In 2004, I suggested a solution to this dilemma in which high throughput might be achieved by parallelization of low-throughput traditional biochemical approaches [1
]. That is, a large number of individual biochemical laboratories might be corralled to test specific computational predictions that lie squarely in their area of expertise. I proposed that a database of predicted functions could be assembled by the computational biologists and that expert biochemists could then browse those predictions and find ones that were easily testable in their laboratories. They could then be awarded a small grant, perhaps $5–10 000, to support a student who, under the expert tutelage of the laboratory, could quickly test the prediction. This idea was sufficiently appealing to NIH (National Institutes of Health) that I was encouraged to organize a workshop in Washington to explore it further. My colleague, Simon Kasif, a computational biologist from Boston University, and I organized such a workshop in 2004 and a report was issued by the American Academy of Microbiology. Many grant administrators came to this meeting, expressed interest in the idea, but told me that they were ill-equipped to administer small grants and so the idea was shelved. However, in 2009 an RFP (request for proposals) was issued by NIH, under their Grand Challenges Initiative, that asked for novel approaches to the functional annotation of genomes. At this point, Simon Kasif and Martin Steffen, from Boston University, and I decided to pursue my original idea and respond to the RFP. We were fortunate to get it funded and the result is COMBREX (COMputational BRidge to EXperiments), a project that aims to provide functional annotation of bacterial and archaeal genomes through a grand collaboration of biochemists and bioinformaticians.
Under the leadership of Simon Kasif, the computational challenge of building the database of predictions is already well under way. A number of expert bioinformaticians have already joined the project and more are expected in the future. At the same time, Martin Steffen and I are organizing the biochemical testing of these predictions. We already have a few collaborations under way and we anticipate that there will be many more in the future. The COMBREX website (http://combrex.bu.edu
) is now open to the scientific community and we are receiving ‘bids’ from biochemists who have the expertise to test predictions. Under the current funding model, a biochemist in the U.S.A. wishing to test a prediction would receive a small grant to pay for the incremental cost of bringing a student, perhaps a rotation student, or even an undergraduate, into their laboratory to perform the necessary biochemistry. The proposal that the biochemist would submit would be fairly brief and would outline the experimental approach to be pursued to test the prediction. At that time, the gene or genes chosen would be off limits for 6 months and no other bids would be allowed until the first laboratory reported its results. Those results will be posted on the COMBREX website, irrespective of whether they are positive or negative, and we would also encourage publication of the results in the scientific literature. Indeed, we are contemplating initiating a journal specifically dedicated to such efforts.
It is important to note that because this initiative is currently being funded with U.S. funds, only U.S. laboratories have the opportunity of obtaining funding through this mechanism. However, the possibility of picking a gene to test will be open to everybody, including laboratories in Europe and the rest of the world. We hope that many U.K. scientists will join us! We are encouraging other funding agencies both within and outside of the U.S.A. to also make funds available to support this project and we have found some interest from both the Wellcome Trust in the U.K. and the Howard Hughes Medical Institute in the U.S.A. As time goes on and if the model proves effective, we anticipate that a number of non-U.S. government funding agencies might also be prepared to throw small amounts of money into the ring to support this effort.
At the present stage, the effort is limited to genes in bacteria and archaea mainly because these are more straightforward to identify. Functional predictions are often easier to make and certainly producing the proteins encoded by these genes in fully functional form and hence available for biochemical testing is generally quite facile. However, if the project is as successful as we imagine, then it could easily be extended to eukaryotic organisms. A more complete description of COMBREX is available [2
It is obvious from the previous description that what we are doing under the auspices of this project is not just providing experimental determination of gene function, but we are also building a community of bioinformaticians and biochemists eager to collaborate on this important project, which will impact all aspects of biology. Importantly, as more functions are elucidated, so our ability to make better predictions will increase. There will be a synergy between the two fields that should benefit everybody. I am reminded of my own early days in biology when collaboration was the norm. Because biology was still a very small field at the time, everyone was quite keen to share results, reagents and ideas so that progress could be made quickly. It is our hope that this COMBREX project can revive some of this collaborative spirit which was so successful in driving the early developments in molecular biology.