Message from ISCB Microblogging the ISMB: A New Approach to Conference Reporting

Neil Saunders1*, Pedro Beltra˜o2, Lars Jensen3, Daniel Jurczak4, Roland Krause5, Michael Kuhn6, Shirley Wu7 1 School of Molecular and Microbial Sciences, University of Queensland, St. Lucia, Brisbane, Queensland, Australia, 2 Department of Cellular and Molecular Pharmacology, University of California San Francisco, San Francisco, California, of America, 3 Novo Nordisk Foundation Center for Protein Research, Panum Institute, Copenhagen, Denmark, 4 Department of , University of Applied Sciences, Hagenberg, Freistadt, Austria, 5 Max-Planck-Institute for Molecular Genetics, Berlin, Germany, 6 European Molecular Biology Laboratory, Heidelberg, Germany, 7 Stanford Medical Informatics, , Stanford, California, United States of America

Cameron Neylon entitled FriendFeed for Claire Fraser-Liggett opened the meeting Scientists: What, Why, and How? (http:// with a review of metagenomics and an blog.openwetware.org/scienceintheopen/ introduction to the human microbiome 2008/06/12/friendfeed-for-scientists-what- project (http://friendfeed.com/search?q = why-and-how/) for an introduction. room%3Aismb-2008+microbiome+OR+ We—a group of science bloggers, most fraser). The subsequent Q&A session of whom met in person for the first time at covered many of the exciting challenges The International Conference on Intel- ISMB 2008—found FriendFeed a remark- for those working in this field. Clearly, ligent Systems for Molecular Biology ably useful tool for taking notes and solutions to problems that we thought were (ISMB) has become an important com- sharing them online. With a core group largely solved from research into single munication hub for bioinformaticians, and of ten contributors, we covered parallel , such as assembly and the core element of the Conference— sessions, leading to a more comprehensive annotation, are being redefined in the presentations of peer-reviewed papers—is set of notes than a single person could context of metagenomics. What better now only one of many activities. Presen- take. Some presentations, particularly the way to start the meeting than to define tation of timely journal publications (the Keynotes, were covered by several people, the open questions and to highlight the Highlights sessions), Special Sessions orga- generating a detailed overview of a talk opportunities for bioinformaticians to con- nized by experts in the respective fields, from different perspectives. A virtual tribute to the field. Tutorials, and Special Interest Group meeting space—a ‘‘room’’—was created Day one closed with David Jaffe’s meetings should attract attendees who containing material relevant to ISMB excellent Introduction to Next-Generation might otherwise prefer smaller, more 2008, in which members posted an item (http://friendfeed.com/e/ focused meetings. In addition to these detailing the upcoming presentation. 685a365c-9ef6-4ec1-b5b7-48e2619a0790/ formal activities, an important aspect is the Notes on the presentation then took the Keynote-2-enormous-amounts-of-sequencing/). informal communication between partici- form of comments posted by the attendees. David illustrated the technology using a pants. This year, about 1,600 participants This proved effective: In addition to notes, range of examples: chromatin modifica- attended the meeting in the conference questions were asked and answered, links tion, high-throughput polymorphism dis- center under the CN Tower in Toronto. to relevant resources were posted, and covery in bacterial and eukaryotic ge- ISMB 2008 also left a footprint on the comments were even added by people not nomes, and, finally, de novo genome Web, via a Web service named Friend- attending the Conference. assembly using short reads. He introduced Feed (http://friendfeed.com/), to capture In addition to the live information some of the new tools required to work highlights from the Conference in near stream, the ISMB 2008 room is a with short reads, such as the ALLPATHS real time. FriendFeed allows users to share permanent, searchable archive from which assembly algorithm, demonstrating again how new technology drives software de- items, either directly or by importing their this meeting report was compiled. It is velopment in bioinformatics. Next-gener- latest content from any Web site that freely available on the Web at http:// ation sequencing continues to gain atten- generates an RSS feed, leading to a friendfeed.com/rooms/ismb-2008/. tion, and David’s enthusiasm for the topic continuous stream of information around led one commenter to note that ‘‘short which communities build (Figure 1). In Keynotes reads sequencing can be used as THE addition, and of most relevance to ISMB, The eight Keynote talks this year were general-purpose tool.’’ FriendFeed acts as a microblogging plat- form: Users post short, typically single- noted for their high quality and breadth is a regular guest at ISMB. sentence messages which generate conver- and were covered online with long com- In light of his major contributions to sations in the ensuing comment threads. ments threads. sequence analysis, it may surprise you to Microblogging, best exemplified by the Twitter service (http://twitter.com/), is Citation: Saunders N, Beltra˜o P, Jensen L, Jurczak D, Krause R, et al. (2009) Microblogging the ISMB: A New popular in the IT/tech sector, but little Approach to Conference Reporting. PLoS Comput Biol 5(1): e1000263. doi:10.1371/journal.pcbi.1000263 used by life scientists. It may be thought of Published January 30, 2009 as the fusion of instant messaging and Copyright: ß 2009 Saunders et al. This is an open-access article distributed under the terms of the Creative traditional blogs: Anyone can follow or Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, join a conversation, and conversations are provided the original author and source are credited. archived. We recommend an article by * E-mail: [email protected]

PLoS Computational Biology | www.ploscompbiol.org 1 January 2009 | Volume 5 | Issue 1 | e1000263 Figure 1. A collage of conversations from the FriendFeed ISMB 2008 room doi:10.1371/journal.pcbi.1000263.g001 learn that his new passion is Image Analysis these days,’’ he presented a diverse set of that image analysis focuses currently on (http://friendfeed.com/e/60f95dea-7f5d- analyses: tubulin tracking in mitosis, modeling and annotation, Eugene out- 454d-a296-abddbeb69cab/Keynote-3-Eugene- stereotypy-tracking cell position during lined the ultimate goal: the ability to Myers-imaging/). The take-home mes- worm development, motion tracking of mine information from this type of sage from Eugene’s Keynote on day two mouse whiskers, and, finally, early results experimental data. was that digital microscopy is now a from ambitious projects to create atlases The second Keynote of day two was a high-throughput technique, with the of fly and mouse brain. His ‘‘worm highlight: the presentation of the Overton potential to generate terabytes of data. straightening algorithm’’ was particularly Prize to . In her presentation Noting that ‘‘my life is very colorful well-received by the audience. Noting Modular Biology: The Function and

PLoS Computational Biology | www.ploscompbiol.org 2 January 2009 | Volume 5 | Issue 1 | e1000263 Evolution of Molecular Networks (http:// opening talk of day four on Intriguing Vienna. These figures were prepared using friendfeed.com/e/29b67aac-38f2-4dd5-8c19- Roles for Small ncRNAs in Cellular Regu- the Web application TagCrowd (http:// b8596895d36e/Keynote-Aviv-Regev-on- latory Networks (http://friendfeed.com/e/ tagcrowd.com/). Modular-Biology/), Aviv looked beyond 29ec0f3d-5a1c-4f2d-807a-5f42e959bf50/ Although this is a crude analysis, several the traditional representation of biological Keynote-7-Hana-Margalit-intriguing-roles- interesting trends can be discerned. The networks as, to quote a FriendFeed for/). In another example of the two-way terms mass, peptide, and ms are on the rise, comment, ‘‘nasty hairball diagrams,’’ to traffic between computational analysis indicating increased use of mass spectrom- ask: Is the concept of a functional module and biological discovery, Hanah’s group etry. In 2008, vector and machine appear, useful? Judging by her research, the developed new algorithms to predict highlighting the current popularity of answer is a resounding ‘‘yes.’’ Aviv showed ncRNAs and their targets. Applying SVMs in bioinformatics. We also see in- the application of the modular concept to these algorithms, they demonstrated that creased usage of the terms accuracy/accurate biological systems including evolution of viruses use miRNAs to downregulate the and performance. Although Phil Bourne the regulation of gene expression and host immune system. With infectious joked during his presentation on Pharmaceu- analysis of gene expression in cancer. enthusiasm, Hanah presented a set of tical Off-Targets (http://friendfeed.com/e/ Unusual for keynotes, she also presented recent experiments that validate this d5b274e5-1d34-4108-853e-5974b2a4cccd/ novel data: a comparative metabolic exciting new discovery. HL29-Phil-Bourne-off-targets-for-some- analysis of carbon utilization across 13 The last of the Keynotes was David major/) that he was ‘‘too old to compare yeast species. The evolution of networks Haussler on 100 Million Years of Evolu- my method to other methods by showing was a key theme and linked well to the tionary History of the that we’re better on a completely differ- final Keynote of the Conference. Aviv (http://friendfeed.com/e/d6af6373-897d- ent dataset,’’ this type of statistical concluded that the temporal and spatial 483d-975e-5204ef127f96/Keynote-8-David- validation seems to be used more widely key variables are still missing from network Haussler-100-million-years-of/). Impro- than ever. Clustering is less popular in models. vising effortlessly through a brief techni- 2008, and increasingly we are referring The Keynote presentations took a more cal glitch, he reminded us just how much to datasets rather than just data and to clinical turn on day three, with Morag new biology can be learned by obtaining networks rather than a single network.We Park on Profiling the Breast Tumor and comparing complete genome se- also discern the ever-growing develop- Microenvironment (http://friendfeed.com/ quences. Examples ranged from the ment of methods and algorithms and more e/04150591-adf3-4e16-a6ed-7fbeaf7d5ac9/ more familiar: ultra-conserved regions emphasis on the biological and experimental. Keynote-5-Morag-Park-profiling-the-breast- in the human genome, the HAR1 gene, tumor/). Proving that the ISMB is as which looks increasingly to be a key Biopathways SIG much about biologically relevant research factor in human brain development, to findings as computation and algorithm the almost-quirky—the discovery of an The Ninth Annual BioPathways development, Morag illustrated how new ancient transposon most similar to coe- (http://www.biopathways.org/) meeting microdissection techniques enable gene lacanth. David concluded with an outline was dedicated to computational methods expression analysis in subpopulations of of his grand challenge: reconstructing the for synthetic biology and was organized in tumor cells. Having identified distinct evolutionary history of each base in the collaboration with Emergence (http:// expression profiles for different cell types, genome of humans and other mammals. www.emergence.ethz.ch/), an EU-funded her research team has developed predictors Once again, it was clear that many consortium that fosters synthetic biology in that correlate with clinical outcomes and opportunities exist for bioinformaticians Europe. Over two days, biological path- permit customization of individual treat- to contribute the tools and analyses ways were discussed under the topics of ment. Her presentation was an excellent required for this task. network reconstruction, database/soft- demonstration that bioinformatics can im- ware development, and network evolution. pact directly the quality of human life. Summary of the Sessions The need for standardization and In the afternoon session, we witnessed machine-accessible pathway data was ap- a tour-de-force Overview of Systems Our notes are a partial record of ISMB parent. talked about recent Biology (http://friendfeed.com/e/ef87914f- 2008, reflecting the relatively small num- developments at the MetaCyc (http:// c3c5-4943-88ad-c518e9991fb4/Keynote-6- ber of bloggers and their interests. One metacyc.org/) and BioCyc (http://biocyc. Bernhard-Palsson-systems-biology-an-era/) can imagine that with more contributors, org/) pathway databases. The number of from Bernhard Palsson. His talk led us an impression of the most popular presen- available reactions and pathways in these through a brief history of systems biology tations and topics could be gained from databases has increased at an impressive to his new paradigm: a comprehensive, the level of online coverage. Sessions from rate due to curation. The nonredundant structured knowledge base that we can the Biopathways SIG, Highlights, Pro- MetaCyc database now holds 7,144 reac- use for metabolic reconstruction. Bern- ceedings, and Special Sessions tracks were tions from 1,138 pathways. Capturing hard stressed the role of community covered online, in particular those discuss- pathway information locked away in annotation in ‘‘filling in the gaps’’ and ing protein–protein interactions, biological published literature could benefit from described how this works, in the form of networks, and the future of scientific voluntary curation. This is the objective annotation jamborees targeted at model publishing. One approach to provide an and the experiment under way at Wiki- organisms. His presentation received the overview of the meeting is a ‘‘word cloud,’’ Pathways (http://www.wikipathways.org/ highest number of comments and was indicating commonly used words mined index.php/WikiPathways). clearly of great interest to non-attendees from Conference Abstracts. Figure 2 The availability of biological informa- following the coverage. shows the 100 most frequently used words tion in a machine-readable format im- Small non-coding RNAs continue to be from the 2008 ISMB Proceedings Ab- proves our ability to move data between a hot topic. A new twist in the tale was stracts, compared with the Proceedings analytical tools. Jennifer Gardy gave an provided by in the from the 2007 ISMB/ECCB meeting in interesting example in Cerebral (http://

