The sequence outputs and read lengths of the most commonly used sequencing plat forms are shown in Figure 1. The run times vary from several hours to several days. At the time of writing, the Illumina platform was by some distance the market leader.
Fig1. Sequence throughput and read lengths for some common massively-parallel DNA sequencing platforms. For some methods the accuracy of sequencing is low for individual sequence reads but can be compensated for by having high sequence coverage (when each region of the DNA is represented by many individual sequence reads). The maximum sequence throughput of close to 1,000,000 Mb of DNA corresponds to 1000 Gb or about 300 haploid human genomes. (Data from Reuter JA et al. [2015] Mol Cell 58:586–597; PMID 26000844.)
Roche/454 pyrosequencing
This, the first of the massively-parallel technologies, was brought to market in 2005. It is based on pyrosequencing (Box 6.4). The procedure is summarized in Figure 2, and described in detail by Margulies et al. (2005) (PMID 16056220).
Fig2. Roche/454 reiterative pyrosequencing. (A) The input DNA is nebulized to 300–800 bp fragments, made blunt-ended, and then ligated to double-stranded adaptor oligonucleotides, A and B, each of which contains primer sequences for PCR and then pyrosequencing. One strand of the B adaptor carries a biotinylated nucleotide, allowing tagged DNA fragments to be isolated by binding to streptavidin-labeled magnetic beads. Double-stranded fragments carrying A and B sequences are captured and then denatured. The single-stranded molecules released into the supernatant comprise the sequencing library. (B) A separate set of agarose beads with a covalently bound sequence complementary to the B adaptor is used to capture the single-stranded end-tagged fragments of the test DNA. The conditions are chosen so that most beads capture either zero or just one DNA fragment. PCR primers, polymerase, and the other components of a PCR mix are added to the beads, and the whole is broken into minute droplets (“microreactors”) in an oil emulsion, so that on average a droplet includes only a single bead. The whole emulsion is then put through 40 cycles of PCR. At the conclusion of the PCR, the emulsion is broken down and the DNA anchored to the beads is made single-stranded by denaturation and washing. The beads, now each carrying several million identical, single-stranded copies of one particular fragment of the test DNA, are loaded into individual wells of a fiber-optic slide (a “picotiter plate” [PicoTiterPlateTM]) that contains 1.6 million wells. Smaller beads carrying immobilized enzymes required for pyrosequencing are layered on top of each well. A CCD camera images the pattern of light emission across the wells as each successive dNTP is washed over the plate.
The strengths of the 454 technology are its relatively long reads and speed. It can produce 400–600 Mb of sequence in a 10-hour run. Disadvantages include its relatively low throughput, high reagent costs, and an inability to size homopolymer runs accurately (a run of eight adenines, for example, cannot be reliably distinguished from a run of nine adenines by the amount of light emitted). In the light of newer technologies, Roche discontinued support for 454 sequencing in 2016.
The ABI SOLiD technique
Applied Biosystems introduced the SOLiD (Sequencing by Oligonucleotide Ligation Detection) system in 2007. Its special feature is the use of an ingenious DNA ligation-based chemistry that checks each base independently twice for mismatches (two-base encoding). This results in a very low error rate for either miscalled or mismatched bases.
The test DNA is fragmented, ligated to adaptor oligonucleotides, bound to beads, and amplified by emulsion PCR. The beads, each carrying millions of clonally amplified single-stranded copies of one particular fragment, are immobilized on the surface of a glass slide and exposed to a cocktail of fluorescently labeled 8-mer oligonucleotides in the presence of DNA ligase. Ligation depends on correct base pairing of the two bases at the 5′ end of the oligonucleotide. After washing away unligated probes, a camera records the color of the ligated probe. The 3′ end of the probe, together with the fluorescent label, is then cleaved off and a further round of ligation started. Successive cycles of ligation and cleavage follow, until the maximum read length is reached. The resulting double stranded DNA is then denatured, the ligation product washed away, and the whole process started over again, but using a sequencing primer displaced one nucleotide from the initial primer. This is done five times in total (Figure 3).
