected during geometry changes such as dragging of nodes in the map as well.
A unique feature of SBGNViz is full support for compound structures, including an automated layout algorithm handling such structures without destroying proper nesting relations (Fig 7). The layout algorithm is executed on the client side using native implementations, making it more suitable for interactive applications.
Editing and highlighting
SBGNViz AS and JS editions inherit many customary network visualization features such as zooming and scrolling, deletion of nodes and edges, and highlighting of desired portions of network (Fig 8) from Cytoscape Web and Cytoscape.js libraries, respectively, adapted for SBGN-PD maps.
Inspection of gene details
SBGNViz AS edition takes biological maps to be displayed in SBGN-ML file format and renders them in SBGN-PD notation. In addition to regular network visualization software capabilities such as zoom/scroll, highlight, and deletion of nodes/edges, the tool has full support for compound structures such as compartments and molecular complexes, including a client-side specialized automatic layout capability. Furthermore, the user is allowed to dynamically inspect gene-specific information from EntrezGene  through BioGene facility.
SBGNViz JS edition has similar features to the AS edition. In addition, the JS edition will work on most common browsers out of the box, without requiring any installations, and may be used on touch enabled devices such as smart phones and tablets. Furthermore, it features novel complexity management techniques for reducing complexity of SBGN maps. This enables users to focus on a part of an SBGN map without worrying about being presented with an incomplete or invalid SBGN map.
These combined features distinguish SBGNViz from tools with similar capabilities such as BioGrapher , SBGN-ED , Cell Designer , and CySBGN . SBGNViz is unique in that it features novel methods for reducing complexity of SBGN-PD maps without destroying the validity and integrity of processes in the map. In addition, it is the only web based visualization tool for SBGN-PD maps with full support for compound structures, including a specialized compound layout capability.
Analyzed the data: OB ED UD. Contributed reagents/materials/analysis tools: MS IB SOS BAA. Wrote the paper: UD ED OB MS IB. Conceived: UD ED OB. Software design and development: MS IB SOS BAA UD. Read and approved the paper: MS IB UD SOS BAA OB ED.
- 1. Demir E, Cary MP, Paley S, Fukuda K, Lemer C, Vastrik I, et al. The BioPAX community standard for pathway data sharing. Nature Biotechnology. 2010;28(9):935–942. pmid:20829833
- 2. Novère NL, Hucka M, Mi H, Moodie S, Schreiber F, Sorokin A, et al. The systems biology graphical notation. Nature Biotechnology. 2009;27(8):735–741. pmid:19668183
- 3. Babur O, Dogrusoz U, Cakir M, Aksoy BA, Schultz N, Sander C, et al. Integrating biological pathways and genomic profiles with ChiBE 2. BMC Genomics. 2014;15:642. pmid:25086704
- 4. Cline MS, Smoot M, Cerami E, Kuchinsky A, Landys N, Workman C, et al. Integration of biological networks and gene expression data using Cytoscape. Nature Protocols. 2007;2(10):2366–2382. pmid:17947979
- 5. van Iersel MP, Villéger A, Czauderna T, Boyd SE, Bergmann FT, Luna A, et al. Software support for SBGN maps: SBGN-ML and LibSBGN. Bioinformatics. 2012;28(15):2016–2021. pmid:22581176
- 6. Demir E, Babur O, Rodchenkov I, Aksoy A, Fukuda K, Gross B, et al. Paxtools: A library for accessing, analyzing and creating biological pathway data. PloS Computational Biology. 2013;9(9):e1003194. pmid:24068901
- 7. Herman I, Melançon G, Marshall MS. Graph Visualization and Navigation in Information Visualization: A Survey. IEEE Transactions on Visualization and Computer Graphics. 2000;6(1):24–43.
- 8. Dogrusoz U, Genc B. A multi-graph approach to complexity management in interactive graph visualization. Computers & Graphics. 2006;30(1):86–97.
- 9. Lopes CT, Franz M, Kazi F, Donaldson SL, Morris Q, Bader GD. Cytoscape Web: an interactive web-based network browser. Bioinformatics. 2010;26(18):2347–2348. pmid:20656902
- 10. Cytoscape.js: A graph theory library for analysis and visualisation;. Accessed: 2015-02-20. http://js.cytoscape.org.
- 11. BioGene: an information tool for biological research;. Accessed: 2015-02-20. http://cbio.mskcc.org/biogene.
- 12. Maglott D, Ostell J, Pruitt KD, Tatusova T. Nucleic Acids Res. 2005 January;33(Database issue):54–58.
- 13. Krause F, Schulz M, Ripkens B, Flöttmann M, Krantz M, Klipp E, et al. Biographer: web-based editing and rendering of SBGN compliant biochemical networks. Bioinformatics. 2013;29(11):1467–1468. pmid:23574737
- 14. Czauderna T, Klukas C, Schreiber F. Editing, validating and translating of SBGN maps. Bioinformatics. 2010;26(18):2340–2341. pmid:20628075
- 15. Funahashi A, Morohashi M, Kitano H, Tanimura N. CellDesigner: a process diagram editor for gene-regulatory and biochemical networks. Biosilico. 2003;1(5):159–162.
- 16. Gonçalves EJV, van Iersel M, Saez-Rodriguez J. CySBGN: A Cytoscape plug-in to integrate SBGN maps. BMC Bioinformatics. 2013;14:17. pmid:23324051