11th Intelligent Systems for Molecular Biology 2003 (ISMB 2003) · 2019. 8. 1. · Comparative and Functional Genomics Comp Funct Genom 2003; 4: 654–659. Published online in Wiley

Comparative and Functional GenomicsComp Funct Genom 2003; 4: 654–659.Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/cfg.336

Meeting Review

11th Intelligent Systems for MolecularBiology 2003 (ISMB 2003)Brisbane Convention Centre, Brisbane, Australia, 29 June–3 July 2003

Catherine A. Abbott*School of Biological Sciences, Flinders University, Adelaide, SA, Australia

*Correspondence to:Catherine A. Abbott, School ofBiological Sciences, FlindersUniversity, GPO BOX 2100,Adelaide, SA, Australia.E-mail:[email protected]

Received: 22 September 2003Revised: 25 September 2003Accepted: 29 September 2003

AbstractThis report profiles the keynote talks given at ISMB03 in Brisbane, Australia by RonShamir, David Haussler, John Mattick, Yoshihide Hayashizaki, Sydney Brenner, theOverton Prize winner, Jim Kent, and the ISCB Senior Accomplishment Awardee,David Sankov. Copyright 2003 John Wiley & Sons, Ltd.

Introduction

This year for the first time the annual meeting ofthe International Society of Computational Biol-ogy (ISCB) was held in the Southern hemisphere.The 11th International Conference on IntelligentSystems in Molecular Biology (ISMB2003) came‘down under’ and was held at the Brisbane Conven-tion Centre, Brisbane, Australia, 29 June–3 July2003. Unfortunately, the promised great Aussiesun did not come out as much as was expectedbut the programming committee delivered a stim-ulating program where the ‘bio’ was once againreturned to bioinformatics.

As is traditional at ISMB, the meeting waspreceded by Special Interest Group (SIGS) satel-lite meetings and tutorials. These SIGS were wellattended, with the numbers of participants rang-ing from 98 for the Bioinformatics Open SourceConference, 100 for Biopathways, 51 for Bio-Ontologies, 28 for WEB03, and 42 for Text Mining(BioLINK). The tutorials were popular, as usual,with over 500 ISMB registrants taking part in the15 tutorials offered. These were held on tradi-tional topics, such as ‘Molecular Phylogenetics and

Evolutionary Analysis’ and ‘Molecular Modelling:building a 3D protein structure from its sequence’,and on more cutting-edge topics such as ‘Bioethicsfor Bioinformaticists’, ‘Artificial Intelligence andMachine Learning Techniques for Bioinformatics’and ‘Data Warehousing in Molecular Biology’.

The official meeting began on Monday with awarm welcome by the co-chairs of the meeting,Gene Myers and Mark Ragan, to the 919 registrantsfrom 42 different countries. This review covers thekeynote and prize-winning talks that were deliveredacross the 4 days of the meeting. All other talks arepublished as papers in the journal Bioinformatics(http://bioinformatics.oupjournals.org/). Notablespeakers at ISMB 2003 included: Sydney Bren-ner (one of three recipients of the 2002 NobelPrize in Medicine, founder of the Molecular Sci-ences Institute and Distinguished Research Profes-sor at the Salk Institute); David Haussler (HowardHughes Medical Institute Investigator and Pro-fessor of Computer and Information Sciences atthe University of California, Santa Cruz); Yoshi-hide Hayashizaki (Project Director of the GenomeExploration Research Group at the Genomic Sci-ences Center at RIKEN); John Mattick (Director of

Copyright 2003 John Wiley & Sons, Ltd.

Meeting Review 655

the Institute for Molecular Bioscience, Universityof Queensland); Ron Shamir (Professor of Com-puter Science at Tel-Aviv University); and MichaelWaterman (Professor of Mathematics, ComputerScience and Biological Science at the Universityof Southern California).

This year’s ISMB focused more on the use ofbioinformatics and computational biology to anal-yse entire biological systems and there was lessemphasis on individual standalone problems, suchas microarray analysis, building phylogenetic trees,motif- and gene-finding algorithms. Papers andkeynotes at the conference were divided into sevenmajor themes: (1) phylogeny and genome rear-rangements; (2) expression arrays and networks;(3) predicting clinical outcomes; (4) protein clus-tering, alignment and patterns; (5) transcriptionmotifs and modules; (6) structure and hiddenMarkov models; and (7) text mining and high-throughput methods; with an additional session forshort papers.

