Andy Marks, Tony Schurtz, F. Mark Ferguson, Leonard Di Sera, Alvin Kimball, Diane Dunn, Doug Adamson, Peter Cartwright, Robert B. Weiss1 and Raymond F. Gesteland1.
Department of Human Genetics, and the [1]Howard Hughes Medical Institute, University of Utah, Salt Lake City, UT 84112.
Automation of a large-scale sequencing process based on instrumentation for automated DNA hybridization and detection is a focal point of our research. Recently, we have devised a method for amplifying fluorescent light output on nylon membranes by using an alkaline phosphatase-conjugated probe system combined with a fluorogenic alkaline phosphatase substrate [1]. The amplified signal allows sensitive detection of DNA hybrids in the sub-femtomole/band range.
On the basis of this detection chemistry, automated devices for detecting DNA on blotted microporous membranes using enzyme-linked fluorescence, termed Probe Chambers, have been built. The fluorescent signal is collected by a CCD camera operating in a Time Delay and Integration mode. Concentrated solutions of probes and enzymes are stored in Peltier-cooled septa sealed vials and delivered by syringe pumps residing in a gantry style pipetting robot. Fluorescence excitation is generated by a mercury arc lamp acting through a fiber optic "light line". Three 30 x 63 centimeter sequencing membranes can be simultaneously processed, currently revealing up to 108 lane sets per multiplex cycle. A probing cycle is completed approximately every eight hours.
Integration of the Probe Chamber into the production pipe line is accomplished through connections to the laboratory data base. A critical component of a high-throughput sequencing laboratory is the software for interfacing to instrumentation and managing work flow. The Informatics Group of the Utah Genome Center has designed and implemented an innovative system for automating and managing laboratory processes. This software allows the model of workflow to be easily defined. Given such a model, the system allows the user to direct and track the flow of laboratory information. The core of the system is a generic, client-server process management engine that allows users to define new processes without the need for custom programming. Based on these definitions, the software will then route information to the next process, track the progress of each task, perform any automated operations, and provide reports on these processes. To further increase the usefulness of our laboratory information system, we have augmented it with hand-help mobile computing devices (Apple Newtons) that link to the database through RF networking cards.
Base calling software has been developed to support our automated, large scale sequencing effort . 1st stage sequence calling identifies putative bands, however, depending on the number of reader indel errors (2-6%), merging 1st stage sequence without the aide of cutoff information can be difficult. To improve our base calling we have employed Fuzzy Logic to establish confidence metrics. The logic produces a confidence metric for each band using band height, width, uniqueness, shape, and the gaps to adjacent bands. The confidence metric is then used to identify the largest block of highest quality sequence to be merged.
This work was funded, in part, by DOE grant DE-FG03-94ER-61817 (R.F.Gesteland, P.I.)
[1] Cherry, J.L., Young, H., Di Sera, L.J., Ferguson, F.M., Kimball, A.W., Dunn, D.M., Gesteland, R.F., and Weiss, R.B. (1994). Enzyme-linked fluorescent detection for automated multiplex DNA sequencing. Genomics 20, 68-74