Community computing Idle cures Taking a coffee break could help find cures for or Aids. Katrina Megget looks at the future of research that harnesses the computing power of the World Community Grid

It’s not every day that just twiddling Institute in Hawthorne, US. ‘It’s like , time and money is of the your thumbs might solve some of In short that with humans and the same holds essence in medical research. humanity’s biggest health problems.  The World Community true for computers.’ This is exactly where the grid But while you are off having your Grid uses volunteers’ PCs comes in. ‘For example,’ says Jasinski, caffeine fix or gossiping over the to run simulations during Together we’re stronger ‘if we have a million molecules we water cooler, behind the scenes your inactive periods Since their conception, computers want to test to see if they will be computer could be helping to find  The vast number of have always had a strong role in effective as new drugs, we need the cure for cancer or Aids. In fact, computers involved can research, performing complex to do one computation for each 1.6 million personal computers in slash computation times algorithms and the like. However, molecule. If I only have one idle mode are currently chugging from years to weeks it has only been in recent years that computer and each calculation through a plethora of virtual  Projects range computing power has been granted takes one minute, it will take me a simulations, from folding to from drug discovery to the average worker. This has led to million minutes or about two years drug–target interactions, as part of the to simulating the way a notable improvement in research to complete my task. On the other world’s largest public humanitarian fold, but – multiple processors can work hand, if I have one million computers supercomputer known as the World could be applied to together to perform computations at a I can complete the task in about one Community Grid (WCG). any computationally faster pace. This allows them to cope minute by having each computer Now in its sixth year, the IBM demanding problem with more sophisticated models and perform one calculation.’ initiative represents a network of graphics-heavy simulations. The individual computers belonging to ‘Until grid technology came reality 530 000 volunteers in more than 80 along, scientists tended to rely on of this is countries around the world. This ‘grid’ stand-alone computers that were already works by harnessing an individual not connected to the outside world evident. For computer’s down time. Rather than and to each other,’ says Jasinski. the ‘Help fight an amusing animated , ‘This began to change with the childhood PCs run various computations using , where computational jobs cancer specially developed software. When could be controlled remotely or project,’ completed, the results are sent back to sent to a colleague. Then peer-to- some 8000 the main server to provide scientists peer networking, the falling cost years of with research information in their of processors and the advent of research quest for cures. multiprocessor PCs began to make should be It might seem like a hands-off feasible.’ cut by using the grid, approach, but the beauty of the grid In many ways the WCG is says Akira Nakagawara, president is the accelerated pace of research revolutionary, combining technology of the Chiba Cancer Centre at Chiba – what could take years in a costly and the collective power of linked University in Japan. ‘In the ordinary time-consuming lab takes just months PCs with the spirit of community to lab, it is absolutely impossible to via the relatively inexpensive and eco- provide cheap, speedy and promising do this kind of research,’ he notes. friendly World Community Grid. The results. But Jasinski denies it is a The project is screening three grid receives seven computational futurist technology, even though million small compounds against results from volunteers’ PCs every decades ago grid computing may have six targets in the hope of finding second of the day. And when the been envisioned as science fiction. about 30 chemicals that could number of computations performed ‘The World Community Grid is a great potentially treat the childhood by the grid each day is equivalent example of how the world can be cancer . Already to one PC computing continuously made smarter by pooling the goodwill, seven chemicals that specifically kill for over 180 years, the genius of grid ideas and resources of people without the neuroblastoma cells have been computing becomes apparent. demanding an excessive amount of finalised for the first target. ‘There is strength in numbers – two money or time,’ he says. And when Likewise, the separate ‘Help heads are better than one,’ explains the world is faced with an ageing conquer cancer’ project – which Joseph Jasinski, programme director population, increasing chronic disease is analysing 86 million x-ray

THINKSTOCK of IBM’s Healthcare and Lifesciences burden and poorly treated infectious crystallography images to aid 52 | Chemistry World | February 2011 www.chemistryworld.org www.chemistryworld.org Chemistry World | February 2011 | 53 Community computing understanding of cancer-related proteins – has been able to dramatically slash research times. ‘It is a long project even on the grid,’ says Igor Jurisica from the Ontario Cancer Institute in Canada. ‘But without WCG we would not be able to make this comprehensive analysis in our lifetime.’ The computation helps the team to identify characteristic gene or protein expression ‘signatures’ that can assist in diagnosing and monitoring diseases, or predicting how well patients will respond to certain treatments. Three of the lung cancer signatures are in preclinical stage already, while similar signatures in leukaemia, ovarian and head and neck cancer are in the pipeline.

