Sequence or Die: Automated Instrumentation for the Genome Era

Date: April 12, 1999Table of DNA Sequencers LI-COR's IR2 Automated DNA Sequencer The first time I ran a sequencing gel using 35S rather than 32P, I was in heaven. The bands on the autoradiogram were incredibly sharp, the background was amazingly clean, and my read length increased by close to 20% to just over 100 bases. Things just couldn't get any better than that! In 1994, The Scientist reviewed the state of DNA sequencing technology and made some predictions about what might be coming next.1 Technologies based on acrylamide slab gels were the only commercially available options. The Applied Biosystems (now the Applied Biosystems Division of Perkin Elmer, commonly called ABI or ABD) model 373 four-color fluorescent sequencer was firmly entrenched as the market leader. Pharmacia's ALF single-color fluorescent sequencer was also available, as was an early single-color machine from LI-COR that operated in the infrared. It was suggested that some novel technologies--capillary gel electrophoresis, sequencing by hybridization, and time-of-flight mass spectrometry--might be making an impact by the end of the decade. Today, while time-of-flight and hybridization approaches are improving,2,3,4 they are not readily used for de novo sequencing, and capillary machines are changing the way we think about genome-level and mega sequencing projects. Traditional approaches to DNA sequencing involved using Sanger's method (the dideoxy or chain-termination technique) to generate the sequence ladders, which were labeled with radioactivity and separated using acrylamide gel electrophoresis. The gels were dried and then imaged using autoradiography. This approach had several drawbacks. The resolution was determined by the nature of the radiolabel used, often leading to fuzzy bands and high backgrounds, and the amount of data was limited by the format. That is, the gel image that the researcher used to "read" the DNA sequence represented a single "time slice" of the electrophoresis--some bands had already migrated off the bottom of the gel while others were too tightly clustered at the top. Modifications to the traditional approaches generally focused on eliminating radioactivity from the process. Similar methodologies used alternative labels to detect the DNA fragments; these included silver staining, linked-enzymatic approaches, chemiluminescence, and fluorescence. While these techniques went a long way toward removing hazards from the laboratory, because they continued to gather information from a single time slice of the gel, they did not significantly improve the amount of information that could be obtained from one sequencing experiment. However, by incorporating a fluorescent label in the newly synthesized DNA strand and by placing an excitation light source and a fluorescence emission detector next to the gel, fluorescence can be monitored as the gel is running--automated fluorescent DNA sequencing. A result of this is that data can be collected for the entire duration of the electrophoresis run, not for just a single time slice. So while it was possible to read a couple of hundred bases from the image of a traditional gel (a gel that was probably run for a couple of hours), measuring the fluorescence present in the DNA bands as they migrate past the detector allows the researcher to run the gel longer. Bands that would normally have left the gel after 2 hours are detected as they pass the detector, as are bands that would be clustered at the top of the gel and unreadable on a conventional image. Suddenly it was possible to run a sequencing gel for 12 to 16 hours, and read lengths of several hundred bases became achievable. Rather than obtaining a photographic image of one time slice of the gel, the researcher receives a trace that depicts the time-dependent amount of fluorescence that passed the detector during the entire run. This "electropherogram" also can be automatically read by the sequencing machine. While readers are available for conventional autoradiograms, having an automated fluorescent sequencer decode the nucleotide sequence after the electrophoresis is complete eliminates one more human intervention from the sequencing process. Automated fluorescent sequencing, however, has its prices. Most notably, it requires extremely rigorous cleaning of the gel plates, much more care in the preparation of the gel reagents and then in the casting of the gel than had been previously necessary, and cleaner template DNA preparations. Molecular