With the sequences of over 1000 genomes nowcompleted, including those of human [13] andmouse [20], the meeting opened with a thought-provoking session on phylogeny and genome rear-rangements, which included two keynote speakers.

David Haussler (University of California;Santa Cruz; http://www.cse.ucsc.edu/∼haussler)gave a fascinating presentation on ‘Identifyingfunctional elements in the human genome bytracing the evolutionary history of the bases: akey challenge for comparative genomics’. He out-lined the principles of using evolution to findgenes and other functional elements [19]. Although75 million years of evolution separate mice andhumans, comparing the genomes of the two speciesprovides a crude way of finding regions of func-tional significance. This type of approach is noisyand, while difficult, it is possible to separate thenoise from the useful information using a cali-bration point. At least half of the human genomeconsists of relics of retrotransposons, i.e. half ofour genome is the rotting carcasses of the selfishDNA that has inserted itself into our DNA over theyears. Comparative genomics will allow us to iden-tify functional elements and, as more species aresequenced, there will be an increase in our power todetect conserved elements. His group has enhancedthese studies by combining hidden Markov andphylogenetic models to create new ways of mod-elling molecular evolution [17].

One of the grand challenges of human molecularevolution is to reconstruct the evolutionary historyof each base in the human genome. Great sums ofmoney were spent on getting a draft sequence ofthe human genome, so we now need to use thisinformation fruitfully. Genome browsers, such asthat at UCSC, will become more than archives ofdata — they will be used as microscopes that canbe used to interpret and discover new things abouthuman genes. The overall goal of future browserapplications is to link gene position to functionalinformation, so that the browser can be used asan engine for discovery — a microscope searchingthe genome for new and exciting discoveries. Thesethoughts were expanded upon later in the meetingin the Overton Prize address by Jim Kent.

The second keynote address in this sessionwas John Mattick’s (University of Queensland;http://www.imb.uq.edu.au/mattick.html), ‘Pro-gramming of the autopoietic development of com-plex organisms: the hidden layer of non-codingRNA’. His presentation generated enormous excite-ment in the audience and attracted an incredibleamount of interest in the form of questions after-wards.

The number of protein-coding genes does notscale strongly with the complexity of an organism.The fly and worm genomes contain 14 000–19 000protein-coding genes, two to three times more thanyeast at ∼6000, but this is not much less than thenumber of mammalian protein-coding genes. More-over, the genome of the inanimate plant, rice, hasmore genes than that of humans. The 98% of non-coding sequence in the human genome actuallyhas a tighter correlation with human complexity.Most of the genetic variation between organismslies in these non-coding regions, indeed only 1% ofcoding genes between mouse and human are dif-ferent. Mattick gave a compelling argument thatnon-coding RNA is the genetic basis of humancomplexity and variation [14,15]. There are enor-mous numbers of non-coding RNA genes in themammalian genome, which are only now begin-ning to be recognized, and which appear to accountfor between one-half and three-quarters of all tran-scripts. At least 50%, and possibly the majority,of the human genome is transcribed. He fieldedthe hypothesis that RNAs derived from processedintrons are involved in gene–gene communicationand networking in real time in eukaryotic cells.The talk summarized the accumulating evidence

Copyright 2003 John Wiley & Sons, Ltd. Comp Funct Genom 2003; 4: 654–659.

656 Meeting Review

that spliced RNA molecules are functional; intronscomprise, on average, 95% of the primary sequenceof protein-coding transcripts in humans, the num-bers and size of introns and non-coding RNA cor-relates with developmental complexity, and someintrons/non-coding RNAs are highly conserved.

The emphasis on systems biology was contin-ued in the next session on expression arrays andnetworks. While there were many interesting pre-sentations, the one that clearly stood out wasthat of Eran Segal (Stanford University; www-cs.stanford.edu/∼eran) on ‘Discovering molecu-lar pathways from protein interaction and geneexpression’, which received the award for Best Stu-dent Paper of the meeting (more details of this workcan be found at http://bioinformatics.oupjour-nals.org/). Segal presented results combining twoSaccharomyces cerevisiae gene expression datasetswith a protein interaction dataset and applying aunified probabilistic model to discover coherentfunctional groups and entire protein complexes.Later in the meeting, he presented further workextending these concepts in his talk, ‘Discoveryof transcriptional modules from DNA sequenceand gene expression’, which was awarded the BestPaper of the meeting.

