Clive Brown is a legend in next-generation sequencing (NGS). As former CTO of Oxford Nanopore Technologies (ONT), he helped drive two decades of innovation that transformed genome sequencing. In 2005, Brown's team at Solexa sequenced the first viral genome, a milestone preceding the company's acquisition by Illumina in 2007 for $600 million. He later led nanopore sequencing development at ONT, helping reduce whole-genome sequencing (WGS) turnaround times from years to hours.

Those advances reshaped precision medicine. A landmark achievement came when Euan Ashley, MD, PhD, and colleagues at Stanford Medicine sequenced and interpreted a patient's genome in five hours and two minutes, demonstrating the feasibility of same-day genomic diagnosis for critically ill patients. More recently, a collaboration among Broad Clinical Labs, Roche, and Boston Children's Hospital completed WGS and analysis in three hours and 59 minutes using Roche's sequencing by expansion (SBX) technology.

Brown, who stepped away from ONT in 2025, has followed these developments closely but is not impressed by the latest speed record. "There's probably a diminishing return beyond a few hours," Brown told Inside Precision Medicine. "Both of those platforms have a real-time readout. ... So, this idea of turnaround time—it took four hours to get the data and then half an hour to analyze it—well, you can actually analyze as you run."

His reaction reflects a broader shift in thinking. Rather than focusing on sequencing genomes ever faster, Brown believes the next frontier in diagnostics lies in directly sensing the biochemical consequences of disease. "If [ONT] wanted ... they could claw that back if they put their mind to it—but does it really matter?" said Brown. "If they literally just designed binders for aberrant proteins, you don't need to sequence them. So, will sequencing last for diagnostics?"

For Brown, the future of precision medicine may depend less on reading DNA and more on detecting the molecular signals that reveal disease in real time.

Proteins as the primary signal

This framework comes into focus in the context of cancer. Liquid biopsy approaches attempt to detect circulating tumor DNA (ctDNA) in blood, but these signals can be sparse, unstable, and difficult to interpret. Tumors also release far more than DNA, shedding proteins, metabolites, and immune-modulating signals that may reflect disease activity more directly.

"It might be easier to make a lot of sense of it based on measuring the full complement of blood proteins, [rather] than trying to find little bits of tumor DNA," Brown said.

The multi-billion-dollar liquid biopsy market (roughly $15 billion today, with projections exceeding $30 billion over the next decade) and its major players—Guardant Health, Foundation Medicine, and Natera—suggest otherwise. Modern liquid biopsies generally achieve excellent specificity (often >95% and, for some applications, approaching 99%), making positive results highly reliable.

Sensitivity, however, depends strongly on the clinical setting: it is relatively high for advanced cancers and recurrence monitoring but remains substantially lower for the earliest-stage cancers, where limited ctDNA makes detection much more challenging.

Michael Snyder, PhD, professor of genetics at Stanford University and a pioneer in precision health, stands by ctDNA as the most mature molecular approach for early cancer detection—at least for the time being. "Right now, ctDNA is more sensitive and can follow around 50 different cancers," Snyder said. "Protein signatures are just emerging and may ultimately take over, but they are not there yet."

Pamela Silver, PhD, professor of systems biology at Harvard Medical School, expects these diagnostic substrates to apply beyond cancer. "Going forward, developments in blood analysis for proteins and metabolites are going to be huge in terms of diagnosis and determining drug action," Silver told Inside Precision Medicine. "We have used blood biomarkers for decades and know that this will be the way." The challenge, she noted, will be interpreting the enormous datasets these assays generate, making artificial intelligence (AI) an increasingly central component of future diagnostics.

Rate-limiting recognition

The bottleneck in moving from sequencing to sensing is not measurement but molecular recognition. As Brown argues, the challenge lies in developing molecules that can reliably bind specific biological targets at scale. "There's just an absence of binders," Brown said. If that capability matures, diagnostics could shift from reading genetic fragments to simultaneously detecting thousands of proteins that directly reflect physiological states.

AI is beginning to make that possible. Advances in protein modeling and molecular design, including work by 2024 Nobel laureate David Baker, PhD, are enabling researchers to computationally create highly specific binding molecules. At the University of Washington's Institute for Protein Design, Baker's group has developed artificial miniproteins that target difficult-to-detect biomarkers, including human hormones. "We've designed binders to over 250 targets just in my group," Baker told Inside Precision Medicine. These molecules also offer practical advantages. "Designed binders are cheaper to manufacture and often more stable than antibodies," he said, "so they have advantages for multiplexed diagnostics."

Brown is optimistic but cautious about clinical translation. "We're living through a time now where AI is enabling people to design binders to proteins, but there's no evidence yet that they're any good," said Brown. "But people are doing it, and they will get good. The iteration of experimental work in AI will improve it dramatically. It's going to become very, very easy."

Silver believes synthetic biology is steadily closing the gap. Biology already provides highly specific recognition systems, while expanding genome databases, improved protein design, and new signal outputs—including electrical, magnetic, and ultrasound-based reporters—are broadening what biosensors can detect. "The ability to design new proteins that can act as sensors grows by the day," Silver said.

