DNA sequencing technology has evolved very rapidly since its inception in the 1970s, and continues to evolve and grow today. This paper will review the major innovations and developments in sequencing technology and briefly summarize their methodologies.

The first group that was able to sequence DNA was the team of Allan Maxam and Walter Gilbert (Maxam and Gilbert). This was a first generation sequencing reaction, and was developed in 1976-1977. This method uses purified DNA and relies on chemical modification of DNA bases (like depurination of adenine and guanine using formic acid and methylation using hydrazine or dimethyl sulfate). The 5′ end is radioactively labeled so that it can be visualized in a gel, and then fragments of modified DNA are electrophoresed. Autoradiography can then be used to visualize the sizes of each DNA fragment. The maximum read length for this technique was approximately 100 bases long.

The next major innovation in DNA sequencing was the Sanger dideoxy chain termination method. This was developed in 1977 by Frederick Sanger (Sanger, Nicklen, and Coulson), and became much more popular than Maxam and Gilbert’s method. Sanger sequencing is a synthesis reaction and uses dideoxy nucleotides to randomly terminate synthesized strands of DNA. The DNA strands that had been terminated with ddNTPs originally were run in 4 different lanes (one for each ddNTP) and were radiolabeled so that they could be visualized with autoradiography. Later innovations made Sanger sequencing even easier when each dideoxynucleotide was labeled with different fluorescent dyes. As such, sequences could be run on a single gel in a single lane. This method was the most popular way of sequencing DNA for many years, and was prevalent until about 2004. While read length was initially about 100 base pairs long, Sanger sequencing now has a read length of about 800 to 1000 base pairs long when run in capillary gels.

With the start of the human genome project, it was necessary to find ways to sequence DNA much more quickly and more cost-effectively than had been done previously. This led to the development of so-called “second generationâ€Â DNA sequencers. It also allowed for the use of smaller samples for sequencing.

One of the first major automated platforms was the Roche 454 (Margulies et al.). This utilizes pyrosequencing, which is a synthesis type sequencing reaction. This also uses emulsion PCR on beads. When a dNTP is incorporated, it releases a pyrophosphate (PPi). ATP sulfurylase is present in the reaction mix, and when PPi is released, converts it to ATP, which can activate luciferase and the emission of light. The Roche 454 can measure the amount of light given off and relate it to the number of nucleotides that have been incorporated. One problem with this type of sequencing is that it can be difficult to accurately characterize sequences of the same nucleotide in a row as the intensity of the pyrophosphate peak given off does not have a linear relationship with the number of homopolymers present. The read length for 454 is approximately 250 base pairs long, and the error mode tends to have indels.

The next major second gen sequencer is the Illumina Solexa platform (Bennett). The chemistry of this platform is that it utilizes reversible terminators and sequences by synthesis. A flow cell is covered with DNA oligonucleotides that are complementary to adaptor sequences that have been ligated to the ends of fragmented genome pieces. As the genome fragments are streamed across the surface of the flow cell, they will randomly bind and go through multiple cycles of denaturation and extension, which creates clusters of clones. After these clusters have been generated, they are loaded into a sequencer which measures fluorescent signals as single nucleotides are incorporated by taking a picture and noting the location of fluorescence. Read lengths are about 26-50 bases on average, and the types of errors that are typically present tend to be SNP errors.

Another important second generation sequencer is the ABI-SOLiD (Sequencing by Oligonucleotide Ligation and Detection) sequencing platform (Valouev et al.). This is another sequencing by synthesis reaction, but unlike Illumina and 454, which use polymerases, this uses ligases. After using emulsion PCR on beads to create clonal clusters, primers base pair to a known adapter sequence that has been ligated to the genomic DNA. Differently labeled probes competitively base pair to the sequencing primer, and sequencing goes through several cycles in which different primers are used each time to bind to positions offset by a single nucleotide each time. DNA bases are added in groups of two in this method. Average read lengths for this technique are on average about 35 base pairs long.

The next second generation sequencing technique is Ion Torrent, which is a sequencing by synthesis technique (http://www.iontorrent.com). When nucleotides are added to a growing DNA chain, pyrophosphate and a hydrogen ion are released. Ion Torrent takes advantage of this by measuring the pH of the reaction mix after flooding a DNA strand with the four bases (one at a time) to determine sequences. One major advantage of this technique is that it doesn’t require a high-cost camera set-up to measure incorporation events. However, because it indirectly measures nucleotide addition through changes in pH, it has difficulty with accuracy in calling sequences of homopolymers, resulting in indel errors (like pyrosequencing). Average read lengths using this technique are about 200 base pairs long.

A more recent innovation is the Helicos-True Single Molecule Sequencing (tSMS) technique (Thompson and Steinmann). It is somewhat similar to Illumina sequencing in that it also uses fragmented DNA, adaptors, and fluorescently labeled dNTPs, but there is no amplification step. This helps eliminate issues with GC bias, which tend to affect amplification steps and can cause errors in base calling. Average read length is greater than 25 base pairs.

Pacific Biosciences’ SMRT technology (Single Molecule Real Time sequencing) immobilizes a DNA polymerase at the bottom of a well and is a sequencing by synthesis technique (Eid et al.). Fluorescently labeled phosphate groups in dNTPs are added to the reaction mix and as the base is added to the growing DNA strand, the machine can measure the light that is given off (each base is labeled with a different fluorescent molecule). The major advantage of this technique is that it can sequence very long reads (more than 1000 bp!) which is very important in de novo sequence assembly. In addition, PacBio can also measure methylation of DNA sequences based on the kinetics of addition of base pairs (using the observation that modified base pairs tend to take longer to incorporate into a DNA strand). Furthermore, this technique can also potentially use a single molecule of DNA, which reduces any GC bias that occurs due to amplification.

The final technique that will be discussed here is nanopore sequencing (Stoddart et al.). The idea behind this is that DNA may be threaded through a nanopore one base at a time. As it’s fed through, the sequencer can measure the change in current as it passes through (which will vary based on what base is moving through the pore). Thus, the sequence can be determined straight from the DNA without the need for modifications or reagents. In addition, because this can be done on a single molecule, there is again no need for amplification and thus no possibility of any GC bias in base calls.