BMC Bioinformatics Methodology article

VisANT: an online visualization and analysis tool for biological interaction data

BMC Bioinformatics

Methodology article

BioMed Central

Open Access

VisANT: an online visualization and analysis tool for biological

interaction data

ZhenjunHu1, JosephMellor1, JieWu2 and CharlesDeLisi*1,2

Address: 1Bioinformatics Program, Boston University, Boston, MA 02215, USA and 2Department of Biomedical Engineering, Boston University, Boston, MA 02215, USA

Email: ZhenjunHu-zjhu@bu.edu; JosephMellor-mellor@bu.edu; JieWu-jiewu@bu.edu; CharlesDeLisi*-delisi@bu.edu* Corresponding author

Published: 19 February 2004BMC Bioinformatics 2004, 5:17

This article is available from: http://www.77cn.com.cn/1471-2105/5/17

Received: 11 September 2003Accepted: 19 February 2004

© 2004 Hu et al; licensee BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL.

Abstract

Background: New techniques for determining relationships between biomolecules of all types –genes, proteins, noncoding DNA, metabolites and small molecules – are now making a substantialcontribution to the widely discussed explosion of facts about the cell. The data generated by thesetechniques promote a picture of the cell as an interconnected information network, with molecularcomponents linked with one another in topologies that can encode and represent many featuresof cellular function. This networked view of biology brings the potential for systematicunderstanding of living molecular systems.

Results: We present VisANT, an application for integrating biomolecular interaction data into acohesive, graphical interface. This software features a multi-tiered architecture for data flexibility,separating back-end modules for data retrieval from a front-end visualization and analysis package.VisANT is a freely available, open-source tool for researchers, and offers an online interface for alarge range of published data sets on biomolecular interactions, including those entered by users.This system is integrated with standard databases for organized annotation, including GenBank,KEGG and SwissProt. VisANT is a Java-based, platform-independent tool suitable for a wide rangeof biological applications, including studies of pathways, gene regulation and systems biology.Conclusion: VisANT has been developed to provide interactive visual mining of biologicalinteraction data sets. The new software provides a general tool for mining and visualizing such datain the context of sequence, pathway, structure, and associated annotations. Interaction andpredicted association data can be combined, overlaid, manipulated and analyzed using a variety ofbuilt-in functions. VisANT Background

The growing catalogue of biological data includes infor-mation discovered by methods that detect interactionsbetween different biological molecules. Some of thesetechniques are direct and experimental (e.g. yeast two-hybrid, chromatin-immunoprecipitation (ChIP)) whileothers are indirect, predictive and computational (e.g.,phylogenetic profiling [1], protein binding prediction[2],

and cis-element detection [3] and gene expression profil-ing) Instances of such interactions are the observed or pre-dicted relationships between genes and proteins, and forthe purpose of computational storage and analysis theycan be represented as networks of functional association.Tools are needed which gather, display and facilitate anal-ysis of these large data structures.

VisANT: an online visualization and analysis tool for biological interaction data

Interaction discovery techniques continue to emerge andevolve. Although they vary in accuracy, the confidence inany particular association is highest when made by a com-bination of measures [4]. What is true for the individuallink in this case is also true for the network; that is, thatthe prediction of functional pathways and regulatory sub-units in the cell is best accomplished by the combinationof many measures of interaction, be they experimental orcomputational, between DNA, proteins, or any moleculein the cell. The results of Ideker and Thorsson, et al[5],Jansen, et al[6], and Yanai and DeLisi[7] suggest thepotential value in combining multiple interaction types inanalyzing global systems. In contrast to biologicalsequence databases, for which uses and applications arewell established, the development of databases and asso-ciated tools for organizing, mining and analyzing molec-ular systems has begun relatively recently. To date, thefocus in this rapidly evolving field has been mainly ontools for describing and visualizing experimental interac-tion networks [8-10], derived from gene expression andprotein interaction data sources. Research on new meth-ods, both computational and experimental, that describeassociations among genes and proteins will continue tonecessitate flexible data models that can grow to fit theneeds of analysis and visualization. The broader problemof multiple data type integration will largely depend onthe usefulness of these emerging data models.

