SALT LAKE CITY — The University of California, San Diego’s technology-transfer office is currently in discussions with several undisclosed tool vendors about licensing an “ordered array” technology that university researchers invented and that has applications in high-throughout sequencing, one of the inventors said last week.
Speaking at the Association for Biomolecular Resource Facilities conference here, Xiaohua Huang, an assistant professor of bioengineering at UCSD, said that at least one company has expressed interest in licensing the patent-pending technology for use with its next-generation sequencing platform.
In his talk, Huang said the “ordered array” approach uses a magnet to direct the assembly of DNA particles into a grid-like pattern on a microfluidic chip. He said it could be a promising alternative for a number of high-throughput sequencing platforms that currently use random arrays of DNA molecules.
The approach is not to be confused with sequencing on pre-fabricated arrays, a strategy that Affymetrix and others are pursuing, where the identity as well as the location of each feature on the array are known (see In Sequence 1/15/2008).
Huang said that Applied Biosystems’ SOLiD system, Illumina’s Genome Analyzer, and Danaher Motion/Harvard University’s Polonater all use random arrays, whose drawbacks include low density and low imaging efficiency, and a demand for complex image analysis to recognize the shape, location, and intensity of signals on the chip.
Huang said that ordered arrays, on the other hand, could eliminate much of this difficulty, since the cameras only need to capture information from specific spots on the chip.
Other next-gen sequencing platforms are already betting on ordered arrays. 454 Life Sciences, for example, uses a picotiter plate, essentially an “ordered array” of tiny reaction wells.
“It doesn’t matter how fast your chemistry is. It matters how fast you can read it.”
Intelligent Bio-Systems, a startup that is commercializing sequencing technology from Columbia University, also plans to use ordered arrays in order to minimize image acquisition time. Last fall, IBS CEO Steven Gordon told In Sequence that the company will use a “proprietary method” to generate the arrays (see In Sequence 10/23/2007).
Huang’s approach uses photolithography to create an array of microwells in a layer of photoresist on a coverslip. The array is then placed within a microfluidic device, and DNA molecules that have been conjugated with superparamagnetic microbeads are introduced into the chamber. A magnet directs the assembly of the beads into the wells so that each well contains a single bead.
The system currently requires 9 pixels to analyze a single feature, which Huang said could translate into a throughout of around 35 gigabases a day using one camera. His team is aiming to improve the resolution of the system to one pixel per feature. With that resolution, and with four cameras, he said the system could analyze around 1,250 gigabases per day.
Huang noted that the array technology could work with nearly any sequencing chemistry, including sequencing by ligation, sequencing by synthesis, or single-molecule sequencing. In an interview with In Sequence following his talk, he said that the key limiting factor for most next-gen systems is not the chemistry, but the imaging.
“It doesn’t matter how fast your chemistry is,” he said. “It matters how fast you can read it.”
Nevertheless, he said his lab is using a $750,000 “$1,000 Genome” grant from the National Human Genome Research Institute to develop a novel sequencing chemistry of its own. The NHGRI awarded the grant in 2005 under its Advanced Sequencing Technology program.
The approach Huang’s lab is pursuing, called sequencing by denaturation, is based on standard Sanger dideoxy sequencing, but gradually increases the temperature to denature dideoxy-terminated fragments from the templates one by one. Huang said that because shorter molecules dissociate first, and longer molecules dissociate last, the system can then plot the signal intensity of the molecules versus time in order to identify the bases.
“In theory, we have proven that we can sequence molecules of up to 50 bases in length,” he said. “In practice,” however, the approach is “harder than we thought because the imaging is not trivial.”