If reliable binders become routine, then diagnostics may no longer depend on natural antibodies or lengthy laboratory development. Recognition systems could instead be rapidly designed and deployed, making biology far more measurable. In that future, sequencing would remain essential, but primarily as the discovery engine for determining what should be sensed.

Decentralized diagnostics

One of the less glamorous but more important constraints in diagnostics is infrastructure. Even with advances, sequencing requires complex workflows like sample preparation, instrument calibration, centralized processing, and specialized interpretation pipelines. Portable sequencers still need expert supervision and controlled environments.

According to Brown, binder-based sensing systems could completely invert this structure. "It might be quite hard to decentralize the sequencer," Brown said. "It's probably quite easy to decentralize binder-based assays." Diagnostics' physical footprint shrinks if molecular recognition is cheap, stable, and programmable. Brown imagines that testing with "little protein arrays with little sensors that bind to blood markers" could take place in places like pharmacies, homes, or even clinics rather than at centralized facilities.

Silver argues that decentralized diagnostics represent one of the greatest opportunities in healthcare. Point-of-care testing speeds up treatment decisions and reduces patient uncertainty by eliminating centralized laboratory delays. The economic effects go beyond developed healthcare systems. Many regions lack centralized laboratory infrastructure, making inexpensive field-deployable diagnostics essential rather than convenient. In field settings, portable DNA sequencing remains rare but is the gold standard for many applications. Instead, biosensors could be used as rapid frontline tests and sequencing for clarification.

If sensing becomes simpler than sequencing, there are major implications for industry players. Large diagnostics companies such as Roche have built entire business models around centralized testing infrastructure. A move toward distributed sensing does not just introduce a new technology—it challenges the economic logic of how diagnostics are delivered.

Continuous monitoring

If molecular sensing becomes cheap and scalable, then diagnostics is no longer an event, but a process. Today, disease detection is often reactive: symptoms appear, tests are ordered, and snapshots of biology are taken at a single moment in time. But biological systems are dynamic, constantly shifting across states that may not be visible in isolated measurements. A continuous sensing model would change that entirely.

In this emerging vision, AI continuously analyzes molecular data over time. Instead of searching for a single disease marker, algorithms track thousands of biological signals simultaneously. "Something weird appears, and it will flag it," Brown said. "It'll be a routine surveillance tool." The shift would mirror trends already underway in digital health, where wearable devices increasingly monitor physiological signals in real time.

Brown's prediction extends the concept into molecular biology. Rather than waiting until a patient develops symptoms, disease could be detected through subtle changes in protein expression, immune activity, or other molecular signatures long before conventional diagnosis. "Instead of feeling sick and then going for a scan," Brown said, "I think it'll just be picked up from frequent blood tests much more cheaply."

Snyder sees continuous monitoring as an extension of the digital health revolution already underway. Wearable technologies, he argues, are largely ready today, whereas many molecular sensing technologies remain under active development. If monitoring becomes routine and widely adopted, then the economics could dramatically improve through scale. "People will gain because they will [detect] disease early, before symptoms, and that will help keep them healthy," said Snyder. "Health systems should gain too because they can manage people's care more efficiently." The principal obstacle, however, is reimbursement. "The biggest challenge from the healthcare standpoint is who pays?" Snyder said. "No one pays to keep people healthy."

What emerges is a model of medicine that resembles infrastructure more than intervention. Disease is not discovered when symptoms appear; it is flagged when patterns deviate. The boundary between health and illness becomes increasingly statistical rather than categorical.

A post-sequencing future?

Sequencing remains foundational for discovering genes, mapping regulatory networks, and identifying disease mechanisms. But its role may shift from a clinical endpoint to an upstream discovery engine. In this model, sequencing generates hypotheses, while sensing enables real-time interpretation. Rather than doing away with sequencing, the next generation of diagnostics may rely less on it as medicine's primary interface with biology.

Brown appears to be moving in this direction through ventures focused on molecular sensing and AI-designed recognition molecules. "I'm interested in molecular sensing using ultra-cheap portable sensors," Brown said, keeping details scarce. "I'm on the third [patent] now, and then I'll probably raise some money at that point with the proof-of-concept data," said Brown. "I'll probably stay in stealth mode until I've got something shaped up that is pretty convincing." Whatever Brown is working on, he said it's "something unusual that will upset some people."

Whether the future centers on sequencing, molecular sensing, wearables, or all three, the shared goal is earlier intervention. Snyder emphasizes detecting disease before symptoms emerge, while Silver highlights affordable biosensors for resource-limited settings. As AI, protein engineering, and biosensors advance, precision medicine may increasingly depend on continuously measuring biology, with sequencing serving as the discovery engine.

---

 

Jonathan D. Grinstein, PhD, North American editor for Inside Precision Medicine, investigates the most recent research and developments in a wide range of human healthcare topics and emerging trends, such as next-generation diagnostics, cell and gene therapy, and AI/ML for drug discovery. He is also the host of the Behind the Breakthroughs podcast, featuring people shaping the future of medicine. Jonathan earned his PhD in biomedical science from the University of California, San Diego, and a BA in neural science from New York University.