Fig3. Principle of two-base encoding used in ABI SOLiD DNA sequencing. The top row shows how one primer is extended in successive cycles of ligation and cleavage (seven cycles are shown here, being color-coded as shown in the key at bottom). Ligation depends on correct base pairing (marked by dots) of nucleotides 1 and 2 of the probe with the template. Nucleotides 3–5 are degenerate. After ligation, the probe is cleaved after nucleotide 5, removing the fluorescent label and the degenerate downstream nucleotides. A second cycle of ligation and cleavage then takes place; this time, ligation depends on exact pairing of nucleotides 1 and 2 of the probe with nucleotides 6 and 7 of the template (marked by dots). After maybe a dozen such cycles, the whole newly synthesized strand is dissociated from the template and washed away, and the cycles of ligation and cleavage are started again using a different sequencing primer that anneals to a slightly different position on the template, as shown in the second row. Overall, five different sequencing primers are used. Each position in the template is interrogated twice, allowing very low mismatch or miscalling error rates.
In 2013, Applied Biosystems introduced a new and much simpler method (“Wildfire”) for preparing the sequencing templates. This did away with beads and emulsion PCR, relying instead on a system of isothermal amplification of DNA fragments anchored to a glass slide, to produce compact clusters of clonal copies. Although the method is different, the effect is similar to the bridge PCR used by Illumina for its next-generation system, as described below. Full details of the Wildfire technology can be found in the paper by Ma et al. (2013) (PMID 23940326).
The particular advantage of the SOLiD system is its high accuracy, including with homopolymer runs. Its main disadvantage is the short read length, up to 75 bp, which makes assembly more difficult. Each full sequencer run takes 7 days and generates around 4 terabytes of raw image data, which are analyzed to produce up to 200 Gb of finished sequence.
Illumina/Solexa sequencing The market-leading Illumina technology was originally developed by the Solexa company (Bentley et al. [2008], PMID 18987734; see Further Reading). The bridge PCR technique (Figure 4) generates clonal clusters of amplified fragments bound to a glass slide, and these are sequenced by synthesis in a way similar to Sanger dideoxy sequencing, but using reversible chain terminators.
Fig4. Bridge (cluster) amplification. (A) The surface of an Illumina flow cell is carpeted with two types of single-stranded oligonucleotide (red and blue bars) that are tethered to the surface at their 5′ ends by a short, flexible linker. The two different types of fixed oligonucleotides represent the two types of adaptor sequence fixed at the ends of the test DNA fragments (1, 2, 3, 4, and so on). Test DNA fragments that have been made single-stranded can bind at one end to a tethered oligonucleotide with a complementary sequence. The DNA templates are present at a low concentration to ensure wide spacing between individual bound DNA fragments. (B) Individual DNA fragments are amplified by cycles consisting of: (i) DNA synthesis using the bound test DNA fragment as a template; (ii) denaturation leading to exit of the original test DNA fragment; and (iii) bridging (where the tethered DNA bends so that the adaptor sequence at the free end hybridizes to a neighboring complementary adaptor-specific oligonucleotide fixed on the surface). (C) The end result is a series of physically separate DNA clusters, each containing multiple copies of just one type of DNA fragment (monoclonal DNA clusters). For clarity, this example shows four clusters, but in practice there will be many millions of DNA clusters.
Illumina/Solexa sequencing uses dye-labeled chain-terminator dNTPs (Figure 5A), as in Sanger sequencing, but in this case no normal (unmodified) dNTPs are present. The reaction therefore stops after incorporation of a single nucleotide. This is imaged to record the color, and then both the blocking group and dye are removed, allowing a second nucleotide to be added. Thus sequencing goes in cycles of incorporation, imaging, and cleavage (Figure 5B and C).
