Next-generation sequencing (NGS) has matured into an essential part of modern drug development and diagnostic laboratories, and its steadily dropping costs are making it affordable for many facilities that a few years ago couldn’t justify the expense.

The coming years will build on that foundation by extracting more and different types of data from samples, leveraging AI both upstream and downstream, and streamlining workflow. The entire workflow—not just the sequencing itself—will become more cost effective, industry experts tell GEN.

The next wave

“The next major breakthrough in sequencing technology will likely come not from the sequencer alone, but from the ability to extract more meaningful biological insights from every sample,” Karen B. Chapman, PhD, associate VP, R&D Genomics, Agilent Technologies, says.

She envisions “the convergence of improved long-read sequencing, intelligent assay design, AI-enabled analysis, and integrated automation across sample preparation, target enrichment, quality control, sample tracking, and data interpretation.”

Multiomics assays will likely be the next big thing.

“Sequencing has traditionally focused on the four canonical bases, but DNA molecules contain much more information, including methylation, damage, and secondary structure, that can reveal different layers of biology,” Mark Van Oene, president and CEO, PacBio, points out. “We see sequencing moving toward a rapid test that can deliver comprehensive, highly accurate results within hours, enabling scientists and clinicians to make decisions faster.”

The experts at New England Biolabs (NEB) concur. “That’s one of the next big opportunities in sequencing: to gather different data types and provide them in a way that allows them to be interpreted usefully in the context of the human genome,” Andrew Barry, associate director, product portfolio management, NEB, says.

Such multiomics assays will “enable richer and more comprehensive biological insights,” Lakmal Jayasinghe, PhD, CSO, Oxford Nanopore Technologies, says. “The next major breakthrough will be the ability to analyze DNA, RNA, and proteins on a common sensing platform along with epigenetics, epitranscriptomics, and post-translational modifications of proteins. The next frontier is understanding how that code is regulated and executed in real biological systems. We see proteomics as a key part of that future.”

As the industry pushes current boundaries, it is also working to improve sequencing performance through parallel workflows and automation that can accommodate a broad range of samples. Opportunities exist “to increase throughput, improve fidelity, and simplify workflows,” Jayasinghe adds.

To that end, Keerthana “Kit” Krishnan, director, NGS applications and product development, NEB, points out: “Library prep may need a bit more work.” On the upstream side, she identifies sample extraction and sample prep as areas with opportunities for improvement. Downstream opportunities, she says, are in data analysis.

For NGS to become a routine clinical tool, however, those analyses must be clinically meaningful and linked to follow-up actions, she says. She envisions NGS assay kits that are tailored to specific clinical applications.

AI adds value today

“The real breakthrough,” Chapman of Agilent Technologies points out, “will be…in delivering deeper, more actionable biological insights in a way that is economically viable, operationally practical, and accessible to a broad range of laboratories and users. AI will play a critical role in enabling this vision, unlocking new possibilities that lower computational and labor costs while enhancing accuracy and increasing operational speed.”

Across the board, instrument developers are leveraging AI both upstream and down. For example:

• PacBio uses AI upstream to transform raw sequencing data into HiFi reads and to detect DNA modifications. Downstream applications include tertiary analysis, which Van Oene says can be used “to identify and prioritize genetic variants associated with rare disease.”
• NEB’s bioinformatics team deploys AI for coding support, while the automation team is exploring its application to streamline multi-instrument automation processes for developers and for field application scientists to anticipate pitfalls customers may face. AI is also becoming an important tool in enzyme discovery.
• Agilent uses AI to learn from large datasets and develop ways to improve probe placement and target-enrichment design to enhance specificity, coverage and uniformity. Other applications may include “improving base calling, variant detection, structural variant characterization, methylation assessment and clinical interpretations,” Chapman says.
• Oxford Nanopore uses AI to help build more accurate models and to directly detect DNA and RNA modifications. Looking ahead, “AI will become increasingly important in helping scientists, clinicians, and biopharma companies interpret complex multiomics datasets,” Jayasinghe says.
• Illumina has embedded AI tools in its secondary analysis pipeline and throughout its software and informatics portfolio. It has also created the peer-reviewed Primate, Splice and Promoter AI genomic models.