Published databases such as BIND [11], KEGG [12], Pre-dictome [13] and STRING[14] provide the conceptualplatforms on which software for leveraging the full con-tent of the interactome could operate. Early efforts in thisarea, such as the protein-protein binding databases DIP[15] and PathCalling [16], demonstrated usefulness indynamic visualization of interaction networks, by allow-ing users to navigate among links in those particular datasets. Recent visualization and analysis tools such as Cyto-scape [8], MintViewer [9] and Osprey [10] have expandedthis concept. They include features for viewing and query-ing larger subsets of the interactome on a more globalscale. The tools typically operate from the viewpoint ofphysical associations between proteins, or correlated geneexpression, and include information that summarizesannotated functions, such as Gene Ontology (GO)[17]groupings, among subnetworks of linked genes or pro-teins. Missing from the current bioinformatics palette,though, is a generic interaction network tool capable ofmanaging and analyzing the more abstract forms of inter-action information that are available and regularly pub-lished. The VisANT tool, while not a complete solution inresponding to this need, is nonetheless robust and usefulfor many different data types and analyses.

Important features that VisANT offers to the research com-munity are (i) navigation of database-driven interaction

and association networks, (ii) visual comparison, manip-ulation and storage of known networks and uploadeduser-defined data, (iii) the ability to uncover orthologousnetworks, and (iv) the ability to perform exploratory datamining and basic graph operations on arbitrary networksand sub-networks, including loop detection, degree distri-bution (the distribution of edges per node) and shortestpath identification between various component genes orproteins.

Results

Design

One of the major design goals is flexibility – both withrespect to the assimilation of new types of data, and theneed for evolving a graphical interface that can fit newtechniques for describing biological networks. For exam-ple, if new computational methods are published foridentifying cis-regulatory elements upstream of yeastgenes, we want putative interactions derived from thesemethods to be easily compared with other interactions,such as those determined by chromatin immuno-precipi-tation (ChIP) assays. For some problems, users might beinterested only in those experimentally determined inter-actions, such as protein-protein or protein-DNA binding– the physical interactome. Where experimental data islimited, or biased towards genes with well-understoodfunction, research of gene networks can benefit from useof systematically derived interactions produced in sil-ico[2]. The increasingly data-driven state of research biol-ogy suggests that analysis of high-throughput data isnecessarily more exploratory than hypothesis-driven.VisANT is designed to allow a wide range of exploratoryquestions. A researcher interested in interaction data forthe uncharacterized gene YLR089C in Saccharomyces cere-visiae, which has uncharacterized function, could explorethe network of known interactions in different species forconserved features. Investigations of biologically directedquestions, such as finding transcription factors putativelylinked downstream of known receptor proteins, could beused in generating hypotheses regarding new molecularfunctions or modes of gene regulation [8,18,19].

Regardless of the motivating problem, users should beable to identify potentially meaningful features of net-works such as shortest paths, dense nodes (i.e. nodes witha large number of connected edges), highly-connectedsubgraphs, or network motifs such as directed loops orfeedforward loops which appear to be biologically impor-tant [20-23].

Implementation

We applied the Model-View-Controller (MVC)[24] designpattern for the architecture of VisANT. The tiered systemseparates the data abstraction (the process by which a par-ticular data type is represented and stored, e.g., a two-

