6. Peeking Deeper into State-of-Art Ion Current Sensing based Nanopore Sequencing
Before discussing FENT nanopore transistor prototype next, I wanted to take a step back and summarize my thoughts (for the sake of completeness) on ion-current blockade-based sequencing, following recent discussions on LinkedIn, Twitter, other.
For ion-current blockade method to be able to sequence DNA effectively, we need to focus on the following three key criteria, in addition to others: How is the base-specific signal generated? What are the carriers/reporters of that signal and the medium of information transfer? High sensing speeds (rate of measurement) imply higher signal losses. What is the maximum speed at which we can resolve bases?
1. Base-specific signal generation at nanopore aperture: As I discussed in posts 3 and 4, ion current sensing is a surrogate for physically measuring each base/strand against the size of the nanopore aperture/channel. Nanopore aperture acts as a molecular ruler to compare the relative sizes of passing bases, with ions as reporters/carriers of information from nanopore to sense electrodes. Protein nanopores have the advantage that they can be designed very accurately by introducing mutations and modifiers.
Solid-state nanopore ‘ion-current sensing’ has a drawback here, because fabricating small diameter pores with exact shape and size to match protein pores is non-trivial. To resolve this, one elegant approach researchers are working on is to use of 2D materials such as graphene, MoS2, BN etc, exploring carbon nanotubes (CNTs) and other 1D nanostructures, that can yield highly structured size and shape-controlled nanopores. Fabrication of sub-nanometer pores has been demonstrated on graphene, where one or few carbon atoms are removed from the 2D sheet. However, many challenges such as high noise background, leakage current and fabrication limitations still remain.
INanoBio's FENT nanopore-transistor is a new kind of nanopore sensor that senses information at the nanopore-aperture using transistor electron-current, via electrostatic field-effect. Therefore, we do not need small nanopores with exact size and shape, unlike in the case of ion-current sensing. FENT has an inherent advantage in that we seek to electrostatically read surface charge / potential on each base for sequencing, as they pass through the nanopore.
2. Measuring the signal with sensitivity sufficient to enable base discrimination: Ion current sensing signal is in tens of pico-amperes (or less), which is fairly low-current. This is where the challenge of sensing just a few electrons/ions at the output electrode comes into play. We need specially designed high-performance ultra low-current sensing ASIC and electronics to achieve this. The area of the electronic signal amplifying circuit that is needed per-pore (on ASIC) and the contact pad area (for signal out) - define the nanopore array pitch, and number of nanopores per sq cm. I believe Oxford Nanopore has pitch around 100 micron (2021), and so can fit close to 10,000 per sq cm. I am sure this will be improved over time.
Roche-Genia has taken a different approach, with respective advantages and disadvantages, that we can glean from their published literature and press reports. They initially used base-specific molecular tagging to increase the signal amplitude (reference link). In another parallel approach, they are developing what’s termed Sequencing by eXpansion (SBX), detailed on their website here. Using Xpandomer technology (Stratos Genomics) they add a molecular spacer (xpandomer) between each base to physically separate out the bases prior to running through a nanopore. This relaxes the need to read bases that are just ~ 3.4 A apart on DNA, and additionally provides an option to tag/barcode the molecular spacers for increasing signal amplitude.
Increasing base signal-amplitude is great because we now do not need to discriminate low pico-amperes of current, as is the case with Oxford Nanopore, so there is no requirement anymore for high-performance low-current ASIC/electronics. A smaller simpler signal-amplification circuit (few sq microns on chip) will suffice. In addition, use of Multiplexing (switching and measuring a sub-array of pores by a fast sense element) becomes possible. So area per pore drops down dramatically. This I think is a key reason for Genia’s (now Roche) public claims of 100,000 - million nanopores on their chip, which is reasonable. The sequencing speed per pore using this approach falls to few bases (or tens of bases) per second, per pore, which would be about 1/100 (to 1/10) of per-pore speed achieved by Oxford Nanopore. However, on the same area of the chip they can now fit 100x more nanopores. So effective readout speed (base call output) per sq cm chip matches that of Oxford Nanopore.
Coming to solid-state nanopore ‘ion-current sensing’, they may have certain advantages in terms of improving signal amplitude over protein nanopores, in that we can increase the number of ions in signal by driving more current (higher voltage operation). In parallel, we can adapt measures to suppress the noise, to increase signal-to-noise ratio. As the platform will be fully solid-state, once realized and fully developed, it will remove any limits on scaled manufacturing - a major advantage over protein nanopores.
Given that INanoBio’s FENT is a brand new solid-state nanopore sensor, we can apply and evaluate it with a range of DNA sequencing approaches and chemistries. For this novel tech, the world is an oyster. We will test approaches that involve detecting tagged bases like that of Roche-Genia, and direct sequencing of unmodified DNA (like Oxford Nanopore) but at >100x higher speeds.
3. Speed of ion-current sensing: When talking about high-speed base readout using ion current sensing (per pore), we arrive at certain fundamental limitations due to physics of ion transport in solutions. As we previously discussed, ion mobility in solution is on the order of 0.001 cm2/ (V·s), which is relatively slow. By optimizing ion current sensing based nanopore sequencing platforms (protein or solid-state nanopores) we can achieve around 1000 bases per second per pore. By pushing these limits, achieving up to 5000 bases per second might be possible - not sure, this might be a stretch.
In general, speed of ion current sensing depends on (i) the speed of transport or mobility of ions in solution (carriers of information), which is limited by ionic collisions due to Brownian motion, and (ii) the speed of the sensing electronics that measures and acquires data at the electrodes. On (ii), by using higher speed electronics (amplifiers, ADCs) we can readily go from 1 million samplings per second (MSps) up to 1 billion samplings per second. However this does Not imply higher base read speeds, an important point. It does not fundamentally solve (i) - the restriction imposed by ion transport physics.
The question to ask is what is the cut-off frequency Fc (3dB loss in signal) of the specific ion-current sensing nanopore platform? Achievable base read speeds should be around ~ 1/10 Fc (see post 4, footnote 3). Fc might be around 10,000 Hz for ion current blockade nanopore sensors.
So, achieving close to one million bases per second read speed (per pore) using ion-current sensing is not feasible. The only way would be to somehow sense the information-carrying ions before they undergo collisions (not feasible - check mean free path of ions in solution). Of course, I might have missed something in this analysis and am happy to be proven wrong.
With INanoBio’s FENT this limitation of high-speed nanopore sensing is fully resolved. Being highly mobile, electrons in silicon channel (about million times faster than ions in solution) respond to charge/potential changes at nanopore in sub-micro-second to nano-second time scales. In the lab, we have achieved FENT nanopore transistor prototype with a cut-off frequency Fc > 100MHz.
With FENT being a CMOS sensor, it can be directly integrated with high-speed ASICs and electronics for ultra-fast data measurement and processing, to achieve over 100 million samplings per second (MSps). This ability for systems on chip (SOC) integration is another advantage.