Microarray Centre

Optimal Cluster Solution

For calculations and development work, the UHN Microarray Facility uses this Apple Workgroup Cluster for Bioinformatics, which includes one server node, two cluster nodes, an Xserve RAID system, and a backup power supply. Atop the biocluster is a web server, a database server, and an Xserve RAID system for running the Microarray Facility’s public database portal and a laboratory content management system.

“Greatest Value Per Dollar”

Because Virtanen and his staff prefer to focus on science and not on configuring computers, he chose an IT solution that helped his department hit the ground running: the Apple Workgroup Cluster. The Apple Workgroup Cluster components are pre-qualified and integrated; no in-depth technical expertise is required to maximize the performance of the system. This nearly turnkey, high-performance Apple Xserve cluster ships with more than 200 tools that are included in the iNquiry package, including such essential bioinformatics applications as BLAST, EMBOSS, and HMMer.

“We chose Apple equipment because it offered the greatest value per dollar in terms of computing power, storage, and simplicity of administration and setup,” says Virtanen. “The hardest thing was physically figuring out how the rack went together,” he laughs. “’Do we have this upside right or right side down, and which way does this little screw go in here?’ That was the only hurdle we ever faced.

“We use so much software,” he adds. “All the usual open-source suspects are there, such as mySQL, Apache web server, Tomcat, PERL and BIOPERL. We run a Plone server to keep track of all the projects that are on the go. We have hundreds of small bioinformatics programs installed along with associated databases. The BioTools toolkit is great for organizing some of these programs. Its biggest advantage is that you can keep track of results that you ran at different times, so it can organize findings even when you sometimes run dozens of searches an hour.”

Minimizing Administration Headaches

For storing the immense amounts of data generated by the UHN Microarray Facility, Virtanen says he needed a solution that was expandable, efficient, and mirrored — and didn’t require a lot of management on the back end. “The other bioinformatician and I come from a UNIX background so we have no problems administering networks and handling server administration,” he says. “But we really didn’t want to spend all that time — we just wanted to get down to actually writing software.”

“With the remote GUI-based administration in Apple Remote Desktop, the basics have become easy and the advanced is accessible.”

Virtanen uses Apple Remote Desktop to remotely administer the cluster, which is housed in a climate-controlled room outside of his general office area. “We use all of those utilities to configure the machine and to check logs and see if there are any error reports and so forth,” he says. “With the remote GUI-based administration in Apple Remote Desktop, the basics have become easy and the advanced is accessible.”

Getting Down to the Science

After installing the Apple Workgroup Cluster, Virtanen found they didn’t need to hire a full-time UNIX systems administrator, so they were able to relocate by hiring more scientifically minded personnel, thereby improving the overall productivity of the department. And as a veteran of many years in the bioinformatics trenches, Virtanen can appreciate the differences Apple technology is making on his daily routine. “Years ago, those of us doing what is now called bioinformatics spent a lot of time configuring things like server-side scripts and reading through dry technical manuals just to get things running to the point where everything was going smoothly and we could actually start analyzing data,” he recalls. “I remember many a late-night running into morning poring over literature to find out how to get something done, like manually configuring a firewall.”

Looking to the future, Carl sees value in harnessing the power of Apple’s Xgrid computing architecture to create their own in-house supercomputer. “One of the things we are looking forward to is outfitting all of our technicians’ desktops with Macs in the near future, which will allow us to grab compute cycles using Xgrid when systems are not in use,” he says.

“I always liked the way Macs could offer a more efficient solution for those of us who don’t want to spend as much time with the nitty-gritty of the hardware and the infrastructure underlying the OS itself. Now we have time to get down to the science.”

< Previous Page 1 2