Future AI value

Algorithms are moving “beyond simple rule-based design and increasingly learn from large datasets to optimize enrichment performance,” Chapman says. “Over time, these approaches could evolve into fully-AI-driven panel design systems that automatically optimize probe selection, predict assay performance, minimize sequencing costs, and continuously improve…as more data becomes available.”

“Looking ahead, the most transformative role for AI may be guiding researchers toward the right biological signals to measure, rather than merely processing larger volumes of data,” she continues. “AI can help to identify the most relevant genomic regions, uncover intricate molecular patterns, integrate genomic and methylation information, and bridge genomics discoveries with protein, cellular, and metabolic biology.”

As AI becomes increasingly ubiquitous, explainable AI will become a requirement, these experts agree. The ability to interrogate, assess, and validate the AI’s reasoning is imperative not only for on-instrument applications such as base-calling, but especially for tertiary analysis where understanding the evidence behind a result is critical.

As Barry says, “To use AI effectively, we have to know what’s going on underneath the hood.”

Continuing, overall cost reduction

The lab cost of genome sequencing broke $100 two years ago, and some labs are advertising 30x whole-genome sequencing and analysis for about $500. And, Van Oene from PacBio says, “There is potential to drive the cost of whole-genome sequencing down by another order of magnitude.”

For example, Van Oene says that increasing sample throughput offers a major opportunity “to reduce instrument amortization by spreading the cost of the instrument across many more genomes.” Other opportunities are emerging through computing advances that speed processing and analysis as well as through improvements in compute logistics, such as data compression, which reduces storage requirements, and cloud computing, which reduces hardware requirements and expands access to data.

“The sequencing part of it is becoming a smaller part of the total cost of a workflow,” Fiona Kaper, PhD, VP, head of assay R&D, Illumina, points out.

So, rather than focusing solely on reducing the cost of sequencing itself, NGS innovators increasingly are targeting efficiencies across the entire workflow, from improving sequencing performance and optimizing library preparation to expanding multiplexing capabilities that generate more insight from every run. “Ultimately,” Chapman, Agilent Technologies, says, “the industry’s focus will shift from the cost of sequencing a genome to the cost of delivering a clinically meaningful answer.”

“There’s still some meat left on the bone in terms of how you can improve some workflows,” Barry, NEB, observes.

Scalability is one example. The challenge of sequencing and analyzing a genome varies by size, and solutions need to address that. Therefore, scaling the upstream environment to meet the need is a key consideration for some instrument makers, along with logistical enhancements, such as lyophilization and assay formats, to improve ease of use globally.

Multiomics and multimodal assays two other examples. “We see a lot more customers starting to layer in additional information like proteomics, transcriptomics, and spatial and single cell analysis,” Kaper says.

This trend suggests a future, modular, environment in which scientists can select and optimize the assays that best address their questions. “Having a black box [solution]…really doesn’t work for everyone,” Krishnan underscores.

Further cost reductions, these experts agree, will stem from automating and consolidating workflows, effectively multiplexing assays and developing multimodal assays to generate more actionable information with each run. Such simplification extends to library preparation, which has expanded from DNA and RNA to include proteomics and data from spatial and single-cell analysis.

Instruments developers also are reducing the quantities of reagents needed. Consequently, Van Oene says, “Reducing reagent volumes in library preparation and sequencing will lower consumable costs as well as shipping and storage requirements—which people don’t usually think about when they think about the cost of a genome.”

“Time is also a form of cost,” Kaper says. The industry recognizes that by increasing data quality, capabilities, and sequencing speed to help its users do more, better, and faster.

“The goal,” as Jayasinghe of Oxford Nanopore Technologies says, “is to help customers derive meaningful answers faster, with fewer steps and greater efficiency.”

Ubiquitous sequencing

Five to ten years from now, NGS may well be the core technology for research and clinical labs both, Kaper, Illumina, predicts. “Anything that has to do with the genome, like oncology, could really benefit from technologies that interrogate the genome and also the epigenome, proteome, etc.,” she says.

“Ultimately, the goal is not simply a cheaper genome, but a more informative genome, at a much lower cost,” Van Oene concludes.