VisANT: an online visualization and analysis tool for biological interaction data

dimensional adjacency matrix represents interactionsbetween pairs of proteins) and retrieval layers from thepresentation schema, which improves data integrity andincreases flexibility. In particular, the tiered system allowsus to put data control logic at the middle-tier to protectthe data. Since the presentation of the stored data is sepa-rated from the data itself, users can modify the visual data(such as x, y coordinates, node size and labels) withoutmodifying the data stored in the database, or makechanges in any tier, without effecting the others, whichmakes the system easily maintainable and extensible.The system is implemented using J2EE technology usinga web service layer driven by the freely available Tomcatserver. This data layer technology is both server- and plat-form-independent. It can be readily adapted to differentcomputer systems and, with additional effort, other datasources. This enables other interaction databases to reuseVisANT as a visual analysis tool by implementing a webservice layer with a database-specific Application ProgramInterface, and the VisANT data transportation format.Technical details of this implementation can be found inthe source code of VisANT, as well as its user manual thatJava Runtime Engine (JRE). The visual tool has been testedon Netscape, Mozilla and Internet Explorer browsers, run-ning Java (JRE 1.1 or greater) on Windows 2000/XP, Linuxand Mac OSX. For the most reliable performance of thesoftware, we recommend using a newer version of Java(JRE ≥on request.

Visual exploration of biological interaction network

The main interface of VisANT, the network visualizationpanel (Fig 1), displays a set of connected nodes or verticescorresponding to user selected gene IDs (the nodes) andthe experimental methods that uncovered the connec-tions (the connecting lines, or edges). Each vertex thuscontains annotation information, and each edge storesthe method used in assigning the link. Different experi-mental methods are captured on the screen by using edgesof different color; consequently different edges can havedifferent meanings. Some represent actual physical inter-actions between proteins (e.g. from yeast two-hybrid.);some connect a transcription factor to the proteinencoded by the gene downstream of the regulatorysequence to which it binds (ChIP); others represent corre-lated functions (e.g. those determined by phylogeneticprofiling[28].) Edges between transcription factors andthe products of the genes they regulate are represented by

arrows, to indicate causal direction. All other edges arecurrently undirected.

A network is constructed by entering ORF IDs, GI num-bers, or even KEGG pathway IDs for an arbitrary numberof genes, and using data obtained by one or any combina-tion of methods shown in the methods menu. Nodes cor-responding to the selected genes will then appear on thescreen, and by left clicking one or more times, they can beexpanded into an increasingly complex set of interactions.Figure 1 is a screen shot of VisANT showing the connec-tions in a segment of the MAPK regulatory network con-structed by data from Lee et. al[29], and correlations inmicroarray experiments published by Hughes, et al. [30]VisANT algorithms find paths between receptors (STE2,SHO1, and MID2) and transcription factors (STE12,SWI4) in the MAPK network, revealing complex feedbackrelationships that possibly contribute to regulatory con-trol in these pathways.

Additional functionality is supported by the Predictomedatabase, which maintains look-up tables that store andassociate synonyms and annotations for the same pro-tein/gene, and which also facilitates the integrative analy-sis of the network with function, structure and sequenceannotation. VisANT also provides functions to load user-defined interaction data with a single mouse-click, ena-bling easy comparison between different data sets. Thenumber of viewable genes, proteins and interactions canrange from few (as shown in Fig. 1) to thousands. To sim-plify and help filter the larger data sets, different layoutalgorithms combined with the built-in basic graph opera-tions, such as closed loops, help to isolate network topol-ogy features that have potential biological implications.[20-22,31,32]

The relaxing layout algorithms implemented in VisANTare all based on a similar core heuristic algorithm[33]which models a two-dimensional network of physicalobjects with mechanical forces operating along the edges.The source code for these algorithms is based on modifi-cations of a layout program distributed by Sun Corpora-tion [27]. Although the algorithms have no biologicalmeaning, they successfully separate the graph by the den-sity of the connections between subgroups of nodes, pro-viding a visual method of identifying relatively densesubgraphs within larger networks. Additional graph oper-ations are generally provided through the various filterswhose functions are detailed in the user manual on theVisANT website.