PLoS Computational Biology | www.ploscompbiol.org 3 January 2009 | Volume 5 | Issue 1 | e1000263 Figure 2. Word cloud analysis of the Proceedings abstracts from ISMB/ECCB 2007 (left) and ISMB 2008 (right) doi:10.1371/journal.pcbi.1000263.g002 www.pathogenomics.ca/cerebral/), a Cy- kinase-substrate recognition obtained altering precisely the activity of one toscape plugin that facilitates pathway from NetPhorest. They explained how cellular component for medical applica- layering according to location informa- integration of the growing list of classi- tions might not be the best approach to tion. Cerebral was also developed for the fiers from NetPhorest might be used to follow. Both , in relation to comparison of multiple quantitative data- predict signaling pathways. cancer therapy, and Phil Bourne, in sets, such as gene expression data across Chris Sander challenged the oft-used relation to prediction of drug off-target different conditions. phrase: ‘‘There is no such thing as effects, mentioned the need to understand Novel network analysis methods were pathway reconstruction, nor is there such the effect of pharmacological interventions described by David Gilbert and Shoshana [a] thing as pathway re-engineering. within the context of complete biological Wodak. The Gilbert group has been work- Pathways are simply biological models.’’ pathways. ing on a language to specify semiquanti- He described an interesting approach to tative model rules that can be used for derive biological models from a significant Web 2.0 for Science Birds of a model verification. This could also be used collection of observations obtained from Feather Session to search pathway model databases for combinatorial perturbation experiments. pathways that can implement a desired Using a cancer cell line, researchers Many scientists make little use of public behavior for engineering purposes. performed a set of 21 drug pair treatments Web 2.0 resources for creativity, sharing, presented a new algo- and observed phosphoproteins and cell and collaboration. Bloggers Shirley Wu rithm to find paths within a metabolic cycle markers. Using a novel computa- (http://shirleywho.wordpress.com/) and network, given start and end point metab- tional approach, they were able to redis- Pedro Beltra˜o (http://pbeltrao.blogspot. olites. cover many of the functional interactions com/) co-chaired a Birds of a Feather Acquisition of pathway information among the genes and to highlight potential (BoF) discussion—Science Blogging and relies increasingly on large-scale datasets. new connections. Web 2.0—attended by about 20 people. A Finding functional relevant pathways Are pathway models obtained from the lively discussion ensued, examining the through analysis of large datasets is some- literature or from novel mining approach- reluctance and the enthusiasm of life times referred to as network reconstruc- es useful for medical or technological scientists to embrace Web 2.0 (http:// tion, a topic covered by Chris Sander, applications? Chris Sander, Phil Bourne, van.embl.de/cb/web_20_talk_series.shtml). Lars Juhl Jensen, and Rune Linding. Lars and Christopher Myers all touched on this The first topic was online reference mana- presented NetPhorest (http://netphorest. subject from different angles. Myers told gers: Connotea (http://www.connotea.org/), info/), a compendium of classifiers for us that all biological pathways are ‘‘slop- CiteULike, (http://www.citeulike.org/), and linear motifs important for phosphorylation- py.’’ Their analysis of different models new entrant Labmeeting (http://www.lab- dependent signaling. Rune described shows that individual parameters are meeting.com/signin), which allow tagging NetworKIN (http://networkin.info/search. typically poorly constrained by collective and online storage of papers. Why don’t php), a tool to predict in vivo kinase fits, even when abundant observations are more scientists use them? A major reason substrates by integrating gene functional available. The focus of experimental seems to be the reluctance to share what association scores from the STRING design should move away from determin- one is reading with the rest of the world, (http://string.embl.de/) database, in vivo ing precise reaction parameters to collec- based on the fear that a reading list is a phosphorylation data, and classifiers for tive behaviors. This also implies that window into your research. More gener-