Dynamics' MegaBACE 1000 Before we review some of the more popular machines that are commercially available today, it is worth looking a little deeper at how the four-color automated fluorescent sequencing process works. (Many of these principles also apply to single- and two-color fluorescent machines.) Most modern protocols continue to use chain termination chemistry to produce the sequence ladders. The quantity of template required has been significantly reduced through the introduction of cycle sequencing approaches. (Similar to single-primer PCR, the DNA template is used for multiple rounds of new chain synthesis.) Different fluorescent labels can be incorporated into the primers, in which case four separate reactions must be performed for each of the four nucleotide-specific terminations, or labels can be incorporated into dideoxy terminators, in which case all four termination reactions can be performed in the same tube. Acrylamide gels are loaded in much the same way as conventional sequencing gels, and the data are collected during the run. While with single-color fluorescent sequencing, each nucleotide-specific sequence ladder must be loaded in its own lane, with four-color sequencing, the four sequence ladders can be loaded in the same lane. One practical benefit of this is that four times as many sequences can be loaded onto the gel. The four dyes that we used for early fluorescent sequencing, called FAM, JOE, TAMRA, and ROX, as well as many modern derivatives of these fluorophores, have some problems. First, the excitation maximums for these dyes are at different wavelengths, making it difficult to obtain effective excitation of all four dyes with a simple light source. (Excitation and emission wavelengths are as follows: FAM: ~ 493 nm, 520 nm; JOE: 528 nm, 554 nm; TAMRA: 560 nm, 583 nm; ROX: 587 nm, 607nm.5 ) The excitation problem has been largely solved with the introduction of energy transfer dyes6,7 such as Amersham's DYEnamic system and ABI's BigDyes. Second, the emission spectra of the four dyes overlap considerably, necessitating careful optical filtering techniques to resolve one dye's signal from the next. Even with good optical filtering, there is still considerable crosstalk between the optical channels. The third problem involves mobility of the labeled DNA strands in the gel.8 As the four dyes have different chemical compositions, they influence the migration of DNA fragments in the gel in different ways. These differences in migration are more pronounced with smaller DNA fragments. The ABI 373 machine excited all four dyes with two lines from an argon-ion laser placed "behind" the gel. Fluorescence emission was collected by a detector located adjacent to the laser. A rotating filter wheel helped reduce crosstalk between the fluorescence from each of the four dyes. The laser and detector assembly scanned back and forth across the gel, and the labeled DNA fragments migrated past the scan line. Data collected in this way consist of four numbers (corresponding to the fluorescence signal collected through each of the four filters), for each of a predetermined number of points across the gel, collected at regular time intervals throughout the electrophoresis. When the run ends, these data have to be processed. The electropherogram that you are probably used to seeing from automated four-color sequencers is in fact highly processed data. The raw data (which are almost impossible to read by eye) are first processed using a matrix inversion. The matrix is a reference of how much each of the four colors "bleeds" into each of the four filtered channels. The matrix inversion allows this bleedover to be corrected, leading to reasonably well-defined peaks and an acceptably flat background. After matrix inversion, the different mobilities, caused by the different fluorophores, have to be eliminated. Mobility shifting works by correcting the observed mobilities in a region-by-region fashion. After the matrix inversion and mobility correction, the data can be basecalled--nucleotides assigned to each peak in the electropherogram. In general, this is the stage where the data are printed as the trace you are probably familiar with. PhorTech International (a market research company) put a figure of $170 million on the 1997 DNA sequencing market, with $145 million of that coming from automated sequencers. As the $3 billion Human Genome Project really gets up to speed, and as a variety of other genome projects and diagnostics applications accelerate, this figure is presumably growing. In today's