Data integration

Our public VisANT implementation currently drawsinformation from the Predictome database, based on datafrom 66 fully sequenced microbial genomes. Higher

VisANT: an online visualization and analysis tool for biological interaction data

Figure 1

Sample view of a VisAnt application. Displayed are connections in a segment of the MAPK regulatory network con-structed by data from Lee et. al.,[29] (Brown lines with arrows, indicating binding of protein to DNA) and correlations in microarray experiments published by Hughes, et al[30] (green lines), as well as links established by protein-protein binding etc. Genes for membrane-bound receptors, and related pathway proteins and transcription factors linked by physical interaction and gene expression relation are shown. Protein/DNA is represented as the nodes. Red nodes represent proteins that are annotated in at least one KEGG pathway (the quick-tip of node STE12 indicates that it maps to KEGG pathway 04010). A "-" indicates that the node is fully expanded (i.e. all connections are shown) while the "+" indicates that some links have not yet been displayed. Correlations between nodal proteins are indicated by connecting lines (edges), different colors corresponding

to different experimental methods.

eukaryotes, including worm, human and mouse, are notyet supported, although we do include parsed versions oftheir genomes, so that networks orthologous to those inmicrobes can be mined. Computational methods in thisdatabase include phylogenetic profiling, gene fusion andgene proximity data. Experimental data drawn from pub-licly available data include protein-protein and protein-DNA interactions (S. cerevisiae), as well as gene expressioncorrelation and association data. VisANT provides a gen-eral platform for the integrative research on interactionnetworks in the context of pathway, sequence, structureand associated annotation. Pathway data is provided bythe KEGG database based on the KEGG Markup Language(KGML)[35] which is currently available only for meta-bolic pathways. The COG[36] database was used to pro-vide homology information for relationships between

VisANT: an online visualization and analysis tool for biological interaction data

Figure 2

Illustration of data integration in VisANT. (A) The MAPK related network constructed from receptors and transcription factors in the pheromone-response pathway. Purple rectangles demonstrate the quick-tip obtained by mouse-overs of the edge between DIG1 and FUS3, and the nodes CHA3 and STE12 respectively. Most integration data are available only after the node has been queried against the databases, and are available under the "Available Links" submenu of the node. (B) Gen-Bank[37] record of human homology protein for CHA1 based on COG database. The homology information is available after the corresponding filter has been processed. (C) STE12 is mapped to KEGG pathway 04010 (MAPK Singling Pathway) and the pathway has been loaded with corresponding nodes highlighted. (D) Functional annotation of STE5 is loaded through the cross-

reference in SGD[49] database.

species. Annotation information is drawn from KEGG andthe Gene Ontology, and cross-referencing of genes andproteins to GenBank [37] and SwissProt is provided.When an interaction/association is discovered by morethan one method, the corresponding edge will be seg-mented with different colors corresponding to the meth-ods. These colors can be customized using the built-inmethod table. An example can be found in Figure 2A,which shows that the interaction between DIG1 and FUS3is recovered by three different experimental methods. Rednodes in Figure 2A indicate that they have been mappedto KEGG pathways. For example, quick-tips show thatSTE12 has been mapped to KEGG pathway 04010, while

CHA1 is mapped to both KEGG pathways 00260 and00272. KEGG pathways are directly referenced fromwithin VisANT, with corresponding nodes highlighted asshown in Figure 2C. For S. cerevisiae proteins, annotationfrom SGD has also been referenced, as shown in Figure2D. GenBank[37] sequence information has been refer-enced in similar fashion.

Interaction networks of arbitrary genes in microbes can beprojected onto groups of orthologous human genes,providing hypothetical relationships between humangenes. This projection is based on the COG ortholog data-base, coupled with filters provided by VisANT. Figure 2Ashows that CHA1 has two ortholog proteins in human

VisANT: an online visualization and analysis tool for biological interaction data

BMC Bioinformatics 2004, 5(19923959 and 5803161, GI number) and Figure 2B dis-plays the GenBank[37] record of human protein19923959 directly referenced in VisANT.