At present, we have vast amounts of datasources, DNA sequence data, gene expression dataand data from proteomic technologies such asyeast two-hybrid systems. The great challenge ishow to use this data to better understand bio-logical systems. On the second day of the meet-ing, these concepts were discussed in the keynoteaddress of Ron Shamir (Tel Aviv University;http://www.math.tau.ac.il/∼rshamir/), ‘Recon-structing genetic networks’. His presentationemphasized the importance of developing compu-tational methodologies that are rigorous, robust andrealistic to help integrate the different datasets thatare available. He emphasized the point that, asthese datasets are highly heterogeneous, the scien-tific community should not expect any single ‘killerapplication’. Tools that are developed will be tunedfor a specific task. His group has implemented theCLICK tool and tested it on a variety of biologicaldatasets, ranging from gene expression to cDNAoligo-fingerprinting to protein sequence similarity.CLICK was successfully used for gene clustering infunctional genomics, finding motif sequences, tis-sue classification and more [18,7]. Other tools thathis group has developed for the analysis of gene

expression data include the new biclustering algo-rithm SAMBA (Statistical Algorithmic Method forBicluster Analysis) and PRIMA (PRomoter Inte-gration in Microarray Analysis), a program forfinding transcription factors whose binding sites areenriched in a given set of promoters. By utilizinghuman genomic sequences and models for bind-ing sites of known transcription factors, PRIMAidentifies transcription factors whose binding sitesare significantly over-represented in a given setof promoters. Another JAVA-based tool his grouphas developed to aid clustering and visualizing ofgene expression data is EXPANDER (EXPressionANalyzer and DisplayER). This visualization toolincludes an implementation of the new CLICKclustering algorithm, as well as for other popu-lar clustering algorithms, such as K-means, self-organizing maps and hierarchical clustering. Oneof the important take-home messages of the entiremeeting, particularly from the point of view of anexperimental biologist like myself, which was high-lighted by Shamir, is the importance of improvednetworking between those developing computingalgorithms and experimentalists, to verify the rel-evance of newly developed tools. These method-ologies are in their infancy and the long-term chal-lenge is to encourage more research and effort touse these methodologies to make a dent in prob-lems related to medicine and health.

This year’s recipient of the ISCB Overton Prizewas Mr William James (Jim) Kent (UC SantaCruz; http://www.cse.ucsc.edu/∼kent). This pres-tigious prize is awarded to a bioinformatician foroutstanding accomplishment in the early phase ofa career. His talk, entitled ‘Patching and Paintingthe Human Genome’, outlined the efforts that havegone into making the human genome more under-standable to humans, i.e. providing visual represen-tations of the human genome. Kent is best knownas the researcher who ‘saved’ the human genomeproject. He wrote GigAssembler [10], a programthat produced the first full working draft assem-bly of the human genome, the Human GenomeBrowser at UCSC (http://genome.ucsc.edu [11]),just before the company Celera was to present acomplete draft of the human genome to the WhiteHouse in 2000. This feat enabled the research com-munity to keep human genomic data freely avail-able in the public domain. Kent’s talk summarizedthe goals of his work and introduced the bioin-formatics tools he has built. He outlined the work


Meeting Review 657

involved in developing tools such as: the Intronera-tor system for exploring the genome of C. elegans[12]; the program WABA, which was one of thefirst pair-hidden Markov models for the alignmentof genomic DNA of two species; Improbiser, anexpectation-maximization method to discover andcluster potential transcription factor binding sites;and the popular BLAT, which rapidly searches fullgenomes at both the DNA and protein levels [9].As discussed by Kent, the UCSC browser is con-stantly evolving to include visualization of othergenomes, such as mouse, and the goal of futureresearch will be to provide tools that allow us tovisualize multiple genomes more easily and willfocus on integrating RNA data and allowing com-parisons between genomes. The ultimate aim ofthese tools is to allow us to use genomic infor-mation more fruitfully.