Number crunchers For Stan Watowich at the University of Medical Branch in

Galveston, US, doing research in this LIBRARY PHOTO SCIENCE / PHANTATOMIX way is a no-brainer – particularly in disease areas considered high risk or One grid project is completing accurate free energy says there are very few things in life financially unattractive. Watowich’s modelling the binding of of binding calculations on highly that are infallible. While there are team is using the WCG to find 300 000 potential HIV distributed computer resources. advantages to electronic simulations, treatment cures for and protease inhibitors ‘If successful, our grid-based drug they are, in the end, only as good as and has been able to discovery approach can be applied the accuracy of the computational cut research times from an estimated to almost any disease with defined model, he says. False positives do 25 years on traditional laboratory molecular triggers,’ he adds. exist, and it requires additional multiprocessor computers to just research to eliminate these. weeks via the WCG. The end of experiments? In the case of finding new drugs, In the first phase of the project, the So does this all point to grid-based Watowich adds that the candidates team investigated about 15 different computer simulations trumping from grid-based research are proteins that could be potential drug traditional lab-based work? Jasinski really no different to leads that targets because of the role each plays says it is important to see the have entered the pipeline through in disease pathogenesis. Some 2.5 two as complementary. ‘Today’s more conventional paths. They will million small drug-like molecules science demands both approaches. still require significant medicinal were screened against these 15 A computer can certainly offer chemistry effort to be developed into proteins, resulting in over 35 million certain efficiencies that reduce drug candidates. However, he says, simulations. Now in the time and expense dramatically. the greater number of grid-generated second phase, this number has been At some point though, it becomes leads does mean an increased whittled down to 5000 promising necessary to perform experiments probability that at least one will candidate molecules and the team with physical matter.’ Likewise, he become an approved treatment. is running free energy calculations for each protein target. From these calculations, 50 of the best binding compounds for each target protein will be tested in biochemical assays to confirm the computer predictions. Active compounds will then be evaluated in culture assays and other preclinical tests with the ultimate goal of moving promising new drug candidates into clinical trials in humans with regulatory drug approval as the end result, says Watowich. Not only does this research aim to attack the high attrition rate for compounds moving through the development process by providing more promising drug candidates that should have an improved probability Volunteers can monitor of regulatory approval, Watowich their contributions using says it also aims to show it is possible screen savers and get

to accelerate drug discovery by ‘points’ for work done NAMAZU-TRON 54 | Chemistry World | February 2011 www.chemistryworld.org Modelling limitations aside, Jurisica believes that computation is going to ‘revolutionise’ medicine. He points out that it was exactly this kind of ‘brute force computing’ that enabled the identification of thousands of prognostic signatures in lung cancer. Indeed, this success has led to Jurisica’s team preparing a new WCG proposal where the main idea is to combine the knowledge of physical protein interactions, protein disorder and other properties with failed crystallisations to predict co-crystallisation experiments.

Tackling the hardest problems Watowich points out that drug discovery projects for any number of neglected, rare and infectious diseases are effectively endless. ‘The computing power of the World Community Grid will allow researchers to beyond the confines of traditional computing and launch large computing efforts that accurately model physical

processes.’ He adds that ‘the DREW KEVIN relatively low cost of the WCG makes it ideal for studying diseases this number down to 30 or 40 Complex calculations like various HIV therapy regimes – a step, that do not appear to offer an compounds that we can then order protein folding take best he says, towards more personalised attractive return on investment to and hand on to our collaborators to advantage of the grid’s treatments. Nakagawara agrees, large pharmaceutical companies’. test experimentally,’ he says. huge computing power saying the grid can be used for more Not only is the WCG an ‘incredible personalised projects like genomics resource’ for researchers who may Wide open possibilities screening or signalling pathway have no other means to test their ideas, The future of research using analysis – both of which could play notes Jasinski, it is also a starting block the WCG isn’t just about seeing a role in next generation cancer for attacking diseases that have been how proteins interact with other treatments. ‘The WCG can be applied traditionally hard-placed to treat. molecules. At New York University, to analyse the huge data sets obtained One such disease is Aids. According US, Richard Bonneau is using the from DNA sequencing, transcriptome, to the Global Health Council, in 2008, WCG to predict how proteins fold. proteomics and chemical structures to 33.4 million people around the world Taking the collection of microbes make personalised medicine possible.’ were living with HIV and 2.7 million found inside and on the skin of But this is just the tip of the iceberg people were newly infected with humans, the team is looking at when it comes to the grid’s future

the virus. While the disease can be whether proteins derived from the research potential, says Jasinski. As IBM treated with drugs, there is no cure. genes of these organisms – many of long as the computational models But perhaps more problematic, is that which have gene sequences not seen IBM’s Joseph Jasinski are in place ‘there is no real limit the virus, often called a ‘sloppy copier’, before – perform novel functions or thinks the possibilities to the kinds of research areas the is constantly evolving new variants if their resulting protein structure is for grid computing are grid can contribute to’. Already the leading to drug resistance. similar to known structures. ‘Using endless grid has been used to investigate Arthur Olson, from the the WCG, we can start to answer environmental chemistry, how to Department of Molecular Biology some of these questions and maybe produce more nutritious food and how at The Scripps Research Institute uncover new enzymes or protein to manage natural water courses for in La Jolla, US, is using the World scaffolds while trying to understand maximum benefit. But, says Jasinski, Community Grid to address this the full extent of protein fold space,’ ‘we also believe the grid can help in issue. ‘From a medical point of says Kevin Drew, a doctoral student weather and climate models and in view, evolution of resistance in in Bonneau’s team. Normally, studies of topics in social sciences and HIV infected patients is a major this would be a costly and time humanities – such as comparative threat to their continued survival. consuming exercise but the WCG studies of literature, language, music. Our approach is to understand the can help push findings into the lab, ‘The grid We are not limited to the physical mechanisms of viral resistance and he says. To date, proteins translated sciences,’ he says. ‘We are only limited to develop therapeutics that can box from more than 94 genomes have will allow by the creativity of different research the virus into a corner from which it been folded in this project. researchers to communities in their ability to create cannot escape.’ The team is screening Indeed, genomes and DNA are accurate computational models.’ Idle more than 300 000 molecules against expected to play an expanding role think beyond or not, what better excuse is multiple variations of two target in future grid research. Already, the confines there to take that lunch break? sites on the HIV protease, which Olson and his team are beginning to is involved with virus maturation. use patient data to understand the of traditional Katrina Megget is a freelance science ‘Typically we hope to whittle individual variation in response to computing’ writer based in London, UK www.chemistryworld.org Chemistry World | February 2011 | 55