What is Grid Computing?

I’ve recently discovered that a grid computing service was available at the University of Liege (where I am working), inside the Alma-Grid structure. The “Alma-grid” name is quite confusing since, if I understand correctly, they offer lab solutions for genomics, proteomics, etc., and bioinformatics is only a part of it.

A Grid is a network of many computers sharing their unused ressources (CPU and/or disk storage) to solve large-scale computation problems. We can see that like a distributed computational and/or storage system. Well-known “grid computing” projects are, a.o., SETI@Home or Folding@Home. The main advantage of a grid is that you reach the power of mainframe computers in terms of CPU power and/or storage, without the cost of buying one since a grid usually uses common desktop computers distributed around the world.

Although I am sure biology will need more and more computing power, I was recently able to “feel” it … I compared proteins expressed in two conditions. For that purpose, we are using 2D-DiGE, a powerful technique enhancing classical 2D gel electrophoresis. For the result analysis, we are using a specific image comparison software that requires a lot of CPU. Sometimes, I must wait a few minutes before seeing the results of a filter or a computation; and we are using an Intel P4 with 1Gb of RAM! Later on, we had to find what proteins we’ve found (sequencing step). Only for this simple experiment (we had about 90 proteins to sequence), we generated at least 100Mb of data. Finally, the more time-consuming step was the data analysis since we only have spreadsheet software for that purpose. I impatiently wait for the day when bio-companies will provide adequate software for data analysis. Oh yes, I have one more requirement: it must be relatively cheap (the image comparison software for 2D-DiGE analysis costs more that 50 000 euros!). The dream would be to have free software (with “free” like in “free speech”). So, I hope you can imagine the power and ressources needed for big projects like sequencing all the human genes…

N.B. At least one other grid project exists in Belgium: begrid. The CERN has interesting and easy-to-understand animations on grid computing (Flash plug-in required).