This year the Society successfully introduced anew parallel stream to the ISMB program and theTuesday afternoon was dedicated to short papersin parallel with meeting reports of the variousSIGS. This new approach was well received byall attendees.

The third day of the meeting opened with a ses-sion on protein clustering, alignment and patternsand the morning concluded with a keynote from therecipient of the inaugural Senior Scientist Accom-plishment Award, David Sankoff (University ofOttawa; http://www.crm.umontreal.ca/cgi/qui?sankoff). This new award was established in orderto recognize a member of the computational biol-ogy community who is more than 12–15 yearspost-degree who has made major contributions tothe field of computational biology through research,education, service, or a combination of the three.Sankoff received this award for the immense contri-butions he has made to computational biology dur-ing his career. Over the last 15 years his work hasfocused on the evolution of genomes as the resultof chromosomal rearrangement processes [1,16].He argued that the increase in large-scale genomicsequence data will give computational biologiststhe ability to compare theoretical work to exper-imental work! The scientific community now hasmany inventories of the random rearrangement andevolution that occur amongst different genomes,and now needs to use mathematical processes tolook at these comparative maps/genomes, to getmore realistic ideas about rates and overall tenden-cies in evolution [6].

The Wednesday afternoon session began withthe next keynote of the meeting, ‘Dynamic pro-gramming algorithms for haplotype block partition-ing’, delivered by Michael Waterman (Universityof Southern California; http://www-hto.usc.edu/people/Waterman.html). His lecture focused onusing computational approaches to analyse molec-ular sequence data collected from human varia-tion and single nucleotide polymorphisms (SNPS)[21,22]. The technology is now available to scorelarge numbers of DNA variants (SNPs) in differentindividuals. The challenge is to be able to asso-ciate SNP data with different disease states whenthe problem suffers from excessive dimensionality.Waterman discussed the importance of developinga means to reduce the number of dimensions to thespace of genotype classes in a biologically mean-ingful way. Linked SNPs are often statisticallyassociated with one another (in ‘linkage disequi-librium’) and the number of distinct configurationsof multiple tightly linked SNPs in a sample is oftenfar lower than one would expect from independentsampling. These joint configurations, or haplotypes,might be a more biologically meaningful unit, sincethey represent sets of SNPs that co-occur in apopulation. Recently there has been much excite-ment over the idea that such haplotypes occur asblocks across the genome, as these blocks suggestthat fewer distinct SNPs need to be scored to cap-ture information about genotype identity. There isa need for formal analysis of this dimension reduc-tion problem, for formal treatment of the hierarchi-cal structure of haplotypes, and for considerationof the utility of these approaches toward meetingthe end goal of finding genetic variants associatedwith complex disease.

Thursday was the last day of the meeting, andboth keynote speakers once again gave presen-tations reminding us of the power of system-atic approaches in science. Yoshihide Hayashizaki(RIKEN Genomic Science Center; http://gen-ome.gsc.riken.go.jp/index.html) reviewed theamazing amount of work that has been accom-plished in analysing the mouse transcriptome inhis presentation, ‘The dynamic eukaryotic tran-scriptome’. The RIKEN centre have just com-pleted their mouse ‘encyclopedia project’, a mapof the mouse transcriptome. This project furtherreinforced the importance of computer scientistsand biologists working together, and with such ahigh level of collaboration it has been possible


658 Meeting Review

to publish the mouse genome and transcriptometogether. Hayashizaki outlined the important prin-ciples behind the RIKEN transcriptome approachvs. the EST approach, which proved an invaluableaid to annotating the human genome. The RIKENgroup decided that, as there was a saturation ofhuman ESTs and as the transcriptome was moredynamic than the genome, it was more suitable forsystematic analysis compared to the proteome. Thesecondary goal of the RIKEN mouse genome ency-clopedia project was to develop a series of new andoriginal technologies to aid investigation of tran-scriptomes [3,4]. The high-throughput sequencingtechnologies developed at RIKEN enabled themto sequence 40 000 samples per day. The FAN-TOM project led to functional annotation of 37 086cDNAs, 20 487 protein-coding genes and 16 599non-coding genes from the mouse [2]. This projectwas a tangible example of how international andinterdisciplinary collaboration can lead to excellentscience. Another of the goals of RIKEN was todevelop a format by which the FANTOM clones(the genome encyclopedia) could be easily dis-tributed to the scientific community. With some-thing that seemed to be in the realms of sci-ence fiction, or else a scientific joke, Hayashizakiannounced that a special issue of Genome Researchhad been produced — a test DNA book in whichsample cDNA clones had been spotted on to water-soluble paper [8]. Incredibly, in this proposed DNAbook, all the FANTOM clones will be clearly anno-tated, so that for each cDNA there would be a spotthat could be punched out and, using PCR, a copyof the cDNA could be made. Thus, the DNA bookprovides an ingenious way for delivering DNA in atimely and cost-effective manner to the community.