Fig5. The Illumina/Solexa sequencing-by-synthesis method involves repeated cycles of incorporation of reversible terminator deoxynucleotides, imaging, and cleavage. (A) An example of a reversible terminator dNTP where the normal 3′ hydroxyl group is replaced by a 3′ O-azidomethyl group (imagine a 3′ O-methyl group but with the three hydrogen atoms replaced by nitrogen atoms). In addition, the base (thymine in this example) has a side chain containing an azidomethyl (–N3 ) group and a fluor group. Once a reversible terminator has been incorporated into the growing DNA chain, the fluorescent signal from the fluor can be imaged, but then a single chemical treatment cleaves both azidomethyl groups (red arrows), leaving free hydroxyl groups. The result is that the fluor group is released, and the 3′ hydroxyl group is restored so that a new nucleotide can be incorporated. (B) The cycle of nucleotide incorporation, imaging, and cleavage. This shows synthesis of three out of the many millions of growing DNA strands synthesized using single-stranded DNA templates (gray bars) using an adaptor-specific primer. The process continues through many cycles, of which only the first two are shown here. (C) Example of four-color imaging. The 19 colored circles in each panel represent 19 of the millions of growing DNA strands. The color given in each circle represents the last nucleotide to be inserted in the growing DNA chain and the six individual panels show imaging after six cycles of nucleotide incorporation. For clarity, two of the growing DNA strands are shown enclosed in white rings so that successive color images (representing nucleotide incorporation) can be readily followed from one panel to the next. For the top one of the two encircled growing DNA strands, the successive color changes (moving between the images from left to right) indicate the sequence CATCGT; the bottom one remains green for six cycles, indicating the sequence CCCCCC. (B and C, adapted from Metzker ML [2010] Nat Rev Genet 11:31–46; PMID 19997069. With permission from Springer Nature. Copyright © 2010.)
The chemical cleavage step leaves an entirely normal 3′ hydroxyl group for addition of the next dNTP, but part of the dye linker remains attached to the base. A modified DNA polymerase is used, both to improve incorporation of the heavily modified dNTPs and to tolerate the “scar” left in the growing newly synthesized strand after cleavage of the dye. Accurate data depend on the incorporation and cleavage reactions being complete across all the millions of clusters on the flow cell, and this limits reads to around 100 nucleotides.
Illumina produces a range of machines to suit either large-scale genome projects or small clinical sequencing services, together with kits to simplify all aspects of library preparation and data acquisition.
Ion Torrent systems
The first Ion Torrent PGM (Personal Genome Machine) was released in 2010. The work flow of Ion Torrent systems has much in common with that of the 454 pyrosequencing system: both use fragmentation, adaptor ligation, capture by beads, emulsion PCR, and deposition of beads, each carrying millions of single-stranded copies of one particular DNA fragment, into wells of a plate. As in 454 technology, individual dNTPs are washed across the plate and sequencing proceeds by synthesis. However, Ion Torrent systems use a radically different, and much simpler, way to follow the synthesis.
Incorporation of a correctly paired dNTP into the growing strand releases not only the pyrophosphate detected by pyrosequencing but also a hydrogen ion, H+. Ion Torrent systems detect the hydrogen ion directly as an electric signal. The Ion Torrent chip is a CMOS silicon chip, like those used in digital cameras. Below each well is an ion- sensitive field effect transistor that generates an electronic signal in response to release of a hydrogen ion. Unlike any of the systems detailed above, Ion Torrent systems do not require fluorescence and camera scanning, which allows higher speed, lower cost, and a smaller-sized machine.
Reads of 200 or 400 bp are delivered in 2.5 or 4 hours, respectively, and the throughput depends simply on the number of wells on a plate—for example, an Ion Torrent 540 chip allows 60–80 million reads of 200 bp, giving 10–15 Gb of sequence. Ion Torrent machines are thus marketed as relatively cheap and simple benchtop machines suitable for applications in microbiology or clinical resequencing. They suffer from the same problems with homopolymer runs as the 454 system.