competitive market, there are essentially four classes of automated machines that can be used for a variety of fragment analyses as well as sequencing: (1) single- and two-color gel-based instruments, (2) four-color gel-based instruments, and (3) single- and multiple-capillary instruments. (As migration from capillary to capillary can be quite different, all capillary machines currently require four distinctly labeled sequence ladders.) We'll look at some representatives from these categories; vital statistics such as bench footprints, ballpark price, and general performance statistics can be found in the table. Capillary arrays for the MegaBACE 1000 DNA sequencing system from Molecular Dynamics (AP Biotech). The capillaries are supplied in preassembled arrays of 16 capillaries each, designed for easy replacement and long life. Image provided by AP Biotech. Three main products are included in this single-color automated sequencer class, which is dominated by small "personal" instruments. One is from Visible Genetics and two are from Amersham Pharmacia Biotech (AP Biotech; is it just me, or is merger fever in the biotechnology world leaving everyone confused about who is who these days?). Many machines in this review employ scanned excitation/detector systems where the excitation source and the detector move from side to side across the gel. The time taken to completely traverse the gel (the duty cycle) introduces an interesting trade-off that has been necessary in many systems: It is necessary to interrogate each point across the gel for sufficient time to obtain an adequate signal (the signal-to-noise ratio improves with the square of the interrogation time), but it is also necessary to return to each point often enough not to miss an event (such as a labeled DNA band passing the scan line). With most slab gels, DNA bands are migrating sufficiently slowly that missing a band is not likely, but as companies find ways to run gels (or capillaries) faster or to cram more lanes on the gel, this point is worth keeping in mind. Visible Genetics makes a one-color "personal" sequencer as part of its Open Gene series: the MicroGene Blaster. Disposable MicroCel cassettes (16 lanes) can be quickly filled with photopolymerizable acrylamide from a SureFill cartridge using Visible Genetics' SureFill Gun, and the gel can be rapidly polymerized in a few minutes in the Gel Toaster. Four single-color sequences (or 16 fragment analyses) labeled with Cy5.5 can be loaded, and sequence can be read to 300 bases in under 30 minutes. The excitation and detector system employs arrays of semiconductor lasers and photodiodes. While the MicroGene Blaster and the MicroGene Clipper (described in the next section) have not been explicitly pitched as de novo sequencing machines (Visible Genetics' focus is on diagnostics), these machines are suitable for resequencing relatively short, diagnostically important stretches of DNA. PC-controlled machine uses a red HeNe laser (633 nm) for excitation and is optimized for Cy5 dye. Moving parts are eliminated by directing the laser beam across the entire width of the gel (running between the plates), exciting all lanes simultaneously. The 40-lane gels, which accommodate 10 sequences, are cast from photoactivated Repro-gel mix and polymerized in AP Biotech's Repro-set apparatus. AP Biotech's new SEQ4x4 machine is the ALF's baby sibling. Sixteen lanes, each excited by its own diode laser and interrogated by its own photodiode, allow for four sequences to be run (hence 4x4). LI-COR's main sequencing platform is its IR2 machine, a sequencing platform that deserves more attention. LI-COR focused on two factors--sequence quality and customer service--and it has secured the reputation for being among the best in the industry in both areas. LI-COR's machine operates in the infrared. The pro is that biomolecules, gels, and gel plates have very little inherent infrared fluorescence, while the con is that the infrared dyes are less commonly used and less understood than the visible dyes used in most other systems. With a little planning, however, the pro easily outweighs the con! LI-COR took the inherently low background of infrared dyes and designed a robust excitation and detection system. A solid-state infrared laser (similar to those in many compact disc players) and a solid-state avalanche photodiode are mounted next to each other and focused on the gel using microscope objectives. These components are duplicated at a second wavelength (spectrally separated from the first by 100 nm), and the