Network storage and sharing

Visualization and comparison of different interaction net-works (networks obtained with separate methods) is animportant means of validation and understanding the rel-ative contribution of different methods to functionalunderstanding. VisANT allows users to enter customizeddata sets through the control panel as shown in Figure 1,and to overlay these data sets upon one another, or uponpublished datasets. Where multiple data sets based onsimilar methods have been published (e.g. yeast 2-hybridscreening in S. cerevisiae), the reference to each source iscited. The data format for user-specified data is simple tab-delimited, and can represent either directed or undirectedassociations. VisANT also provides password-protectedsaving of each customized graphical workspace to allowfurther analysis of a particular network at any time fromanywhere on the internet. In addition, these individualworkspaces can be securely shared, to promote collabora-tion within and among research groups.

Discussion

Although networks and pathways can be visualized andnavigated using clickable images [15], the data miningprocess requires more than visualization. Visual data min-ing is mediated by a collection of interactive methods thatsupport exploration of data sets by adjusting parametersto see how they affect the information being presented.The functionalities provided by VisANT reflect thisapproach, especially as it applies to biological networks.Both genome-wide and conventional interaction data canbe noisy and error prone [38]. The integration of interac-tion data from various data sources is critical for improv-ing the accuracy of these data [38-42]. Data integrationalso requires the unification of heterogeneous data (suchas expression data, sub-cellular localization information,and functional category etc.) into one general data modelso that different analyses can be carried out easily. Cluster-ing of gene expression, for example, may be guided byknowledge of protein localization, or participation ofgenes in the interaction network [43].

Other visual integration tools, such as Cytoscape [8],Osprey [10] and GenMAPP[44], are able to display vary-ing aspects of physical interaction and expression dataand relate this to functions and pathway annotation.VisANT differs conceptually from these tools in the notionthat all such information – interaction, expression, func-tion – can be represented and analyzed as a network. Thedimensions of these networks can be very large, thus pre-senting a major and still incompletely met challenge for

http://www.77cn.com.cn/1471-2105/5/17

visual integration and computation. Table 1 summarizesthe differences between the three programs.

Future Directions

The goal of the VisANT project is to provide a general plat-form for visually mining process-level annotation[45].This annotation, sometimes called functional, relates thegenome to cellular processes: growth, apoptosis, differen-tiation and so on. Our first step focuses on protein/geneinteraction mining and visualization. As the interactionnetwork turns to functional modules/pathways and net-works, corresponding functions will be implemented tosupport further analyses, including simulation of cellularactivities.

Specifically, VisANT enhancements in the near future willinclude the following:

1. Visualization and graph manipulation. The data modelwill be further generalized to represent different types ofbio-objects and the interactions between them. Visual rep-resentation of nodes and edges will be enhanced andstandardized [46,47]. An immediate goal is to produce adata model and related functionalities that supportabstract groupings and modularity based on function orexperimental evidence, in order to facilitate the full inte-gration with groupings such as KEGG pathways, GOannotations, and diverse objects such as protein com-plexes [23]. These groupings enable a more modular anal-ysis of structure within interaction networks.

2. Inclusion of the full complement of KEGG pathways.3. Support for higher eukaryotes including worm, humanand mouse. Analysis and comparison of interactionsacross species will continue to be improved. Specifically,we are interested in the concept of cross-species mappingto facilitate direct comparison of the conservation of net-works between different organisms.

4. The implementation of additional features for integrat-ing data sources. For example, VisANT will be able to loadmicroarray data either from standard databases (GEO [48]etc.), or from a user's local file. Third party open-sourcesoftware, such as TM4 [34], will be integrated to enabledirect analysis of expression data in context of minednetworks.

5. VisANT's architecture will be further enhanced to ena-ble pluggable parsers and filters, providing the flexibleinterfaces to facilitate the integration of heterogeneousdata sources and third-party's analysis. Correspondingly,VisANT will be able to run as both a signed on-line javaapplet and standalone application.

