Sameh El-Difrawy

Basecalling using hidden Markov models

Abstract In this paper we propose hidden Markov models to model electropherograms from DNA sequencing equipment and perform basecalling. We model the state emission densities using artificial neural networks, and modify the Baum–Welch reestimation procedure to perform training. Moreover, we develop a method that exploits consensus sequences to label training data, thus minimizing the need for hand labeling. We propose the same method for locating an electropherogram in a longer DNA sequence. We also perform a careful study of the basecalling errors and propose alternative HMM topologies that might further improve performance. Our results demonstrate the potential of these models. Based on these results, we conclude by suggesting further research directions.

High throughput system for DNA sequencing

A 768-lane DNA sequencing system based on micromachined plates has been designed as a near-term successor to 96-lane capillary arrays. Electrophoretic separations are implemented in large-format (25cm by 50cm) microfabricated devices with the objective of proving realistic read length, parallelism, and the scaled sample requirements for long-read de novo sequencing. Two 384-lane plates are alternatively cycled between electrophoresis and regeneration via a robotic pipettor and switching optical system. The instrument minimizes the DNA sample requirement to “1∕32×” Sanger chemistry, equal to typical genome center operation, and a 16-fold improvement in scaling relative to previous microfabricated devices. The 40-cm-long channels permit an increase in read length (several hundred base pairs) relative to previous multichannel microfabricated devices. These advances directly address the cost and automation model for adaptation of the technology.

MEMS-based systems for DNA sequencing and forensics

We outline some of the most important scientific problems that were overcome in this multiyear project. These include (a) the high-yield microfabrication of zero-defect bonded glass plates at sizes greatly exceeding semiconductor fabrication, (b) the integration of electrodes, reservoirs, and high-pressure seals into these plates, (c) detector optimization for the required very high rate data acquisition and (d) microfluidics issues and optimizations related to channel geometry.

Uniform acquisition for high-throughput DNA sequencing

The demand for higher throughput and lower cost DNA sequencing did not dissipate with the successful completion of the first human genome but, instead, is actually accelerating in a period of functional genomics. We have developed a high-throughput DNA sequencing instrument utilizing MEMS technology to sequence DNA samples in 384 separate channels in parallel. The performance of the system relies on the quality and uniformity of the data acquisition system. In this work we describe the different constraints of the system and our solution that balances the conflicting requirements of a high data throughput, low noise and uniform performance across all channels.

BIOMEMS-768 DNA Sequencer

The Whitehead Institute has developed an automated DNA sequencer that will go into final Genome Center testing during the summer of 2001. The system comprises a total of 768 separation channels distributed over two plates. The working elements are 50-cm × 25-cm, 384-lane microfabricated glass elements, which undergo alternating electrophoresis and regeneration under use with an exchangeable sieving matrix. The microfluidic and sample transfer devices needed to service the plates are integrated into the same compact instrument. Concurrently with the instrument development we have developed all the protocols, materials and surface preparations to achieve matrix-limited separations of single-stranded DNA in the microfabricated plate format.