entire optical system is mounted on a scanning translation stage. The gel is interrogated at 768 points across its width (allowing band morphologies to be accurately recorded), and data are recorded at 16 bits (65,000 gray scales per pixel, allowing an enormous dynamic range in the intensity of the detected bands). Taken together, this system allows the longest read lengths in the industry: 1,000 bases per run is considered "average," and 1,400 base reads are not uncommon. The machine comes with a wide variety of options and allows different scan speeds for different applications; the most common installation uses 64-lane gels to read 16 sequences. But there's more. That second excitation/detection system scans back and forth alongside the first. While this does allow two of the four reactions to be run in a single lane, LI-COR has developed a technique it calls SBS: simultaneous bidirectional sequencing. For example, a forward primer labeled with the 700 nm dye and a reverse primer labeled with the 800 nm dye can be simultaneously added to each of the four base-specific reactions and a 2 kb fragment sequenced in its entirety from both ends. While the IR2 is not a machine designed for high-throughput sequencing of thousands of templates per day, 32 kb per gel, in big chunks, is well worth a second look. Visible Genetics also produces two-color machines. The MicroGene Clipper is a two-color version of the MicroGene Blaster. The use of two dyes (Cy5 and Cy5.5) allows for eight sequencing samples to be run on each gel or allows for four samples to be sequenced from both ends. The MicroGene Clipper also uses a separate laser diode for each lane; two colors are accommodated by two photodiodes for each lane. The read lengths are very short compared to the IR2, but the Clipper is fast: 400 bases at 98.5% accuracy in 30 minutes. The MicroGene Tower is the Clipper's taller twin. Almost identical to the Clipper, the Tower allows for different gel lengths, the longest optimally producing read lengths of 1,000 bases in about 4 hours. An SBS reaction generated the 1,817 base sequence shown above. Chromatograms for both forward (top) and reverse (bottom) sequences are shown at bases 897-926 and bases 892 -921, respectively. Image provided by LI-COR. For four-color sequencing in a slab-gel format, the ABI PRISM 377 DNA Sequencer is an unquestionably impressive machine. Built around the ABI's PRISM technology, the 377 allows for a bewildering array of options. So what is ABI PRISM? The 373, as described above, used a filter wheel to resolve the different fluorescence emissions. One downside of this approach was that if you wanted to look at dyes that had different emission profiles, the filter wheel had to be replaced. ABI PRISM overcomes this limitation. Excitation in the 377 comes from an argon-ion laser, with all except the 488 and 514 lines eliminated by optical filtering. Using mirrors, the laser beam is directed through the gel, parallel to but behind the gel and its surrounding plates. Some of the instrumentation scans back and forth across the gel. The laser beam is redirected toward the gel by a mirror on the translation stage and focused into the exact plane of the gel by a lens. Fluorescent emission is collected from the laser's focal point by a second lens, scattered laser light is removed by filtration, and the fluorescence is then spectrally resolved by a diffraction grating. This spectrally resolved fluorescence is finally directed onto a CCD camera. The beauty of the ABI PRISM system is that the entire spectrum (514 to 680 nm) is collected on the CCD, and it can then be "virtually filtered" rather than optically filtered. Pixels on the CCD that correspond to the peaks of the four fluorescent dyes can be summed ("binned"), and the other regions can be ignored. Changing filters or fine-tuning the system becomes as simple as reprogramming (in the software) which pixels are kept and which are discarded. As the emission spectrum of a fluorescent dye can subtly change with the temperature, gel buffer, and the nature of the coupling chemistry, and so forth, ABI PRISM allows for rapid optimization of different dyes and conditions. The 377 has the lion's share (about 85%, according to the company) of the automated slab-gel sequencing market (LI-COR's IR2 is the runner-up). This success is guaranteed in part by ABI's dominance of the reagents side of the fluorescent sequencing industry (patent battles over proprietary dye technology notwithstanding), but is also supported by a large number of different configurations that the machine will accommodate. The commonest uses 36-lane gels with 36 cm plates. A typical "fast sequencing" run on this setup should yield 450 bases per lane from a 2.5-hour electrophoresis. Plates are also available in 12 cm and 48 cm sizes: 48 cm plates combined with a "long sequencing" run should yield 800 bases in about 9 hours. The standard configuration can run both 24- and 36-lane gels. The XL upgrade is a software option that increases the lane density to 48 or 64, and more recently (after extensive experimentation with track-etched gel plates), a 96-lane standard slab gel upgrade has become available. For those who want a state-of-the-art four-color sequencer but don't need high throughput, ABI offers a budget 18-lane version. Most of us are familiar with slab gels, but capillaries are a completely different kettle of fish. Several good overviews of capillary sequencing have been published,9,10 and understanding capillaries in more depth is worthwhile because they will probably form the basis for full (fluidic) integration and chip-based analysis.11,12 Emission spectra of DNA sequencing dyes for the ABI Prism 3700 DNA Analyzer. Image provided by Perkin Elmer. Typically between 50 and 100 µm in diameter, capillaries used for DNA sequencing are anywhere from 30 cm to 1 m long glass tubes that are generally rendered somewhat flexible by a polyimide coating. Casting a gel of these dimensions is difficult (but possible), so various nongelling viscous sieving polymers are used to effect DNA separation; in the last few years, polymers that have low viscosity (allowing relatively low-pressure refilling of the capillary) combined with adequate resolving power (allowing separations out to several hundred bases) have become available. The tiny cross-sectional area of a capillary reduces joule heating to negligible amounts, allowing electrophoresis to be conducted as high as 30 kV (compared to 1,500 V for a typical slab gel). Under these conditions, not only is the sample migrating through the gel, but the entire gel is slowly squirming its way through the tube due to electro-osmotic flow. Constructing a fluorescence detector for a round tube is also harder than for a flat glass plate. Luckily, the manufacturers have taken care of these problems (at least to some workable degree). What hasn't been done for you, though, is sample preparation. Capillary samples are loaded electro- kinetically--maybe 10 µl of sample enters the separation matrix due to an applied electric field--so any other molecule that can carry a charge (salt, protein, primer, and so on) will reduce labeled fragment loading and therefore decrease your signal. Capillary sequencing samples require careful preparation and clean-up, depending on the sensitivity of the detection optics in the instrument. To get around this, Perkin Elmer chose to use sheath flow in the 3700, to increase sensitivity while not increasing sample preparation time. Perhaps not a giant leap, but ABI's 1995 introduction of the single-capillary fluorescent DNA sequencer--the ABI PRISM 310 genetic analyzer--showed that capillary methods, already long accepted in other analytical roles, were appropriate for sequence analysis. Another machine in the ABI PRISM series, the 310 had a proven excitation/detection system and allowed researchers a low-impact transition from manual, radioactive sequencing to automated fluorescence-based methods. The 310 combines some elegant features (such as the capillary flushing/refill system that also served as part of the electrical path) with some features that feel a little "clunky" (such as the microtiter dish manipulator). The 310 runs a single capillary in about 2 hours. This cycle includes electrokinetically injecting the sample, performing the electrophoresis, collecting the data, pumping out the spent separation matrix (the polymer), and flushing/refilling the capillary with fresh polymer. The machine accepts samples in 96-well microtiter plates, although unattended operation for 96 samples is not possible due to the volume of polymer that can be stored in the reservoir. However, unattended operation for 48 hours--over the weekend--is possible at 9 or 10 sequencing runs per day. While it is possible to read 600 bases from a 165-minute run using POP-6 (ABI's performance-optimized polymer for sequencing) even with rigorous sample clean-up using a variety of methods including magnetic beads or minicolumns from Princeton Separations, usable read lengths of 300-400 bases are more common. Clearly this is not a production machine, but the ease of unattended operation paved the way for multiple-capillary machines. In addition, the 310 has been adopted by researchers with small budgets; the company reports sales of over 3,000 instruments in three years. Multiple-capillary machines look set to become the genome industry darlings. While read lengths currently don't match those attainable from long slab gels, for huge projects it seems to be more effective to crunch tens of thousands of (shorter read length) samples through the system than to worry about getting the most out of each sample. As capillaries can be automatically refilled without human intervention, unattended operation for thousands of samples is realistic. These machines are likely to remain in specialist labs; just securing the reagent costs alone for a couple of these monsters becomes a daunting proposition, and a small army of graduate students (or at least a battalion of robotic workstations) seems necessary to generate the templates. Molecular Dynamics won the race to introduce the first multiple-capillary sequencer when it unveiled the MegaBACE 1000 DNA sequencing system at the Hilton Head genome meeting in 1997, developed through a joint project between Molecular Dynamics and AP Biotech, who officially joined forces last year. The MegaBACE 1000 uses a scanned confocal detector (in a confocal detector, both the excitation light and the emitted fluorescence pass through the same optical components) to interrogate an array of 96 capillaries, each of which is separating a four-color sequencing ladder. The turnaround time (injection to injection) is under 2 hours, so more than 1,100 samples can be processed by a single machine in 24 hours. (We humans are not yet redundant, though, because as it stands, the MegaBACE 1000 needs an operator to manually exchange sample trays between runs. However, a collaboration between Molecular Dynamics and Beckman Coulter announced last year aims to integrate robotic workstations to feed four sequencing machines simultaneously.) As capillaries tend to foul after a couple of hundred runs, Molecular Dynamics produces 16-capillary arrays to allow modular replacement. Using its long-read sequencing matrix and AP Biotech's sequencing reagents, Molecular Dynamics claims average read lengths of over 500 bases per capillary, and in many labs read lengths are routinely exceeding 600 bases. Half a megabase of raw sequence a day ain't bad ... Other 96 capillary machines are available, for example, from Spectromedix, and researchers have proposed even larger capacity (1,000-capillary) instruments, but ABI's 96-capillary machine, hinted at in 1997, has been eagerly awaited but only recently announced. The 3700 DNA analyzer is the third machine in the ABI PRISM series, so it features similar integration with ABI's sequencing reagents as the 310 and 377 machines. Even though machines have been shipped under ABI's early access program, precise details about the 3700's specifications and performance are unavailable. Clearly it is a 96-capillary machine, and 24-hour unattended operation (at 96 templates per run but using 384-well plates) is indicated in the literature. A rather slower injection-to-injection time of 3.5 to 4 hours was mentioned by one representative, but this seems unduly long, so it may be incorrect. Some noteworthy features of the 3700 are a wider spectral range, enabling the use of a fifth laser, a 96 plus eight capillary system, with eight capillaries in reserve in the event of a fouled capillary. A January press release claimed that third-party orders for 170 units had been received, in addition to the 230 units ordered by Celera Genomics, the company founded by TIGR's Craig Ventor and Perkin Elmer. Assuming the 3700 performs up to company specs (but with a 4-hour turnaround), and assuming 20-fold redundancy to go from shotgun genome to finished sequence, Celera's 230 machines should complete the 60 gigabases of raw sequence needed for a human genome in about 500 fully operational days. Just don't ask me to do the template preps! Beckman Coulter's CEQ 2000 That level of throughput is more than most researchers need or can support. Perhaps an eight-capillary machine is more realistic for many labs? Beckman Coulter has leveraged its expertise and reputation by modifying its time-tested P/ACE capillary system to handle DNA sequencing, giving us the eight-capillary CEQ 2000. The CEQ 2000 features an array of eight 50 cm long, 100 µm i.d. capillaries. The CEQ 2000, similar to the LI-COR IR2, reaps significant gains from operating in the near-infrared. Beckman Coulter has synthesized its own proprietary derivatives of Cy5 and Cy7, resulting in a series of dyes that have well-spaced emission maxima spread between 675 and 815 nm; these dyes are excited by two separate lasers in the CEQ 2000. The low inherent fluorescence of infrared dyes allowed Beckman Coulter to create an unusual detector: All eight capillaries are interrogated simultaneously using a filter wheel and photomultiplier, and the lasers are scanned back and forth across the capillaries (exciting each capillary at about 8 Hz). In the visible region, scattered light and fluorescence would render this scheme inoperable, and even in the infrared, crosstalk between the capillaries might be a problem with strong sequences or overloaded samples. A 2-hour injection-to-injection time allows unattended operation for 24 hours--12 loadings from a 96-well plate. With its fluorescent dye-terminator cycle sequencing kit and premixed gel matrix, Beckman Coulter claims the CEQ 2000 will produce average read lengths of 500 bases: about 50 kb of raw sequence per day from a production pipeline that could realistically be kept full by a small team. So where do you start when the time comes to move away from the local core facility and go it alone for your next sequencing project? Is it worth purchasing your own personal machine or teaming up with other groups and sharing a larger one? Based on the scope of your project, it's easy to figure out roughly how many templates you'll need. From there, determining how many your group can produce on a daily or weekly sustained basis and how much you can afford for reagents will give you an idea of the machine throughput you'll need. In the larger scheme of things, the capital cost of the instrument is a fairly small portion of the total cost of the project, so it shouldn't be the major concern. Most labs that are not already sequencing extensively but want to do so in a fairly serious way will probably be looking at the LI-COR IR2, the ABI PRISM 377 DNA Sequencer, the Beckman Coulter CEQ 2000, or Visible Genetics' MicroGene Tower. Your sequencing strategy, and whether you'll need long read lengths or lots of short runs, will probably be the next consideration. Finally (as reports from actual users often fall short of the marketing department's expectations), if you can, get a demo machine and see how it works with your templates and techniques. In honor of a National Institutes of Health "visit" to the Utah Genome Center, someone had T-shirts printed that read "Sequence or Die." One of the NIH team, during a seminar, elicited some nervous laughter when he quipped that the slogan reflected how the NIH was reviewing grant proposals at the time. But larger-scale sequencing looks like it's here to stay--if you can stand the heat. Table of DNA Sequencers S. Veggeborg, "DNA sequencing: today's technology and beyond," The Scientist, 8[20]:17, Oct. 17, 1994. F. Kirpekar et al., "DNA sequence analysis by MALDI mass spectrometry," Nucleic Acids Research, 26:2554-9, 1998. D.J. Fu et al., "Sequencing exons 5 to 8 of the p53 gene by MALDI-TOF mass spectrometry," Nature Biotechnology, 16:381-4, 1998. S. Parinov et al., "DNA sequencing by hybridization to microchip octa- and decanucleotides extended by stacked pentanucleotides," Nucleic Acids Research, 24:2998-3004, 1996. ABI's description of the dyes used in the PRISM system are found at www.perkin-elmer.com/ab/about/dna/377/377a2c.htm. C.R. Heiner et al., "Sequencing multimegabase-template DNA with BigDye terminator chemistry," Genome Research, 8:557-61, 1998. L.G. Lee et al., "New energy transfer dyes for DNA sequencing," Nucleic Acids Research, 25:2816-22, 1997. O. Tu et al., "The influence of fluorescent dye structure on the electrophoretic mobility of end-labeled DNA," Nucleic Acids Research, 26:2797-802, 1998. N.J. Dovichi, "DNA sequencing by capillary electrophoresis," Electrophoresis, 18:2393-9, 1997. A.E. Karger, "Separation of DNA sequencing fragments using an automated capillary electrophoresis instrument," Electrophoresis, 17:144-51, 1996. H. Swerdlow et al., "Fully automated DNA reaction and analysis in a fluidic capillary instrument," Analytical Chemistry, 69:848-55, 1997. C.S. Effenhauser et al., " Integrated chip-based capillary electrophoresis," Electrophoresis, 18:2203-13, 1997. The author, Bob Sinclair, can be contacted at bobsinclair@tech-write-edit.com.

Sequence or Die: Automated Instrumentation for the Genome Era

Interested in reading more?

Become a Member of