Monkey Talk: Cluster Opinions and Insights from Cluster Monkey.

By Douglas Eadline

The Perfect Match

In case you did not know, Biologists have spent a great deal of time putting together a text file that is three billion characters long (The Human Genome). And, next on the list is to figure out what it all means. Not an easy problem. Insights often come from searching the genome with the hope it will yield some biological sentence or paragraphs that helps us understand the mystery of life. This searching it turns out has an insatiable appetite for compute cycles.

The workhorse program used to examine the genome (human or otherwise) is called BLAST, which stands for Basic Local Alignment Search Tool. BLAST is used to answer a fairly straightforward question: I. In a biological sense, this question is not a simple I or . Results produced by BLAST are a statistical measure of how well two strings match in a biological sense. Even though BLAST was designed to be as efficient as possible, it can seriously strain any computing resource due to the shear size of the data. Clusters have become the platform of choice for BLAST searches.

Here is an interesting thought, What would the bioinformatics community do if there was no cluster market/community today? I suggest that clusters and cluster technology most certainly would have been developed by the bioinformatics community. Surely the necessity for large amounts of high performance computing would have fostered the invention of a commodity solution. How fortuitous then is the rise of cluster computing for bioinformatics.

The biology/cluster synergy continues today. Biologists can now take advantage of existing technologies that are ready and waiting for their problem. Indeed, there is the BioBrew package that provides a turnkey bioinformatics software distribution based on the popular ROCKS cluster distribution. The BioBrew package is freely available and is hosted at the bioinformatics.org site where you will find other interesting projects like Ghemical. As the name implies, Ghemical, is an open source application for studying the quantum and molecular mechanics of both small and large molecules. It also has a GUI interface for viewing results. In addition, if fast BLAST processing is what you need, then you may want to check out a parallel version of BLAST called mpiBL! AST.

So what is so special about these projects? First, they are all freely available. Like much of the software needed for clusters, the user if able to freely modify and possibly re-distribute this type of software. Second, because modifying software to fit your needs is critically to your success, open software is a necessity. (One could certainly argue that science should always use open software, so that there are no "black boxes" involved.) The end result, is that you control your destiny and can help achieve a common goal. I would be willing to guess that, like open commodity clusters, had the there been no open application platform, the bioinformatics community would have developed their own. Sharing is just too powerful a mechanism. The adage "share a little get a lot" is true here as well.

And, there is money is made by supporting, configuring, designing, and modifying all these open applications and clusters. Because assembling the pieces correctly is just as important as have the pieces in the first place.

Douglas Eadline can be found swing around the binary trees at Cluster Monkey

This work licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 2.5 License