The future of precision medicine may depend less on reading DNA and more on detecting the molecular signals that reveal disease in real time. While genome sequencing has accelerated dramatically over the past two decades, leading experts now argue that the real frontier lies not in sequencing speed but in developing smarter, faster ways to sense what's actually happening in a patient's body right now.

Why Is Sequencing Speed No Longer the Main Goal?

For years, the genomics field celebrated speed records. Researchers at Stanford Medicine sequenced and interpreted a patient's entire genome in five hours and two minutes, demonstrating same-day genomic diagnosis for critically ill patients. More recently, a collaboration among Broad Clinical Labs, Roche, and Boston Children's Hospital completed whole-genome sequencing and analysis in three hours and 59 minutes using Roche's sequencing by expansion technology.

But Clive Brown, the former Chief Technology Officer of Oxford Nanopore Technologies who helped pioneer modern DNA sequencing, sees diminishing returns in chasing faster turnaround times. "There's probably a diminishing return beyond a few hours," Brown explained. "Both of those platforms have a real-time readout. So, this idea of turnaround time,it took four hour to get the data and then half an hour to analyze it,well, you can actually analyze as you run".

Brown's skepticism reflects a broader shift in thinking among precision medicine leaders. Rather than focusing on sequencing genomes ever faster, the field is beginning to ask a more fundamental question: do we even need to sequence DNA for many diagnostic applications?

What Could Replace DNA Sequencing in Diagnostics?

Brown argues that the next frontier in diagnostics lies in directly sensing the biochemical consequences of disease. Instead of hunting for tiny fragments of tumor DNA in blood, doctors could measure the full complement of proteins and metabolites that tumors actually release. "If they literally just designed binders for aberrant proteins, you don't need to sequence them," Brown noted. "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

This approach addresses a real limitation of current liquid biopsy technology. While modern liquid biopsies achieve excellent specificity, often exceeding 95% accuracy for positive results, their sensitivity for detecting early-stage cancers remains substantially lower. The challenge is that early tumors release sparse, unstable circulating tumor DNA (ctDNA) that is difficult to detect and interpret.

Michael Snyder, a professor of genetics at Stanford University and pioneer in precision health, acknowledges that ctDNA remains the most mature molecular approach for early cancer detection today. "Right now, ctDNA is more sensitive and can follow around 50 different cancers," Snyder stated. "Protein signatures are just emerging and may ultimately take over, but they are not there yet".

How Is AI Enabling Protein-Based Diagnostics?

The bottleneck in moving from sequencing to sensing is not measurement technology but molecular recognition. The challenge lies in developing molecules that can reliably bind to specific biological targets at scale. "There's just an absence of binders," Brown explained.

Artificial intelligence is beginning to solve this problem. Advances in protein modeling and molecular design, including work by 2024 Nobel laureate David Baker, 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. "Designed binders are cheaper to manufacture and often more stable than antibodies, so they have advantages for multiplexed diagnostics."

David Baker, Nobel Laureate and Researcher, University of Washington Institute for Protein Design

These AI-designed molecules offer practical advantages beyond specificity. They are cheaper to manufacture, more stable than traditional antibodies, and can be rapidly designed and deployed for new targets. Pamela Silver, a professor of systems biology at Harvard Medical School, expects these diagnostic substrates to apply far 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 noted. "We have used blood biomarkers for decades and know that this will be the way".

Steps to Understanding the Shift From Sequencing to Sensing

• Real-Time Detection: Instead of waiting hours for sequencing results, protein and metabolite sensors can provide immediate readouts of disease activity, enabling faster clinical decisions and reducing patient uncertainty.
• Multiplexed Measurement: AI-designed binders can simultaneously detect thousands of proteins that directly reflect physiological states, providing a more complete picture of disease than isolated DNA fragments.
• Decentralized Infrastructure: Binder-based sensing systems could be deployed in pharmacies, homes, or clinics rather than centralized laboratories, making diagnostics faster and more accessible to underserved regions.

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," he said. "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".

What Does This Mean for Healthcare Infrastructure?

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.

Binder-based sensing systems could completely invert this structure. "It might be quite hard to decentralize the sequencer," Brown explained. "It's probably quite easy to decentralize binder-based assays." Diagnostics' physical footprint could shrink dramatically if molecular recognition becomes cheap, stable, and programmable.

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.

The challenge ahead is not technological but iterative. As AI systems design more binders and researchers test them experimentally, the feedback loop will improve rapidly. The shift from sequencing to sensing is not imminent, but the pieces are falling into place. Sequencing will likely remain essential as the discovery engine for determining what should be sensed, but the clinical frontline of precision medicine may soon look very different.