Saturday, March 15, 2008

AMIA Summit on Translational Bioinformatics

Hundreds of clinical scientists, biologists, bioinformaticians, and policy gurus descended on the swanky Intercontinental Mark Hopkins hotel for the first AMIA-sponsored Summit on Translational Bioinformatics last week. Stanford's Atul Butte rallied impressive troops for this inaugural meeting, including the leaders of all of the National Centers for Biomedical Computation (NCBCs, 7 or so total). Since translational bioinformatics is not simply about research, but about translating research into tangible benefits (clinical diagnostics, therapeutics, and standard of care), this meant a many faceted conversation involving basic researchers, large-scale integrative projects (e.g. caBIG, the NCBCs), clinical scientists, informaticians, and government agencies. This was reflected by the structure of the meeting, which consisted of tutorials; policy, technology, and organization panels; primary paper sessions; and posters covering topics ranging from how to establish collaborative projects to ontologies and phenomics.

Given the breadth of the audience, I'm sure the highlights of the conference vary from person to person. Below are some of mine:

Eitan Rubin from Ben Gurion University, Israel (Talk highlight). "Reverse translational bioinformatics: a bioinformatics assay of age, gender and clinical biomarker." A self-proclaimed biologist, Eitan presented some intriguing work in what he called "reverse translational bioinformatics" - using clinical/medical data to make useful discoveries about biology. As an additional aim, he strove to show that existing bioinformatics tools could be applied to clinical data with little modification. To do this, he took an immense data set - thousands of variables collected for tens of thousands of individuals (part of a nutrition and lifestyle survey that was epidemiological in nature), including laboratory tests, questionnaire answers, and medication data - and essentially turned it into a microarray after binning by age. Note that this was a proxy for clinical data since no such data is currently publicly available. He then subjected this array to the same kinds of analyses one would perform on an array of molecular biological data: normalization, calculation of median values, clustering by age and variable. The results encompassed both the expected and the surprising. For example, when he clustered by age, he found distinct boundaries between somewhat intuitive ages - at 12 yrs and 16 yrs for both sexes, at 40 yrs for women and again around 49, and around 45 for men; these could point to interesting biological changes going on at these age boundaries. He also plotted the median values for variables like serum lead level vs age and found distinct patterns. At this point, he has only begun to analyze the enormous amounts of data, and more interesting patterns are sure to emerge. In the meantime, it helps drive home the potential behind open data and data (and methods!) re-use.

Yael Garten from Stanford University (talk highlight). "Pharmspresso: a text analysis tool for linking pharmacogenomic concepts." [Disclaimer: Yael and I are colleagues in the same lab and I helped to critique her presentation.] Yael's work on a semantic, scoped search engine for pharmacogenomics is worth mentioning because of its immediate and potential utility. Pharmspresso allows a user to query a corpus of documents (currently about a thousand pharmacogenomic-related articles previously curated by the PharmGKB team) for keywords, genes, drugs, and/or polymorphisms occurring in the same sentences. Based on the Textpresso ontology created for mining the C.elegans literature, Pharmspresso includes semantic support for human genes, drugs, and genetic polymorphisms and additionally improves upon more general search engines such as Google and PubMed by limiting the scope of the hits to the sentence-level and returning hits color-coded within each sentence for easy evaluation of search results. Pharmspresso has already helped the PharmGKB curators and in the future will be incorporated into an automatic curation pipeline.

Selected papers to be published in BMC Bioinformatics. At the close of the conference, the surprise announcement was made that 15 of the 27 presented papers had been selected to be published in a summer issue of BMC Bioinformatics as a joint agreement between the Open Access journal and AMIA, who would foot the bill. The papers would need to be expanded and updated for submission but the peer review process had happened for the conference and so they were already considered accepted for the journal. A couple of big conferences already do something similar - ISMB/ECCB and RECOMB - but it would be great if every major conference had some kind of arrangement like this with a journal. It seems like it would be a win-win for everyone - peer-review already taken care of, an increased audience for that issue of the journal, and a nice CV boost for the authors (and no more hard decisions between presenting at a conference vs publishing in a journal). Given the fact that this was the very first meeting for this conference, it was a very nice surprise indeed.

Thoughtful A/V setup. This is simply a logistical highlight. We've all sat through our share of technical difficulties, but this conference (at least in the main room) was astonishingly free of them. A large part of this was due to the presence of dedicated A/V staff who knew just when to dim and raise the lights, cue mood music, and put up the "transition screen" - a screen blank except for the AMIA logo. This screen went up whenever a presenter's slides were NOT up, and prevented those awkward moments when the audience could see the desktop of the presenter's laptop or the view of the Powerpoint application. It was also nice not to have to see the blue or black screens when video input was changed. All in all, it imparted a much-appreciated professional touch to the conference which other meetings would do well to emulate.

In summary, there were some informative panels on various policies and the NCBCs, interesting research, and nice extras that made this first Summit on Translational Bioinformatics a big success!

No comments: