Nanopore Sequencing: Reading DNA by Ionic Current Dips

What It Is and Where the Reading Happens

Nanopore sequencing is a single-molecule, real-time technique in which an intact strand of DNA or RNA is read as it physically passes through a nanoscale channel. The reading event occurs at a protein nanopore — a barrel-shaped channel just over a nanometer wide at its narrowest point — inserted into an electrically insulating synthetic membrane (an amphiphilic polymer, not a natural lipid bilayer, in commercial flow cells).

The membrane separates two chambers of salt buffer (typically ~0.3–1 M KCl). Each pore sits above its own electrode and amplifier, so a MinION flow cell hosts up to 512 independent channels and a PromethION scales to thousands. A holding potential of about 180 mV drives K⁺ and Cl⁻ ions through the open pore, producing a steady baseline current near 100 pA. Because DNA is negatively charged, the same field also pulls the strand toward and into the pore.

• Cis chamber: where the DNA library and motor enzyme wait
• Trans chamber: where the strand emerges after threading
• Constriction (reader head): the ~1.2 nm zone that gates the current

The Mechanism, Step by Step

Reading a strand is a controlled, ratcheted translocation, not a free fall:

• 1. Capture. A sequencing adapter ligated to the DNA carries a tethered motor protein (a processive helicase-like enzyme). The electric field concentrates and captures the adapter at the pore mouth.
• 2. Unwinding and ratcheting. The motor enzyme sits atop the pore and, powered by ATP-fueled conformational cycling as it moves along the strand, feeds single-stranded DNA through one nucleotide at a time. This slows translocation from microseconds-per-base (far too fast to read) to a measurable ~400 bases/second.
• 3. Current modulation. As each set of bases occupies the constriction, it partially blocks ion flow, dropping the current by a characteristic amount. Crucially, the signal reflects not one base but a k-mer — several adjacent nucleotides (~5 in the constriction) contribute simultaneously.
• 4. Squiggle to sequence. The stepwise current trace (the squiggle) is digitized at ~4–5 kHz and decoded by a neural-network basecaller (e.g., recurrent or transformer models) trained to map current levels back to base sequence.

Key Molecules and Characteristic Numbers

The sensor chemistry has evolved through several named pore proteins:

• α-hemolysin — the Staphylococcus aureus toxin used in the founding experiments; a mushroom-shaped heptamer with a ~1.4 nm constriction but a long β-barrel that smears many bases together.
• MspA (Mycobacterium smegmatis porin A) — an octameric funnel with a short, sharp ~1.2 nm constriction that reads a much narrower k-mer; key to Jens Gundlach's proof-of-principle DNA reads.
• CsgG / CsgF — the curli-secretion channel from E. coli, a nonameric β-barrel that is the basis of Oxford Nanopore's R9 and R10 flow cells. The R10 pore is longer with two reader heads, improving homopolymer resolution.

Characteristic values: constriction ~1.0–1.7 nm; membrane thickness a few nm; open-pore current ~100 pA falling to tens of pA when blocked; four bases give four measurably distinct blockade levels, but because ~5 bases sit in the pore at once there are up to 4⁵ = 1024 k-mer states the basecaller must distinguish. Raw single-read accuracy on R10.4.1 chemistry is about 99% (Q20).

How the Signal Is Studied, Basecalled, and Controlled

The raw output is not sequence but a stream of picoampere measurements. Turning it into bases is a signal-processing and machine-learning problem:

• Segmentation: the squiggle is split into discrete current levels ('events'), each corresponding to a ratchet step of the motor.
• Basecalling: models such as ONT's Guppy and Dorado use recurrent/transformer neural networks trained on ground-truth genomes to convert current traces to sequence with per-base quality (Phred Q) scores.
• Adaptive sampling ('Read Until'): because reading is real-time, the software can basecall the first few hundred bases, decide the read is unwanted, and reverse the voltage to eject it — enriching targets without new chemistry.

The same raw current also encodes base modifications: 5-methylcytosine and N6-methyladenine perturb the blockade slightly differently from unmodified bases, so methylation is called directly — no bisulfite conversion needed. Direct RNA sequencing works the same way, threading native RNA (with modifications like m6A) rather than a cDNA copy.

Comparison to Related Sequencing Methods

Nanopore is the flagship of third-generation (long-read) sequencing, and it differs sharply from its cousins:

• vs. Illumina sequencing-by-synthesis: Illumina images fluorescent reversible-terminator bases added to millions of amplified clusters, giving very short (150–300 bp) but extremely accurate (Q30+) reads. Nanopore reads the native molecule with no PCR, yielding reads thousands of times longer at slightly lower per-base accuracy.
• vs. PacBio HiFi: PacBio detects fluorescence pulses as a polymerase adds bases in a zero-mode waveguide, circularizing molecules to read them many times for consensus. It is single-molecule like nanopore but still enzyme-synthesis-based, whereas nanopore performs no synthesis at all.
• vs. Sanger sequencing: Sanger uses chain-terminating ddNTPs and capillary electrophoresis — one short read at a time. Nanopore is massively parallel and reads far longer fragments.

The defining nanopore advantage is read length (spanning repeats, structural variants, and whole bacterial genomes in one read) plus direct modification detection. Its historical weakness — homopolymer errors, where long runs of identical bases give a flat current and ambiguous length — is much reduced but not eliminated by the dual-reader R10 pore.

Significance, Applications, and Open Questions

Nanopore sequencing reshaped genomics by making sequencing portable and real-time. The USB-sized MinION has been run in the field and even on the International Space Station, and was deployed for rapid pathogen surveillance during Ebola (2015) and SARS-CoV-2 outbreaks.

• Ultra-long reads (megabase-scale) were essential to the Telomere-to-Telomere (T2T) consortium's first complete, gapless human genome (2022), resolving centromeres and segmental duplications that short reads cannot span.
• Native methylation maps let researchers read genetic sequence and epigenetic marks from the very same molecule.
• Direct RNA sequencing captures full-length isoforms, poly(A) tail length, and RNA modifications without reverse transcription.

Open questions and limits remain: pushing raw accuracy toward Q30 to match short reads; further taming long homopolymer and repeat errors; controlling motor-enzyme slippage and translocation variability; and engineering solid-state and hybrid nanopores that could one day replace fragile biological pores. Protein sequencing through nanopores — reading amino acids rather than nucleotides — is an active frontier that could extend the same ionic-current principle to the proteome.

Nanopore sequencing versus other major sequencing platforms

Feature	Nanopore (Oxford)	Illumina (SBS)	PacBio HiFi
Read length	10–100 kb typical, >2 Mb possible	150–300 bp	10–25 kb
Raw single-read accuracy	~99% (R10.4.1, Q20)	>99.9% (Q30+)	~99.9% (consensus HiFi)
Detection principle	Ionic current dip through pore	Fluorescent reversible terminators	Fluorescence in zero-mode waveguide
Native modifications (5mC, m6A)	Yes, directly from raw signal	No (requires bisulfite)	Yes (kinetics-based)
Amplification / synthesis	None — reads native strand	PCR + bridge amplification	Rolling-circle, no PCR
Device footprint	USB MinION to benchtop PromethION	Benchtop to large instruments	Benchtop (Revio/Sequel)