Since 1986 - Covering the Fastest Computers in the World and the People Who Run Them

Language Flags
March 13, 2012

CycleCloud BigScience Challenge Boosts Stem Cell Research

Tiffany Trader

Cycle Computing has proclaimed the winner of the 2011 CycleCloud BigScience Challenge. Victor Ruotti, a computational biologist at the Morgridge Institute for Research, will receive $10,000 in credit from Cycle Computing and four hours of CycleCloud engineering support, plus an additional $2,500 in credit from Amazon Web Services. The award will be used for cutting-edge stem cell research.

The Challenge, which was revealed in detail at the SC11 conference, was open to non-profit researchers who could harness the power of utility supercomputing to answer big science questions that have the potential to offer real benefits to humanity. The results are being announced after a careful evaluation of the five finalists. HPC in the Cloud spoke with Cycle Computing CEO Jason Stowe and the winning finalist, Victor Ruotti, to learn more.

Located on the campus of the University of Wisconsin-Madison, the Morgridge Institute for Research is a private, not-for-profit interdisciplinary biomedical research organization that seeks to accelerate the movement of science from the laboratory to the clinic. Ruotti works in the Thomson Laboratory, run by stem cell pioneer James A. Thomson. Thomson was part of a team that first transformed adult cells into stem cells called iPS cells in 2007. This was a huge breakthrough and has had a significant impact on science and medicine in the years since.

Ruotti’s research group is working on developing a knowledge base indexing system for human embryonic stem cells and their derivatives. The science is based on a fascinating regenerative process called dedifferentiation, which allows the researchers to take an adult cell and turn it into a human embryonic cell, and then further transform that into different cell types.

“You start with a cell and treat it with a certain differentiation factor and these cells which are human embryonic stem cells turn into a particular cell. This is a very complicated process because sometimes we don’t know what cell type they are turning into,” says Ruotti.

He explains this requires RNA sequencing to find more information based on genetic markers and morphology using 3-dimensional pictures. But still it’s difficult to tell what cells they are turning into. After performing over 1,000 different RNA sequences, Ruotti came up with the idea of creating a sort of dictionary to assist in the identification of cell types. This knowledge base indexing system will provide a percent probability that a certain cell is neural, or cardiac, or smooth muscle, or any other cell. The work they are doing now is laying the foundation for their ultimate goal, which is enabling advances in real-world regenerative biology.

Stowe chimes in: “The thing that got us excited about Victor’s work is the huge potential of the knowledge base that he’s putting together. It says if I start out with an undifferentiated cell and want it to end up in a particular direction, here are the probabilities for that to happen. But the primary blocker here in terms of doing the analysis is raw compute horse power. Taking advantage of a really large numbers of compute hours, a quarter million computer hours should really benefit his research.”

When Ruotti initially went to lab founder James Thomson with a detailed explanation of the knowledge base proposal, he was met with raised eyebrows: “You can do that?” Thomson asked?

“We can if we get this number of compute nodes,” replied Ruotti.

“Oh, great! Then do that,” Ruotti recalls Thomson telling him.

The basis of their research is identifying the differentiated cells, but to do this, the team must first perform a series of very computationally-intensive analyses. The science was hinging on the computational power. This is exactly the kind of project Cycle CEO Jason Stowe had in mind when he formulated the BigScience Challenge.

“There are a huge number of potential clinical applications for helping people build treatments based on differentiated cells. It’s a great fit, answering the big questions that couldn’t be answered without utility supercomputing,” says Stowe.

In addition to the grand prize winner, the contest judges selected a final runner-up, Alan Aspuru-Guzik, from the Harvard Clean Energy Project, for his material science analysis aimed at creating more efficient photo-voltaic cells.

All finalists were awarded both an initial $500 credit from Cycle Computing and an additional $1,000 credit from Amazon Web Services (AWS). Aspuru-Guzik, as the runner-up, will also receive access to some of the idle capacity that Cycle generates as part of executing and building its software.

The top projects were selected based on their creativity, benefit to society and on the appropriateness of a running their workloads on Cycle clusters in the AWS cloud. In addition to the top two choices, there were three other finalists in the pool: Jesus Izaguirre from the University of Notre Dame (diabetes research); Soumya Ray from Harvard Medical School (Parkinson’s research); and Martin Steinegger from TU Munich ROSTLAB (mapping genomic diversity). Tasked with having to sort through all these worthy candidates were judges Jason Stowe, CEO, Cycle Computing; Kevin Davies, editor-in-chief, Bio-IT World; Matt Wood, technology evangelist for Amazon Web Services; and Peter S. Shenkin, vice president, Schrödinger.

The next step, according to Stowe, will be to connect Ruotti with Cycle engineers to give them a better idea of the specific workloads and the technical requirements. Then it will be up to Ruotti and his team from an execution standpoint. The other finalists will also be given the chance to advance their research with the awards they received, and HPC in the Cloud will be sure to report on future findings.