VisANT: an online visualization and analysis tool for biological interaction data

BMC Bioinformatics 2004, 5http://www.77cn.com.cn/1471-2105/5/17

Table 1: Comparison of VisANT against Cytoscpe and Osprey

VisANT

Graph ManipulationData Generalization*Input

Developed in house

Node can either be Protein/Gene, chemical compound, KEGG pathway Up-to-date Predictome database, both experimental/computation data On-line network file saved per user Tab-delimited user data through copy/paste

On line saving of work space per user Export as tab-delimited file Jpg image

On-line searching using either name, GI number, ORF ID, KEGG pathway ID etc.

Various layouts

Network motif detection including feed-forward, feedback, shortest path, auto-regulation etc.

Filtering the data with the different methods

Compare different data sets

Select nodes with certain range of connections

Support 66 species

Enable orthology searching against Human genome KEGG database Genbank database SGD database On-line java applet1.1 aboveOpen source

Cytoscape

Based on yWorks packageProtein/Gene/Compound

Osprey

Developed in houseProtein/Gene

Local file with its own format for both Up-to-date Grid database with network and expression data.experimental data only

local file with its own format Local file

Image printing

Search node name on the graph Various layouts

Local file

Image printing Vector graph

Search node name on the graph Various layouts

OutputSearch

Network Operation

Filtering

Flexible filters with different attributes of node and edge

Filtering the data with the different methods/resources

Compare different data setsSupport 4 species Go database

On-line java applet Stand-alone application1.4 above

Commercial, free to academy

Multi-speciesIntegrationRun ModeJava versionLicense

N/A Go database

Stand-alone application1.4 aboveOpen source

*All three programs can display any string as a node. The components listed in this row are those that have biological annotations

We expect that these and other directions of VisANT willalso be augmented and assisted by feedback from theresearch community.

3.4.5.

Authors' contributions

JM and ZH contributed to software concept. ZH imple-mented the system and performed major programmingwork. ZH, JM and JW contributed to the underlying Pre-dictome database. This work was directed by CD. Allauthors have read and approved the final manuscript.

6.7.8.9.10.11.12.

Acknowledgements

This work was funded by NIH grant 1P20GM066401-01.

References

1.

Pellegrini M, Marcotte EM, Thompson MJ, Eisenberg D, Yeates TO:Assigning protein functions by comparative genome analy-sis: protein phylogenetic profiles. Proc Natl Acad Sci U S A 1999,96:4285-4288.

Jansen R, Yu H, Greenbaum D, Kluger Y, Krogan NJ, Chung S, EmiliA, Snyder M, Greenblatt JF, Gerstein M: A Bayesian networksapproach for predicting protein-protein interactions fromgenomic data. Science 2003, 302:449-453.

2.

Frith MC, Hansen U, Weng Z: Detection of cis-element clustersin higher eukaryotic DNA. Bioinformatics 2001, 17:878-889.

Marcotte EM, Pellegrini M, Ng HL, Rice DW, Yeates TO, EisenbergD: Detecting protein function and protein-protein interac-tions from genome sequences. Science 1999, 285:751-753.

Ideker T, Thorsson V, Ranish JA, Christmas R, Buhler J, Eng JK, Bum-garner R, Goodlett DR, Aebersold R, Hood L: Integrated genomicand proteomic analyses of a systematically perturbed meta-bolic network. Science 2001, 292:929-934.

ansen R, Greenbaum D, Gerstein M: Relating whole-genomeexpression data with protein-protein interactions. Genome Res2002, 12:37-46.

Yanai I, DeLisi C: The society of genes: networks of functionallinks between genes from comparative genomics. Genome Biol2002, 3:research0064.

Ideker T, Ozier O, Schwikowski B, Siegel AF: Discovering regula-tory and signalling circuits in molecular interactionnetworks. Bioinformatics 2002, 18 Suppl 1:S233-40.

Zanzoni A, Montecchi-Palazzi L, Quondam M, Ausiello G, Helmer-Citterich M, Cesareni G: MINT: a Molecular INTeractiondatabase. FEBS Lett 2002, 513:135-140.