Hayashizaki concluded his talk by reminding theaudience that life science research in the twenty-first century is increasingly about connecting thegenome to the phenome, and to facilitate this wewill need to combine experimental approaches andcomputational approaches to allow new discover-ies.

These comments served as a primer for SydneyBrenner’s (Salk Institute; http://www.salk.edu/faculty/faculty/details.php?id=7) presentation,‘The evolution of genes and genomes’. Brennerprovided a terrific message, reminding the audienceof the simple concept that genes, including humangenes, have not evolved independently from oneanother. By sequencing whole genomes we have

gained much information but it needs to be consid-ered in a completely fresh light, in a more system-atic fashion which will involve computational biol-ogists working in close conjunction with biologists.For a long time we have known that the mammaliangenome is not homogenous in its GC composition,and that this base heterogeneity, or isochore struc-ture, is correlated with other important genomicfeatures, such as the insertion of repetitive ele-ments and gene density [5]. Recently, it was shownthat mutational rates are variable along the genome,pointing to a mosaic model of genome evolution.Large-scale comparisons of humans, rodents andfrog genomes will be useful for learning about theevolution of mammalian genes and knowledge ofgene chromosome position will allow visualizationof the degree of mutational variation in the genome.

Conclusions

This was my third and most scientifically enjoy-able ISMB meeting to-date. This was due tothe increased emphasis on systems biology andgenomics throughout the entire program. For amolecular biologist and a newcomer to the fieldof bioinformatics like myself, ISMB once againprovided a melting pot of opportunity to minglewith others new to the field, computer scientistsand major players in the area, thus gaining a realfeel for both the challenges and recent successesof the burgeoning bioinformatics scientific com-munity. In addition, the meeting was a chance torenew acquaintances and meet new colleagues, andthe experience reinvigorated my enthusiasm forcomputational biology and bioinformatics. Moreimportantly, the take-home message of the meet-ing, for me, was the importance of providing manyopportunities for computational biologists to workside-by-side with experimentalists to ensure thatthe work in both fields gives maximal outcomes inclinical diagnostics, human health and agriculture.These collaborations will ensure that the bioinfor-matics community comes up with robust and rig-orous solutions to the key scientific challenges thatgenomic researchers face in the future.

By the end of the extremely lively and enthusi-astic Glasgow ‘ISMB 2004’ presentation by DavidGilbert, my bags were packed and I was readyto go. In another first, the next meeting will beheld jointly with the European Conference on


Meeting Review 659

Computational Biology (ECCB) and in conjunctionwith Genes, Proteins and Computers VIII. Thusnext year’s ISMB meeting promises to deliver aprogram with a broader scope, but which will con-tinue to focus on bringing the ‘bio’ back into bioin-formatics. So, ISMB Glasgow 2004 in Scotland, allthat rain and those castles can’t wait!

References

1. Blanchette M, Kunisawa T, Sankoff D. 1999. Gene orderbreakpoint evidence in animal mitochondrial phylogeny. J MolEvol 49(2): 193–203.

2. Bono H, Kasukawa T, Furuno M, Hayashizaki Y, Okazaki Y.2003. FANTOM-DB: database of functional annotation ofRIKEN mouse cDNA clones. Seikagaku 75(2): 149–152.

3. Bono H, Yagi K, Kasukawa T, et al. 2003. Systematicexpression profiling of the mouse transcriptome using RIKENcDNA microarrays. Genome Res 13(6B): 1318–1323.

