The Promise of Nanopore Technology
For those following DNA sequencing trends closely, nanopores have been something of a buzzword for a number of years, representing a theoretical platform for fast, cheap, and ubiquitous DNA sequencing. Nanopore sequencing is now becoming a reality, but since the concept was introduced, other technologies have reduced sequencing costs by several orders of magnitude, raising the bar for success.
The importance of nucleic acids in genetic inheritance has been known for more than a century, but until the early 2000s, the cost and difficulty of sequencing substantial amounts of DNA meant that, as a practical matter, most sequence information was inaccessible. Yet, over the past decade, the introduction of new high-performance sequencing platforms has led to a new paradigm in which obtaining many short sequences has become trivially inexpensive.
A common comparison is with Moore’s law of semiconductor scaling, which describes how digital computing has become exponentially faster and cheaper than it was a generation ago. Just as Moore’s law has brought us into an era of big computing, next-generation sequencing has ushered in big genomics, with high-performance platforms searching for information in huge data sets that were previously impossible to collect and analyze. The same semiconductor advances have also enabled a personal computing revolution, in which low-cost personal and mobile electronic devices are tightly woven into our lives. Analogously, there is rapid growth in applications of personal genomics, as it becomes practical to use genomic sequencing as a routine medical tool. DNA sequencing has become so efficient that it is increasingly finding new uses in fields as broad as forensics, genealogy, agriculture, and identification of bacterial and viral infections, to name a few.
Nanopore sequencing technology represents a fundamental change in the way that genomic information is read, and it continues to have the potential for orders of magnitude of further improvement in the speed and cost of DNA sequencing. However, despite the buzz that nanopore sequencing has generated, it remains relatively immature. Nanopores are anticipated to combine several highly desirable features in sequencing platforms: high throughput, fast run times, long read lengths, single-molecule sensitivity, and low cost. To understand the source of these potential improvements as well the challenges in producing a viable nanopore sequencing platform, lets take a closer look at nanopore technology.
The Nanopore Sensor Concept
A nanopore is an exceedingly small hole through a thin membrane, submerged in a salt buffer solution. When an electrical potential is applied to the solutions on either side of the nanopore, an electric field develops at the pore, driving dissolved salt ions (such as potassium and chloride) through the pore and establishing an electrical current. The device’s distinguishing feature is its nanoscale size; since its diameter is not much larger than many individual molecules, one molecule passing through the pore can measurably change the ionic current. In most conceived nanopore systems (though not all), the information contained in this varying ionic current signal determines the usefulness of the platform.
Nanopores for Single-Molecule Sequencing
The quintessential vision of nanopore sequencing is for an applied electric field to force long intact DNA molecules to thread single file through a nanopore, while each of the four bases is identified in real time by a unique ionic current profile. While this is an attractive concept, it has not yet been achieved, thanks to the fast and stochastic motion of single DNA molecules. DNA molecules can pass through nanopores at speeds greater than 1 million bases per second, while the small difference in measured current between bases makes achieving acceptable signal-to-noise ratios at these timescales very difficult.
Several solutions have been proposed for this problem. One incorporates an enzyme molecule at the pore (such as DNA polymerase), which ratchets a DNA molecule one base at a time, yielding a stepwise motion of the DNA through the pore at speeds much slower than a freely moving molecule. Another strategy attaches bulkier molecules to the middle of a sequence, which get stuck at the pore and slow down its motion. A third strategy uses enzymes that sequentially cleave off portions of the molecule being sequenced, and these cleaved products pass through the pore independent of the rest of the molecule. Still other approaches chemically reencode a DNA sequence into larger surrogate molecules and then use nanopores to detect these larger molecules. While all of these approaches remain in active development, thus far, enzyme ratcheting appears to be the closest to a viable commercial sequencing product.
Comparisons with Existing DNA Sequencing Technologies
There are already a number of commercially available DNA sequencing platforms. It is reasonable, then, to ask why nanopore sequencing attracts such attention. What could make a nanopore sequencing platform so powerful?
First, nanopores are single-molecule detectors. Rather than measuring the aggregate response of thousands of molecules (an ensemble), the signals from nanopores arise from individual DNA molecules. This allows sequencing to proceed directly from a genomic DNA sample, without biases introduced from sample preparation steps, which replicate many copies of each molecule.
Working with single molecules also allows long, uninterrupted reads. Modern high-throughput sequencing begins by chopping up a DNA sample into large numbers of short fragments, which are each sequenced separately. Overlaps in these fragments are then used to digitally reassemble the original sequence, and the length of each read is an important factor in the accuracy of this assembly process. Read lengths in ensemble methods are typically limited to tens or hundreds of bases by the inevitable misalignment of the signals from the many molecules in the ensemble. Since nanopore sequencing involves only one analyte molecule at a time, in theory, it can produce reads thousands of bases long.
Working with individual molecules may also reduce the run-time of nanopore sequencing systems, as there is no need to throttle the number of bases per second to ensure synchronization between ensembles of molecules.
The fact that nanopore measurements are electrochemical rather than optical provides an additional advantage. The up-front costs of next-generation sequencing instruments can be tens to hundreds of thousands of dollars, with large costs for the optical components and cameras for high-performance fluorescent imaging. In comparison, electrochemical instrumentation is quite inexpensive.