Breitkreutz BJ, Stark C, Tyers M: Osprey: a network visualizationsystem. Genome Biol 2003, 4:R22.

Bader GD, Betel D, Hogue CW: BIND: the Biomolecular Inter-action Network Database. Nucleic Acids Res 2003, 31:248-250.Kanehisa M, Goto S, Kawashima S, Nakaya A: The KEGG data-bases at GenomeNet. Nucleic Acids Res 2002, 30:42-46.

VisANT: an online visualization and analysis tool for biological interaction data

BMC Bioinformatics 2004, 513.Mellor JC, Yanai I, Clodfelter KH, Mintseris J, DeLisi C: a database of putative functional links between proteins.Predictome:14.Nucleic Acids Resvon Mering C, Huynen M, J 2002, 30:306-309.

STRING: a database of predicted functional associationsaeggi D, Schmidt S, Bork P, Snel B:15.between proteins.Xenarios I, Fernandez E, Salwinski L, Duan XJ, Thompson MJ, Mar- Nucleic Acids Res 2003, 31:258-261.

cotte EM, Eisenberg D: 16.teins: 2001 update.17.

CuraGen Corporation. Nucleic Acids ResDIP: The Database of Interacting Pro- 2001, 29:239-241.Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, http://www.77cn.com.cn .

Davis AP, Dolinski K, Dwight SS, Eppig JT, Harris MA, Hill DP, Issel-Tarver L, Kasarskis A, Lewis S, Matese JC, Richardson JE, Ringwald M,Rubin GM, Sherlock G: of biology. The Gene Ontology Consortium.Gene ontology: tool for the unification18.

25: Nat Genet 2000,Roberts CJ, Nelson B, Marton MJ, Stoughton R, Meyer MR, Bennett

25-29.

HA, He YD, Dai H, Walker WL, Hughes TR, Tyers M, Boone C,JFriend SH: revealed by a matrix of global gene expression profiles.Signaling and circuitry of multiple MAPK pathways19.enceSteffen M, Petti A, Aach J, D'Haeseleer P, Church G: 2000, 287:873-880.

Sci-modelling of signal transduction networks.20.2002, Milo 3:34.

BMC BioinformaticsAutomatedNetwork motifs: simple building blocks of complexR, Shen-Orr S, Itzkovitz S, Kashtan N, Chklovskii D, Alon U:http://www.77cn.com.cnworks.Shen-Orr SS, Milo R, Mangan S, Alon U: Science 2002, 298:824-827.

transcriptional regulation network of Escherichia http://www.77cn.com.cnwork motifs in the22.Genet Ferrell J. E., Jr.: 2002, 31:64-68.

Natpositive feedback, double-negative feedback and bistability.Self-perpetuating states in signal transduction:23.Curr Opin Cell BiolBader GD, Hogue CW: 2002, 14:140-148.

molecular complexes in large protein interaction networks.An automated method for finding24.BMC BioinformaticsBuschmann Frank: 2003, 4:2.

25.system of patterns.Pattern-oriented software architecture : a26.VisANT Manual.27.VisANT. http://visant.bu.edu/vmanual/ Chichester ; New York, Wiley .; 1996:xvi, 467. 28.Java.Wu http://www.77cn.com.cn/ Http://visant.bu.edu/ .

.between genes using phylogenetic profiles., Kasif S, DeLisi C: Identification of functional links29.

19: Bioinformatics 2003,Lee TI, Rinaldi NJ, Robert F, Odom DT, Bar-Joseph Z, Gerber GK,1524-1530.

Hannett NM, Harbison CT, Thompson CM, Simon I, Zeitlinger J, Jen-nings EG, Murray HL, Gordon DB, Ren B, Wyrick JJ, Tagne JB, VolkertTL, Fraenkel E, Gifford DK, Young RA: networks in Saccharomyces cerevisiae.Transcriptional regulatory30.