PLoS Computational Biology | www.ploscompbiol.org 4 January 2009 | Volume 5 | Issue 1 | e1000263 ally, scientists are hesitant to use the About half of the participants knew about ond Life (http://secondlife.com/). In the open Web as an incubator for ideas and the Conference microblogging. Everyone future, might all scientific conferences live would rather rely on a tight circle of agreed that coverage of ISMB on the on long after the meeting in virtual trusted individuals. Conference Web site could be improved environments? Those interested should Other issues discussed included uncer- by something similar. The Conference read Jean-Claude Bradley’s description tainty surrounding intellectual property of organizers also agree and hope to include (http://usefulchem.blogspot.com/2008/02/ Web content, the problem of critical mass live microblogging at ISMB 2009 in acs-island-on-second-life.html) of how this for social networking sites, the reliability Stockholm. can be achieved. and expertise of bloggers, and the possi- bility of domination by shallow collabora- Poster Sessions Summary tions. Pedro stated that the problem is not one of too many superficial connections, The scale of ISMB is illustrated by the Microblogging platforms and other but that too few connections are being poster sessions: approximately 800 posters, tools for videos, podcasts, and virtual made. Tools such as Epernicus (http:// presented over two evenings. How might environments provide an untapped poten- www.epernicus.com/) could remedy this poster presentation before, during, and after tial for science conferences. Our experi- by connecting, for example, biologists and meetings be improved? Shirley Wu dreams ment using FriendFeed to cover ISMB bioinformaticians working on complemen- of a future (http://friendfeed.com/e/ 2008 was educational and surprisingly tary research. 44118735-65e8-0980-64c3-36247dd8dfab/ successful. We found that it enhanced Many participants had a positive view Off-to-ISMB-2008/) without the problems our note-taking skills, allowed us to of online tools such as social networking of printing and transport, where rolls of compile notes from parallel sessions, and blogging. Roland Krause described paper in cardboard tubes are replaced by attracted wider interest from non-attend- how a connection made on a social memory sticks and flat-screen displays. ees, and, in addition to the ‘‘live’’ aspect, networking site resulted in collaboration Shouldn’t posters be permanently ar- generated a permanent archive of the and a journal publication. Participants chived? Sharing posters digitally using an meeting. agreed that blogging or joining communi- image hosting service met with limited suc- ISMB/ECCB 2009 will be held in Stock- ties such as FriendFeed, Nature Network cess (http://friendfeed.com/e/4e094687- holm. We look forward to the new develop- (http://network.nature.com/), and Open- f68e-5c51-f3ce-fac663f18c23/STRING- ments in Web usage by scientists that are WetWare (http://www.openwetware.org/ and-STITCH-known-and-predicted/). A sure to emerge between now and then. We wiki/Main_Page) can result in authorship, search for Web sites that host posters also anticipate new and exciting ways to job offers, and a vastly increased audi- located the e-posters archive (http://www. report from Stockholm as it happens; ence—at least compared with your papers. eposters.net/). However, this site serves as perhaps the ISMB/ECCB 2009 Web site Consensus from those who participate in a PDF repository and lacks the visual and will look something like this: http://www. science online is that the Web is a great community features expected. Clearly, an bork.embl.de/,jensen/ismb2008/keynotes. opportunity exists for a well-designed equalizer: Students and post-docs can php.html? easily strike up ‘‘conversations’’ with more service catering to those who wish to established scientists, and contributions archive and share posters on the Web. Acknowledgments have less to do with status and more to Imagine that you were able to host a do with ideas. Web participation also poster session at any time, with visitors We would like to thank all those who partici- opens your eyes to other perspectives and from anywhere in the world. Organiza- pated in the Conference coverage through enables you to learn from people with tions including Nature Publishing Group reading, commenting, and discussing at Friend- more expertise than you. and the American Chemical Society are Feed and elsewhere on the Web. The BoF session ended with a discus- already running virtual poster sessions, sion about coverage of ISMB Toronto. meetings, and interactive displays in Sec-

PLoS Computational Biology | www.ploscompbiol.org 5 January 2009 | Volume 5 | Issue 1 | e1000263