Types of Nanopore Sensors
Physically, there are several different types of nanopore sensors. Some nanopores are actually repurposed biological ion channels, such as alpha hemolysin, a molecule secreted by Staphylococcus aureus (the bacteria responsible for staph infections), and MspA, an ion channel found in the bacteria Mycobacterium smegmatis. To use these biologically derived nanopores, they must first be inserted into suspended films that resemble the composition of cell membranes.
Other types of nanopores are fabricated by forming holes in solid-state insulating materials such as silicon nitride, silicon dioxide, or plastic polymers. Currently, the most popular method for forming solid-state nanopores is by drilling through a thin freestanding dielectric using a focused electron beam inside a transmission electron microscope; other approaches rely on ion beams or electrochemical etching.
While ion channel proteins and insulating solid-state pores have represented the majority of nanopore demonstrations, other more exotic nanopores are also under active development, such as those built using self-assembled DNA origami (using DNA as a structural element) or with electrically conductive membrane materials such as graphene.
Biological nanopores have an advantage over solid-state nanopores when it comes to repeatability; every protein is atomically identical, and the molecules naturally self-assemble. However, making structural modifications to biological proteins can be difficult, the process of inserting biological nanopores into lipid bilayers is not always predictable, and biological nanopore signal levels can be comparatively weak. Solid-state nanopores, meanwhile, have struggled with repeatability and low fabrication throughput; if their fabrication can be improved, they may have advantages in signal amplitudes, mechanical stability, and cointegration with microelectronics.
Electronics and Fluidics for Nanopores
Instrumentation for nanopore sensors has much in common with patch-clamp electrophysiology systems for ion channel recordings, and both systems are measured using amplifier circuits that apply a constant voltage while recording ionic current as a function of time. Signal levels for nanopores are typically in the picoampere to low-nanoampere range, and the signal-to-noise ratio is paramount. Important sources of noise in nanopore measurements include low-frequency fluctuations of pore conductance; thermally induced fluctuations; and high-frequency noise related to the amplifier and parasitic capacitance. Noise is a function of the measurement time resolution, and thus it can be reduced by filtering the measurement to make it slower; however, this reduces the signal as well as the noise. Fundamental improvements in signal-to-noise levels must come from increasing the signal amplitude, slowing down the analyte molecules, or addressing noise at its source rather than through filtering.
Nanopores are inherently three dimensional, which introduces new challenges in designing the fluidics of large-scale multiplexed nanopore sequencing platforms. Appropriately scaled arrays of amplifiers can be designed, but unlike other arrayed platforms, nanopores’ electrodes need to physically contact two sides of the sensor as well as ensure electrical isolation between adjacent channels. Arrays of biological nanopores have been demonstrated in grids of microwells, each one sealed by its own lipid membrane. Unfortunately, this approach does not easily translate to solid-state membranes, and parallelized solid-state nanopore platforms are still in development.
Signal Processing for Nanopores
There is an important role for statistics and digital signal processing in nanopore sequencing, thanks to fast stochastic signals and prominent noise contributions. Molecular signals from nanopores tend to appear at irregular intervals, and signal-to-noise ratios are often marginal; the development of more sophisticated filtering and inference algorithms are bound to yield improvements. Depending on the details of each nanopore sequencing implementation, there will also be other uses for digital signal processing. For example, signals from biological nanopores have been shown to be sensitive to not just one but three to five adjacent bases; to decode the sequence from the raw traces can require algorithms inspired by convolution coding algorithms developed for wireless communications.
Remaining Challenges and Possibilities
While much very real progress has been made, nanopores’ breakthrough commercial success remains just around the corner. Nonetheless, nanopores’ tremendous promise remains intact, and their potential to greatly improve sequencing speed, cost, instrument size, and read length have kept nanopores on the sequencing community’s radar screen. Oxford Nanopore has been the most visible of a number of companies striving to bring nanopore sequencing to market. Nanopores hold promise as ultrasensitive platforms for many applications outside the scope of DNA sequencing as well.
An oversimplification of nanopores’ remaining challenges is that to be competitive they need higher signal-to-noise ratios and improved reliability. These are multidimensional challenges that can be addressed by engineering nanopores with larger signal amplitudes; by slowing down DNA motion through nanopores; by reducing noise and speeding up measurement electronics; and by creating automated systems for producing and measuring large parallel arrays of nanopores.
Nanopore sequencing remains a hotly pursued area, and, no doubt, while this article is in press there will be new announcements of progress. Meanwhile, other sequencing technologies will continue to be improved and refined, raising the threshold for nanopores’ success but also fueling new applications for low-cost sequencing and providing ever-more reason to root for nanopores to find their way outside the laboratory.
For Further Reading
- M. Wanunu, “Nanopores: A journey towards DNA sequencing,” Phys. Life Rev., vol. 9, no. 2, pp. 125–58, 2012.
- J. Shendure and E. L. Aiden, “The expanding scope of DNA sequencing,” Nat. Biotechnol., vol. 30, pp. 1084–1094, 2012.
- D. Branton, D. W. Deamer, A. Marziali, H. Bayley, S. A. Benner, T. Butler, and J. A. Schloss, “The potential and challenges of nanopore sequencing,” Nat. Biotechnol., vol. 26, no. 10, pp. 1146–1153, 2008.