Learn about programs/initiatives happening across our campus through these @HMS spotlight stories!
Backing Up Science on the Quad
Rainer Fuchs and IT team help HMS to get a new "backbone"
By ANGELA ALBERTI
Surrounded by Stone Age tools lies a tooth that belonged to an ancient human ancestor. It contains just enough genetic material to extract and study, possibly providing new clues about how humans evolved. A team of geneticists will use specialized computers to sequence the prehistoric DNA and statistical algorithms to compare genotypes across present-day populations. They’ll share the result data directly with collaborators across the globe.
A cell biologist pipettes a mixture of proteins into a thin translucent tube, one of which is given a special biological “tag” allowing all the other proteins interacting with it to be identified. This single protein may be related to Amyotrophic Lateral Sclerosis (ALS), but there are thousands of proteins networking in complex ways that may hold the cause of, and possibly cures for, such debilitating diseases. Data about each complex protein network must be collected, stored, sorted and analyzed to find these clues.
A community health worker treating patients in a remote area uses his mobile phone to take photos of a patient’s post-operative wound and sends it to a Boston physician. The physician, who monitors the wounds to ensure it is infection-free, is also a social medicine professor, and receives dozens of similar photos from other health workers. She adds them to an online presentation, which will be used to teach new care givers across the globe to become experts at assessing the wound healing process.
These are examples of just some of the work conducted at Harvard Medical School, and although the research differs vastly, each scientific endeavor has one foundational element in common: data.
Modern scientific discovery relies on the ability to create, analyze, store and share massive amounts of data. Supporting science and education through technology is what inspires Rainer Fuchs, chief information officer at Harvard Medical School, and his team.
Data Doesn’t Sit Still
“Having the chance to talk to researchers about the science they do not only satisfies my inner science geek; it also gives me glimpses of what's coming down the pike,” Fuchs says. Trying to see the future is an everyday job for him in leading Harvard Medical School’s Information Technology department. By extrapolating current trends and anticipating future needs, he and his team have redesigned the entire IT infrastructure over the past two years, with the goal of ensuring HMS’s position as a preeminent leader in scientific discovery, biomedical education and clinical care.
“Data doesn’t like to sit still; it wants to move around,” Fuchs says. “It needs to be shared among local teams and with external collaborators all over the globe.” Scientists at HMS, for example, transfer data from instruments where it is generated to the School’s high performance computing environments or specialized analysis cores; or even to remote resources, such as supercomputer centers or the Amazon Cloud.
Over the last year, Fuchs and his team have created a new 100 gigabit network “backbone” that connects the School’s data centers and the Internet. Now, over the next few months, all buildings on the HMS campus will be connected to the new backbone, and each HMS department will be required to briefly stop in its tracks and shut down all computers and networks while new equipment is installed and then reconnected. Fuchs says his team is completing most of the work on weekends or overnight in order to minimize the workday impacts.
What will the changes mean for end users? The new network will allow researchers to move massive amounts of information collected from lab instruments and other sources to computers or other tools for analysis, or to send to collaborators much more efficiently than before. Even less data-intensive desktop computing will be speedier and more reliable, due to improvements in the wireless building infrastructure.
“Think about data as cars traveling along a communication highway; we have ripped out country roads and replaced them with the Autobahn,” says Fuchs. But “widening the road” is only one half of the IT systems equation.
The Data Deluge
Imagine a room stacked to the ceiling with different sized boxes. When you try to add a new box to the room, you might be able to shift things around in order to squeeze an additional box inside, but then there’s no room to move. Fuchs was faced with finding room for dozens of virtual “boxes” of research and other kinds of data.
Having received his in PhD in biochemistry in 1989, Fuchs has seen the astronomical increase in research computing data from first hand. “I spent three years cloning and sequencing a gene for my PhD,” he says. “Today, you can do that in an afternoon. Over the last decade, technologies have become available that let us produce sequence data at a speed and volume previously unknown – and cheaply.”
But it’s not just DNA and RNA sequences. New technologies have flooded every aspect of life science, including basic science research and education. Cryo-electron microscopy, lattice light sheet technology, and other new research tools are generating data at a pace that often outgrows the technical infrastructure required to manage and analyze it. On the education front, shifts to electronic technologies and online learning are now the norm, and the HMS Scholars in Medicine Program requires students to embark on research of their own, generating additional streams of data.
According to Fuchs, both average users and those who conduct “Big Data” experiments at HMS are doubling the amount of data they generate each year. Between 2012 and 2016, the amount of data kept on HMS systems has grown from 2 to 30 petabytes. (A petabyte is 1 000 000 000 000 000 bytes, or a hundred times the printed collection of the U.S. Library of Congress in 2012).
This data revolution combined with expensive and limited storage options required Fuchs and his team to dramatically rethink how HMS managed its data. The IT team has consolidated data centers and introduced new data storage and archiving options, which has lowered storage costs per unit by more than 50 percent. Yet, the fact that the rate of data production continues to outpace drops in storage costs is perhaps the most important challenge for Fuchs, not only for his team but for the School overall.
Investing in Science
“I think we were in danger of really losing the ability to do some of the best science in the world,” says Fuchs, “because it relies so much on our ability to deal with very large data. If you're comparing thousands of genomes to understand how mankind evolved, or if you’re studying thousands of cryo electron microscopy images to determine the structure of a fickle protein, you can't do that on a little server under your desk.”
He hopes that the IT system redesign and updated network will not only be adequate to support the School’s current work, but will better position HMS for future endeavors and keep it a magnet for top talent in biomedical research. “Upgrading our infrastructure is probably the single-biggest investment the School has made into supporting our researchers and education business, short of new buildings,” he says.
How will HMS continue to keep up with the onslaught of new data-generating technologies? Ultimately, Fuchs says, HMS must bend the curve of cost that is required to support its research efforts and this is just the beginning. He is confident that as new data creation tools emerge, so too will new networking and storage solutions. He envisions the future of computing as more and more cloud-based and the School’s increased network bandwidth will enable faster transfers of data into and out of the cloud.
Fuchs is always focused on the School’s mission and understanding how business processes and deploying new technology will impact end users. On one visit to a lab, a researcher shared videos with Fuchs documenting the aging of flat worms. “In my head I was doing the math of how much data was being created and what would happen if ten more people did the same,” he says. “This is rooted in learning about the mission, about the science. When I meet with faculty, of course I'm happy to talk about IT problems, but what I really want to learn about is the science we do here. At the end of the day, that’s what it’s all about.”
Editor: Amy Magin
Assistant Editor: Angela Alberti
Contributors: Shannon Patterson, Chelsea Keyes, Linda Miklas, Tania Rodriguez, Susan Vomacka