4. Carninci P, Waki K, Shiraki T, et al. 2003. Targeting acomplex transcriptome: the construction of the mouse full-length cDNA encyclopedia. Genome Res 13(6B): 1273–1289.

5. Castresana J. 2002. Genes on human chromosome 19 showextreme divergence from the mouse orthologs and a high GCcontent. Nucleic Acids Res 30(8): 1751–1756.

6. Eichler EE, Sankoff D. 2003. Structural dynamics of eukary-otic chromosome evolution. Science 301(5634): 793–797.

7. Elkon R, Linhart C, Sharan R, Shamir R, Shiloh Y. 2003.Genome-wide in silico identification of transcriptionalregulators controlling the cell cycle in human cells. GenomeRes 13(5): 773–780.

8. Kawai J, Hayashizaki Y. 2003. DNA book. Genome Res13(6B): 1488–1495.

9. Kent WJ. 2002. BLAT — the BLAST-like alignment tool.Genome Res 12(4): 656–664.

10. Kent WJ, Haussler D. 2001. Assembly of the working draftof the human genome with GigAssembler. Genome Res 11(9):1541–1548.

11. Kent WJ, Sugnet CW, Furey TS, et al. 2002. The humangenome browser at UCSC. Genome Res 12(6): 996–1006.

12. Kent WJ, Zahler AM. 2000. The intronerator: exploringintrons and alternative splicing in Caenorhabditis elegans.Nucleic Acids Res 28(1): 91–93.

13. Lander ES, Linton LM, Birren B, et al. 2001. Initial sequenc-ing and analysis of the human genome. Nature 409(6822):860–921.

14. Mattick JS. 2001. Non-coding RNAs: the architects ofeukaryotic complexity. EMBO Rep 2(11): 986–991.

15. Mattick JS, Gagen MJ. 2001. The evolution of controlledmultitasked gene networks: the role of introns and othernoncoding RNAs in the development of complex organisms.Mol Biol Evol 18(9): 1611–1630.

16. Sankoff D. 2001. Gene and genome duplication. Curr OpinGenet Dev 11(6): 681–684.

17. Seipel A, Haussler D. 2003. Combining phylogenetic andhidden Markov models in biosequence analysis. Proceedingsof the 7th Annual International Conference on Research inComputational Molecular Biology (RECOMB’2003).

18. Sharan R, Shamir R. 2000. CLICK: a clustering algorithmwith applications to gene expression analysis. Proceedingsof the International Conference on Intelligent Systems forMolecular Biology; ISMB 8: 307–316.

19. Thomas JW, Touchman JW, Blakesley RW, et al. 2003.Comparative analyses of multi-species sequences fromtargeted genomic regions. Nature 424(6950): 788–793.

20. Waterston RH, Lindblad-Toh K, Birney E, et al. 2002. Initialsequencing and comparative analysis of the mouse genome.Nature 420(6915): 520–562.

21. Zhang K, Deng M, Chen T, Waterman MS, Sun F. 2002. Adynamic programming algorithm for haplotype block parti-tioning. Proc Natl Acad Sci USA 99(11): 7335–7339.

22. Zhang K, Sun F, Waterman MS, Chen T. 2003. Haplotypeblock partition with limited resources and applications tohuman chromosome 21 haplotype data. Am J Hum Genet73(1): 63–73.


Submit your manuscripts athttp://www.hindawi.com

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Anatomy Research International

PeptidesInternational Journal of


Hindawi Publishing Corporation http://www.hindawi.com

International Journal of

Volume 2014

Zoology


Molecular Biology International

GenomicsInternational Journal of


The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014


BioinformaticsAdvances in

Marine BiologyJournal of



Signal TransductionJournal of


BioMed Research International

Evolutionary BiologyInternational Journal of



Biochemistry Research International

ArchaeaHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


Genetics Research International


Advances in

Virolog y

Hindawi Publishing Corporationhttp://www.hindawi.com

Nucleic AcidsJournal of

Volume 2014

Stem CellsInternational



Enzyme Research


International Journal of

Microbiology

Documents

11th Intelligent Systems for Molecular Biology 2003 (ISMB 2003) · 2019. 8. 1. · Comparative and Functional Genomics Comp Funct Genom 2003; 4: 654–659. Published online in Wiley