Programmable devices, smaller than the size of a man’s arm, may soon be replacing multiple and costly computer systems normally found in large medical research labs.
Researchers at Mount Sinai Hospital’s Samuel Lunenfeld Research Institute and University of Toronto’s Department of Electrical and Computer Engineering have jointly developed a hardware-based computing system that would fit data analysis information and solutions on a single computer.
This feat will be accomplished through the use of digital chips called field-programmable gate arrays (FPGA).
The goal, according to Dr. Christopher Hogue, principal investigator with The Blueprint Initiative, a research program at Mount Sinai, is to enable a complete analysis of the human genome in less than one second. The FPGA will eliminate the need for space-consuming processor clusters and generate significantly faster results.
FPGAs are devices that can be programmed to perform logical functions such as data analysis and comparison. Users create the file according to the required function through a software system provided with the FPGA system. The binary file created is then downloaded into the FPGA board.
In a statement, Dr. Hogue explained that the FPGA design his team is working on uses several transistors organized into logical units called lookup tables (LUT) that are 20 times more than other FPGA solutions currently available in bio-informatics.
“Ours addresses a real everyday problem and provides a solid, inexpensive platform for a useful implementation,” he said.
In its initial study, Hogue’s research team tested the FPGA-based system by analyzing data generated from mass spectrometry experiments, a method used in scientific and clinical drug development and medical diagnostics studies. The process was completed and desired results achieved in 1.6 seconds, with a cost of about $7000.
The same procedure, using the traditional 64-bit processor cluster would cost $80,000 and a staff to run it. The high cost of the current cluster-based system is also associated with the need for continuous cluster acquisition and maintenance as the demand and requirements increase.
“It is conceivable that FPGA accelerator boards can be as cheap, ubiquitous and high-performing as high-end video graphics boards,” Dr. Hogue says.
Randall Willis, communications manager for The Blueprint Initiative, said while research currently focuses on using FPGA for mass spectrometry analysis, it can also be used to process other types of data.
As the project progresses, the FPGA-based system may be used in other areas of biology research. Several research centres and hardware suppliers have already expressed interest in this initiative, Willis said.