298: Science 2002,Hughes TR, Marton MJ, Jones AR, Roberts CJ, Stoughton R, Armour799-804.

CD, Bennett HA, Coffey E, Dai H, He YD, Kidd MJ, King AM, MeyerMR, Slade D, Lum PY, Stepaniants SB, Shoemaker DD, Gachotte D,Chakraburtty K, Simon J, Bard M, Friend SH: via a compendium of expression profiles.Functional discovery31.

102: Cell 2000,Bu D, Zhao Y, Cai L, Xue H, Zhu X, Lu H, Zhang J, Sun S, Ling L,109-126.

Zhang N, Li G, Chen R: protein-protein interaction network in budding yeast.Topological structure analysis of the32.Acids Res NucleicFox http://www.77cn.com.cnworks., Hill CC: 2003, 31:From topology to dynamics in biochemical2443-2450.

Eades P: Chaos 2001, 11:809-815.

34.1984, A heuristic for graph drawing. Congressus Numerantium35.TM4.42:36.

KGML. http://www.77cn.com.cn/software/tm4/index.html142-160.

.Tatusov RL, Natale DA, Garkavtsev IV, Tatusova TA, Shankavaram http://www.genome.ad.jp/kegg/xml/ .

UT, Rao BS, Kiryutin B, Galperin MY, Fedorova ND, Koonin EV: COG database: new developments in phylogenetic classifica-Thetion of proteins from complete genomes.37.2001, Nucleic Acids ResBenson DA, Karsch-Mizrachi I, Lipman DJ29:22-28.

38.

GenBank., Ostell J, Wheeler DL:Edwards AM, Kus B, Jansen R, Greenbaum D, Greenblatt J, Gerstein Nucleic Acids Res 2003, 31:23-27.

M: Bridging structural biology and genomics: assessing pro-

http://www.77cn.com.cn/1471-2105/5/17

tein interaction data with known complexes.2002, Trends Genet39.Bader GD, Hogue CW: 18:529-536.

action data obtained from different sources.Analyzing yeast protein-protein inter-40.2002, Nat BiotechnolGerstein M, Lan N, 20:991-997.

41.interactomes.Sprinzak E, Sattath S, Margalit H: Science 2002, ansen R: 295:284-287.

roteomics. Integrating42.

protein-protein interaction data?How reliable are experimentalTong AH, Drees B, Nardelli G, Bader GD, Brannetti B, Castagnoli L, J Mol Biol 2003, 327:919-923.Evangelista M, Ferracuti S, Nelson B, Paoluzi S, Quondam M, ZucconiA, Hogue CW, Fields S, Boone C, Cesareni G: imental and computational strategy to define protein inter-A combined exper-action networks for peptide recognition modules.43.

2002, Brown MP, Grundy WN, Lin D, Cristianini N, Sugnet CW, Furey TS,295:321-324.

ScienceAres M., Jr., Haussler D: ray gene expression data by using support vector machines.Knowledge-based analysis of microar-44.Proc Natl Acad Sci U S ADahlquist KD, Salomonis N, Vranizan K, Lawlor SC, Conklin BR: 2000, 97:262-267.

GenMAPP, a new tool for viewing and analyzing microarray45.data on biological pathways.Stein L: Nat Genet 2002, 46.Rev GenetGenome annotation: from sequence to biology.31:19-20.

Kohn KW: 2001, 2:493-503.

Natcycle control and DNA repair systems.Molecular interaction map of the mammalian cell47.10: Mol Biol Cell 1999,Pirson I, Fortemaison N, Jacobs C, Dremier S, Dumont JE, Maenhaut2703-2734.

C: 48.Trends Cell BiolThe visual display of regulatory information and networks.49.

GEO Database. 2000, 10:404-408.

SGD Database. http://www.77cn.com.cn/http://www.ncbi.nlm.nih.gov/geo/ .

.J

本文来源：https://www.bwwdw.com/article/ooum.html