Image Sensors for Massively Parallel DNA Sequencing

December 16, 2014, IEDM, San Francisco—Anette Grot from Pacific BioSciences defined the requirements and possible roadmap for the image sensors that will be used in high-throughput, massively-parallel DNA sequencing. The technology trends and improvements in capabilities related to Moore’s Law have enabled tremendous increases in DNA processing.

First, a background on gene sequencing. A sample of a gene segment or a protein will have from 100 to 1,000 base pairs. DNA sequencing requires first slitting the strand of DNA, synthesizing a complementary set of pairs, and recombining the two sets. Current technologies can take about 100 samples in parallel of 800 base pairs. The analysis requires a reference sequence and a long segment will take many reads to confirm the structure. Faster and more highly parallel analysis will enable the sequences to be concatenated into a “movie” of the and provide analysis of traces of n X m separate segments of length f.

Pyro sequencing is a technique that immobilizes DNA strands on a bead. This technique allows the introduction of a single base at a time, followed by a rinse step to clear away unused molecules. The analysis uses fluoresce as a function of the number of bases that attached at that step. The imaging requirements are for a high pixel count, ;pw dark current, low read noise, and high resolution.

A variant is to use reverse terminators which use different fluorescent tags per base and can be read at any time. In this application, the imager needs to have good color separation to differentiate the different colors in the DNA sample. One way to improve this analysis is to have a line scan per color to minimize registration issues.

The technique is capable of single molecule detection with no additional chemistries needed and capable of 100 nm resolution. The measurement of incorporation time takes lace at a 100 frames per second rate. A long read may incorporate 20,000 bases and the actual image requires the processing of about 100 electrons at a time.

One incorporation of this technology uses 4 cameras for the imaging. All of the cameras need to have very low signal-noise ratios and high quantum efficiency. The control and timing electronics need to have extremely good pulse width and jitter specifications, since these parameters can affect the detected incorporation rates.

The advances in semiconductor technologies enable greater speed and precision in the detectors, while the advances in processors and computer architectures permits greater parallelism in the signal processing. Both of these trends are speeding up and reducing the costs of DNA sequencing.